Author: Helen Law

Even as machines known as “deep neural networks” have learned to converse, drive cars, beat video games and Go champions, dream, paint pictures and help make scientific discoveries, they have also confounded their human creators, who never expected so-called “deep-learning” algorithms to work so well. No underlying principle has guided the design of these learning systems, other than vague inspiration drawn from the architecture of the brain (and no one really understands how that operates either).

Quanta Magazine

---

About

Original story reprinted with permission from Quanta Magazine, an editorially independent publication of the Simons Foundation whose mission is to enhance public understanding of science by covering research developments and trends in mathematics and the physical and life sciences.

Like a brain, a deep neural network has layers of neurons—artificial ones that are figments of computer memory. When a neuron fires, it sends signals to connected neurons in the layer above. During deep learning, connections in the network are strengthened or weakened as needed to make the system better at sending signals from input data—the pixels of a photo of a dog, for instance—up through the layers to neurons associated with the right high-level concepts, such as “dog.” After a deep neural network has “learned” from thousands of sample dog photos, it can identify dogs in new photos as accurately as people can. The magic leap from special cases to general concepts during learning gives deep neural networks their power, just as it underlies human reasoning, creativity and the other faculties collectively termed “intelligence.” Experts wonder what it is about deep learning that enables generalization—and to what extent brains apprehend reality in the same way.

Last month, a YouTube video of a conference talk in Berlin, shared widely among artificial-intelligence researchers, offered a possible answer. In the talk, Naftali Tishby, a computer scientist and neuroscientist from the Hebrew University of Jerusalem, presented evidence in support of a new theory explaining how deep learning works. Tishby argues that deep neural networks learn according to a procedure called the “information bottleneck,” which he and two collaborators first described in purely theoretical terms in 1999. The idea is that a network rids noisy input data of extraneous details as if by squeezing the information through a bottleneck, retaining only the features most relevant to general concepts. Striking new computer experiments by Tishby and his student Ravid Shwartz-Ziv reveal how this squeezing procedure happens during deep learning, at least in the cases they studied.

Tishby’s findings have the AI community buzzing. “I believe that the information bottleneck idea could be very important in future deep neural network research,” said Alex Alemi of Google Research, who has already developed new approximation methods for applying an information bottleneck analysis to large deep neural networks. The bottleneck could serve “not only as a theoretical tool for understanding why our neural networks work as well as they do currently, but also as a tool for constructing new objectives and architectures of networks,” Alemi said.

Some researchers remain skeptical that the theory fully accounts for the success of deep learning, but Kyle Cranmer, a particle physicist at New York University who uses machine learning to analyze particle collisions at the Large Hadron Collider, said that as a general principle of learning, it “somehow smells right.”

Geoffrey Hinton, a pioneer of deep learning who works at Google and the University of Toronto, emailed Tishby after watching his Berlin talk. “It’s extremely interesting,” Hinton wrote. “I have to listen to it another 10,000 times to really understand it, but it’s very rare nowadays to hear a talk with a really original idea in it that may be the answer to a really major puzzle.”

According to Tishby, who views the information bottleneck as a fundamental principle behind learning, whether you’re an algorithm, a housefly, a conscious being, or a physics calculation of emergent behavior, that long-awaited answer “is that the most important part of learning is actually forgetting.”

The Bottleneck

Tishby began contemplating the information bottleneck around the time that other researchers were first mulling over deep neural networks, though neither concept had been named yet. It was the 1980s, and Tishby was thinking about how good humans are at speech recognition—a major challenge for AI at the time. Tishby realized that the crux of the issue was the question of relevance: What are the most relevant features of a spoken word, and how do we tease these out from the variables that accompany them, such as accents, mumbling and intonation? In general, when we face the sea of data that is reality, which signals do we keep?

“This notion of relevant information was mentioned many times in history but never formulated correctly,” Tishby said in an interview last month. “For many years people thought information theory wasn’t the right way to think about relevance, starting with misconceptions that go all the way to Shannon himself.”

Claude Shannon, the founder of information theory, in a sense liberated the study of information starting in the 1940s by allowing it to be considered in the abstract—as 1s and 0s with purely mathematical meaning. Shannon took the view that, as Tishby put it, “information is not about semantics.” But, Tishby argued, this isn’t true. Using information theory, he realized, “you can define ‘relevant’ in a precise sense.”

Imagine X is a complex data set, like the pixels of a dog photo, and Y is a simpler variable represented by those data, like the word “dog.” You can capture all the “relevant” information in X about Y by compressing X as much as you can without losing the ability to predict Y. In their 1999 paper, Tishby and co-authors Fernando Pereira, now at Google, and William Bialek, now at Princeton University, formulated this as a mathematical optimization problem. It was a fundamental idea with no killer application.

“I’ve been thinking along these lines in various contexts for 30 years,” Tishby said. “My only luck was that deep neural networks became so important.”

Eyeballs on Faces on People on Scenes

Though the concept behind deep neural networks had been kicked around for decades, their performance in tasks like speech and image recognition only took off in the early 2010s, due to improved training regimens and more powerful computer processors. Tishby recognized their potential connection to the information bottleneck principle in 2014 after reading a surprising paper by the physicists David Schwab and Pankaj Mehta.

The duo discovered that a deep-learning algorithm invented by Hinton called the “deep belief net” works, in a particular case, exactly like renormalization, a technique used in physics to zoom out on a physical system by coarse-graining over its details and calculating its overall state. When Schwab and Mehta applied the deep belief net to a model of a magnet at its “critical point,” where the system is fractal, or self-similar at every scale, they found that the network automatically used the renormalization-like procedure to discover the model’s state. It was a stunning indication that, as the biophysicist Ilya Nemenman said at the time, “extracting relevant features in the context of statistical physics and extracting relevant features in the context of deep learning are not just similar words, they are one and the same.”

The only problem is that, in general, the real world isn’t fractal. “The natural world is not ears on ears on ears on ears; it’s eyeballs on faces on people on scenes,” Cranmer said. “So I wouldn’t say [the renormalization procedure] is why deep learning on natural images is working so well.” But Tishby, who at the time was undergoing chemotherapy for pancreatic cancer, realized that both deep learning and the coarse-graining procedure could be encompassed by a broader idea. “Thinking about science and about the role of my old ideas was an important part of my healing and recovery,” he said.

In 2015, he and his student Noga Zaslavsky hypothesized that deep learning is an information bottleneck procedure that compresses noisy data as much as possible while preserving information about what the data represent. Tishby and Shwartz-Ziv’s new experiments with deep neural networks reveal how the bottleneck procedure actually plays out. In one case, the researchers used small networks that could be trained to label input data with a 1 or 0 (think “dog” or “no dog”) and gave their 282 neural connections random initial strengths. They then tracked what happened as the networks engaged in deep learning with 3,000 sample input data sets.

The basic algorithm used in the majority of deep-learning procedures to tweak neural connections in response to data is called “stochastic gradient descent”: Each time the training data are fed into the network, a cascade of firing activity sweeps upward through the layers of artificial neurons. When the signal reaches the top layer, the final firing pattern can be compared to the correct label for the image—1 or 0, “dog” or “no dog.” Any differences between this firing pattern and the correct pattern are “back-propagated” down the layers, meaning that, like a teacher correcting an exam, the algorithm strengthens or weakens each connection to make the network layer better at producing the correct output signal. Over the course of training, common patterns in the training data become reflected in the strengths of the connections, and the network becomes expert at correctly labeling the data, such as by recognizing a dog, a word, or a 1.

In their experiments, Tishby and Shwartz-Ziv tracked how much information each layer of a deep neural network retained about the input data and how much information each one retained about the output label. The scientists found that, layer by layer, the networks converged to the information bottleneck theoretical bound: a theoretical limit derived in Tishby, Pereira and Bialek’s original paper that represents the absolute best the system can do at extracting relevant information. At the bound, the network has compressed the input as much as possible without sacrificing the ability to accurately predict its label.

Tishby and Shwartz-Ziv also made the intriguing discovery that deep learning proceeds in two phases: a short “fitting” phase, during which the network learns to label its training data, and a much longer “compression” phase, during which it becomes good at generalization, as measured by its performance at labeling new test data.

As a deep neural network tweaks its connections by stochastic gradient descent, at first the number of bits it stores about the input data stays roughly constant or increases slightly, as connections adjust to encode patterns in the input and the network gets good at fitting labels to it. Some experts have compared this phase to memorization.

Then learning switches to the compression phase. The network starts to shed information about the input data, keeping track of only the strongest features—those correlations that are most relevant to the output label. This happens because, in each iteration of stochastic gradient descent, more or less accidental correlations in the training data tell the network to do different things, dialing the strengths of its neural connections up and down in a random walk. This randomization is effectively the same as compressing the system’s representation of the input data. As an example, some photos of dogs might have houses in the background, while others don’t. As a network cycles through these training photos, it might “forget” the correlation between houses and dogs in some photos as other photos counteract it. It’s this forgetting of specifics, Tishby and Shwartz-Ziv argue, that enables the system to form general concepts. Indeed, their experiments revealed that deep neural networks ramp up their generalization performance during the compression phase, becoming better at labeling test data. (A deep neural network trained to recognize dogs in photos might be tested on new photos that may or may not include dogs, for instance.)

It remains to be seen whether the information bottleneck governs all deep-learning regimes, or whether there are other routes to generalization besides compression. Some AI experts see Tishby’s idea as one of many important theoretical insights about deep learning to have emerged recently. Andrew Saxe, an AI researcher and theoretical neuroscientist at Harvard University, noted that certain very large deep neural networks don’t seem to need a drawn-out compression phase in order to generalize well. Instead, researchers program in something called early stopping, which cuts training short to prevent the network from encoding too many correlations in the first place.

Tishby argues that the network models analyzed by Saxe and his colleagues differ from standard deep neural network architectures, but that nonetheless, the information bottleneck theoretical bound defines these networks’ generalization performance better than other methods. Questions about whether the bottleneck holds up for larger neural networks are partly addressed by Tishby and Shwartz-Ziv’s most recent experiments, not included in their preliminary paper, in which they train much larger, 330,000-connection-deep neural networks to recognize handwritten digits in the 60,000-image Modified National Institute of Standards and Technology database, a well-known benchmark for gauging the performance of deep-learning algorithms. The scientists saw the same convergence of the networks to the information bottleneck theoretical bound; they also observed the two distinct phases of deep learning, separated by an even sharper transition than in the smaller networks. “I’m completely convinced now that this is a general phenomenon,” Tishby said.

Humans and Machines

The mystery of how brains sift signals from our senses and elevate them to the level of our conscious awareness drove much of the early interest in deep neural networks among AI pioneers, who hoped to reverse-engineer the brain’s learning rules. AI practitioners have since largely abandoned that path in the mad dash for technological progress, instead slapping on bells and whistles that boost performance with little regard for biological plausibility. Still, as their thinking machines achieve ever greater feats—even stoking fears that AI could someday pose an existential threat—many researchers hope these explorations will uncover general insights about learning and intelligence.

Brenden Lake, an assistant professor of psychology and data science at New York University who studies similarities and differences in how humans and machines learn, said that Tishby’s findings represent “an important step towards opening the black box of neural networks,” but he stressed that the brain represents a much bigger, blacker black box. Our adult brains, which boast several hundred trillion connections between 86 billion neurons, in all likelihood employ a bag of tricks to enhance generalization, going beyond the basic image- and sound-recognition learning procedures that occur during infancy and that may in many ways resemble deep learning.

For instance, Lake said the fitting and compression phases that Tishby identified don’t seem to have analogues in the way children learn handwritten characters, which he studies. Children don’t need to see thousands of examples of a character and compress their mental representation over an extended period of time before they’re able to recognize other instances of that letter and write it themselves. In fact, they can learn from a single example. Lake and his colleagues’ models suggest the brain may deconstruct the new letter into a series of strokes—previously existing mental constructs—allowing the conception of the letter to be tacked onto an edifice of prior knowledge. “Rather than thinking of an image of a letter as a pattern of pixels and learning the concept as mapping those features” as in standard machine-learning algorithms, Lake explained, “instead I aim to build a simple causal model of the letter,” a shorter path to generalization.

Such brainy ideas might hold lessons for the AI community, furthering the back-and-forth between the two fields. Tishby believes his information bottleneck theory will ultimately prove useful in both disciplines, even if it takes a more general form in human learning than in AI. One immediate insight that can be gleaned from the theory is a better understanding of which kinds of problems can be solved by real and artificial neural networks. “It gives a complete characterization of the problems that can be learned,” Tishby said. These are “problems where I can wipe out noise in the input without hurting my ability to classify. This is natural vision problems, speech recognition. These are also precisely the problems our brain can cope with.”

Meanwhile, both real and artificial neural networks stumble on problems in which every detail matters and minute differences can throw off the whole result. Most people can’t quickly multiply two large numbers in their heads, for instance. “We have a long class of problems like this, logical problems that are very sensitive to changes in one variable,” Tishby said. “Classifiability, discrete problems, cryptographic problems. I don’t think deep learning will ever help me break cryptographic codes.”

Generalizing—traversing the information bottleneck, perhaps—means leaving some details behind. This isn’t so good for doing algebra on the fly, but that’s not a brain’s main business. We’re looking for familiar faces in the crowd, order in chaos, salient signals in a noisy world.

Original story reprinted with permission from Quanta Magazine, an editorially independent publication of the Simons Foundation whose mission is to enhance public understanding of science by covering research developments and trends in mathematics and the physical and life sciences.

Tech

Project Loon has already proven its real-world usefulness once this year.

Tech

Sometimes a news story is so complete in its absurdity that it’s hard to be on anyone’s side. That’s the only way to see the current legal battle between the City of Everett, Washington, and a group of young women who make their livings whooshing out pumpkin spice lattes while customers ogle their barely-covered bodies.

If you’ve ever been in the Pacific Northwest, you’ve seen the drive-up coffee kiosks in the parking lots of nearly every strip mall. It’s a super-inexpensive path to entrepreneurship–I’ve seen these kiosks for sale for only $ 20,000. No larger than the average coat closet, each kiosk contains a single server and an espresso machine.

With so many kiosks everywhere, competition is fierce and owners look for creative ways to stand out. Some offer a range of surprising add-ins (Nutella is my favorite). Some also sell unexpected food items, such as biscuits and gravy. Others hand out free biscotti, or freshly-made donut holes, or dog biscuits for canine passengers.

So it’s a no-brainer that some kiosks seek to woo customers with scantily-clad baristas. Since most kiosks have big windows in all directions, it’s almost a low-level peep show you get free with your hot beverage.

That’s where the trouble begins. Eight years ago–after a lengthy undercover investigation in a town that has plenty of other problems–five Everett bikini baristas were arrested and charged with prostitution because they accepted money for such things as a “whipped cream show” (two baristas lick whipped cream off each other) and “basketball” (in which customers throw money that the baristas catch in their underwear).

Since then, the city has tried to contain the baristas using its lewd conduct laws, which Assistant City Attorney Ramsey Ramerman, claims was “simply ineffective.” And so the Everett City Council unanimously passed a law requiring baristas and all other fast food servers to wear clothing that covers “minimum body areas.” It continues:

“Such clothing shall not be see-through and must fit adequately so that undergarments and all minimum body areas remain covered at all times including when the wearer is sitting, standing, bending reaching or performing other work duties.”

Wondering what constitutes a “minimum body area”? Never fear–the City Council has provided a definition:

“‘Minimum body areas’ means the upper and lower body (breast/pectorals, stomach, back below the shoulder blades, buttocks, top three inches of leg below the buttocks, pubic area and genitals).”

For good measure, Everett also enacted a city-wide code defining a lewd act (among other things) as:

“1. An exposure or display of one’s genitals, anus, bottom one-half of the anal cleft, or any portion of the areola or nipple of the female breast; or

2. An exposure of more than one-half of the part of the female breast located below the top of the areola; provided, that the covered area shall be covered by opaque material and coverage shall be contiguous to the areola.”

Just to be extra clear, it added:

“Body paint is not ‘opaque material.'”

These patently silly laws were met with an even sillier lawsuit by seven bikini baristas and one kiosk owner. Not satisfied with challenging the laws on the grounds of restraint of trade or fairness–servers in restaurants and private clubs aren’t included–attorneys for this group went straight for the First Amendment, arguing that the right to expose most of one’s skin constitutes self-expression.

As the Seattle alternative weekly The Stranger puts it, the free speech arguments in the complaint are “absurd in the lengths they go to avoid saying bikini baristas are meant to serve horny people.”

For example, it says this about the baristas and their bikinis:

“They express messages of freedom, openness, acceptance, empowerment, and individuality. By exposing who they are as people through tattoos, scars, and the bikinis that they choose to wear, the Baristas exchange conversations with customers about life experiences, personal choices, and other topics that would not otherwise occur. The Baristas cannot express these messages and prompt these discussions without the unique expression that wearing a bikini provides.”

Furthermore:

“The Baristas use bikinis to portray a fun and happy-go-lucky image that gives customers a quick break from their daily lives. The bikini allows customers to imagine for a moment that they are relaxing at the beach or on vacation. The Baristas could not portray this message with another uniform.”

Not only that, the individual baristas explain what wearing a bikini means to them. Each repeats that the bikinis have nothing to do with sex and everything to do with empowerment. One explains that her bikini reveals scars from a childhood accident, which she talks to customers about and “they open up with their own stories.” Another says, “Millions of women fought for our rights and right to vote, and it’s my right to wear what I want.”

This is where the baristas lost me because the suffragists of 100 years ago went to prisons and workhouses and went on hunger strikes and endured the torture of having six-inch rubber hoses forced down their throats and nasal passages along with near-universal derision and disdain. I don’t think they went through all that out of a fervent hope that someday their female descendants would be empowered to serve coffee while wearing bikinis and playing “basketball” in the ostensible pursuit of self-expression, and the actual pursuit of larger tips.

I don’t want to be a killjoy, and neither should the City of Everett. Since the rule at most bikini stands seems to be that customers must remain in their cars, the worst that can likely result from most scantily-clad barista stands is the occasional whipped-cream show or fully nude coffee serving. In a city that’s also suing a pharmaceutical company over its rampant opioid problem, that just doesn’t seem like such a big deal.

Free speech, on the other hand, is a big deal. I’m not sure if the bikini baristas or their attorney noticed that they filed their complaint during Banned Books Week, an event that reminds us that classics from The Adventures of Huckleberry Finn to Toni Morrison’s Beloved and even The Diary of Anne Frank have been censored in American schools. All over the world, men and women risk their freedom, their health, and sometimes their lives for the right to write, film, or otherwise share the truth as they see it. That’s worth fighting and dying for. The right to show off the bottom half of one’s anal cleft? Not so much.

Tech

Negotiations have dragged on since February.

Tech

There’s a big, buzzy trend sweeping the planet. Countries and municipalities like the United Kingdom, Norway, France, the Netherlands, India, and China have all made a similar pledge: They’re going to kick their gasoline-powered car habit in the next few decades. It is a nice idea, maybe even a world-saving one. The transportation sector accounts for 14 percent of the world’s greenhouse gas emissions, which humanity must cut by 40 to 70 percent by 2050 to prevent that disastrous 2-degree Celsius temperature hike.

Which explains why the state of California is shouting Me too! “I’ve gotten messages from the governor asking, ‘Why haven’t we done something already?’” Mary Nichols, head of the California Air Resources Board, told Bloomberg last week. “The governor has certainly indicated an interest in why China can do this and not California.” She said a California ban on the sale of non-electric vehicles is at least a decade away.

Just one little problem. “There is not, in any of these places, policy follow-up to ensure that these goals are met,” says Nic Lutsey, who studies electric vehicle policy at the International Council on Clean Transportation. No plans, no blueprints, no teeth. These pronouncements are the governmental equivalent of pledging to drop 100 pounds without considering details like diet or exercise.

Sure, electric vehicles have come a ways in the last decade. They’re quick. Some of them are cool. Thanks to fast-improving battery tech and charging infrastructure, they can go further and to more places than ever before. Even legacy carmakers like General Motors, Ford, BMW, and Jaguar Land Rover are joining the faithful (Tesla, Nissan) in the burgeoning EV market.

But EVs are still just a bitty fraction of vehicles sold worldwide—4.3 percent in 2016. Charging infrastructure is spreading, but it’s far from everywhere. And even mass market EVs aren’t cheap.

Still, pay attention. “The announcements are hugely important,” says Nic Lutsey, who studies electric vehicle policy at the International Council on Clean Transportation. “They show what governments are thinking and what all their research groups are telling them: that to clean up transportation, to avoid the worst climate change impact, they will need to electrify the fleet. It’s showing that a consensus is clearly emerging.”

The positive news for countries, states, cities, and any other jurisdiction looking to boot the polluters? There is a way forward. The long road to electro-bliss is paved with incentivizing nudges from local governments, international coordination with carmakers; hard-hitting, quota-filled national programs, and a heaping spoonful of consumer education. And the work starts now.

Today

Electric cars may be more expensive than their gas-powered counterparts, but governments around the world have found combinations of carrots and sticks to convince their citizens to go with the plug.

In the Bugs Bunny snack category, Norway’s electric vehicle owners dodge municipal parking charges, purchase and import taxes, and tunnel and ferry fees—a serious money-saver in a nation streaked with fjords. They can charge or refuel their electric and hydrogen cars for free. They get to drive in bus-only lanes. Ha det bra, traffic! The result: More than a quarter of the vehicles sold in Oslo each year are electric. This investment in the future has its cost, of course. “The foregoing of the sales tax of EVs costs Norway a lot of money, but furthered their goals of making them the electric vehicles leaders of the world,” says Costa Samaras, who studies alternative energy at Carnegie Mellon University.

In Shanghai, where 11 percent of 2015 new vehicle sales were electric, buyers receive a $ 4,400 purchase subsidy. But the biggest money-saver kicks in after the trip to the dealer. To restrict the number of cars on the road, the local government auctions off license plates to the highest bidders; they can fetch as much as $ 12,000. Electric vehicles get to dodge these auctions, saving their owners big bucks.

For a taste of the stick, look to London. Come 2020, a central swath of the British capital will form an “ultra low emission zone.” Any vehicle that wants in and doesn’t meet stringent emission standards will face a daily charge. (License plate-reading cameras will ensure all wheeled things are in compliance.) Shanghai has a similar plan in place that applies to commercial vehicles.

Finally, places like California hoping to goose electric sales ahead of a gas ban could look to … California, which is ramping up the charging infrastructure that makes EV ownership feasible. “Money is not the only reason people are not buying electric vehicles today,” says Lutsey. “There’s also the convenience.” San Jose, the humming heart of Silicon Valley, is already the fifth most charged-up city in the world, with 500 charge points per million people. San Francisco has about 450 per million. Good, but not good enough. Fortunately, California has a $ 800 million electric charging windfall coming its way, in the form of a settlement with Volkswagen over its diesel emissions cheating scandal. The money goes straight to the coffers of the California Air Resources Board, which will spread the money out over the next 10 years. Slowly but surely, it should get easier to own an electric vehicle in California. And that makes it easier for California to ban gasoline for good.

Tomorrow

Baby steps today are necessary, but they don’t get you out of the long-term work a gas ban requires. China is on the right track. Last week, the world’s largest auto market—where a third of all EVs sold last year found their owners—rolled out a much-anticipated electric vehicle plan. The scheme pushes an aggressive cap-and-trade-like policy that will force foreign and domestic carmakers to up the number of zero- and low-emission vehicles they produce by 2020. (The plan, it should be noted, looks like California’s own ZEV scheme, which forces automakers to build electric models or purchase credits from those doing so.)

These policies are producing results. This week, General Motors announced it would build at least 20 new fully electric vehicle models by 2023, and eventually phase of non-electric vehicles. Volvo is going (mostly) electric. BMW, too. No big automaker can afford to lose access to such lucrative markets.

Way Down the Line

The trickiest part of pulling off this sort of move is doing it without hurting your most vulnerable citizens. In this case, those are the low-income, largely rural folks who rely on the affordability of the internal combustion engine to get them to their jobs, their grocery stores, their community centers. It takes a long time for innovations to trickle down to the country’s poorest households.

“Banning sure could increase the number of electric vehicles on the road, but would sure come at another cost,” says Samaras. “We need to make sure mobility and access at lower income population could be improved.” That means investment in public transportation. A lot of it. Also, consumer education for the folks who know nothing about what driving an electric car entails. Annoyingly, there’s no easy answer for that. “There’s a huge diversity of car buyers,” says Lutsey. “Trying to meet every consumer’s demand is incredibly difficult. You need a car that’s the right brand, the right reputation, the right cost point, the right convenience.”

Maybe that’s why no other American states or government agencies have floated gas car bans. But do not count out California, which has a long history of leading on emissions. As federal legislators tussled over the 1963 Clean Air Act, they wrote in provisions allowing the state to create its own stricter regulations, acknowledging it was already way out ahead on environmental standards. Nine other states have signed onto California’s Zero Emission Vehicle mandate, which means the it controls over a quarter of the American vehicle market. Even if California can’t receive a special waiver from the federal government to ban gas, the state seems to think it can pull it off by manipulating vehicle registration rules or ordering gas vehicles off state highways.

Still, more ideas are needed. Banning fossil fuel gluggers isn’t impossible. But moving them to zero is going to take a lot of strategery. And the game has just begun.

Tech

Nearly 10 million Americans saw ads on Facebook that have since been linked to Russia and its alleged plot to influence the 2016 U.S. presidential election.

The statistic, revealed by Facebook on Monday, comes as the company takes fire for the role its automated advertising platform played in spreading the online ads.

Facebook said Sunday that it would deliver to U.S. lawmakers copies of the roughly 3,000 Russian-linked ads. In a blog post on Monday Facebook vice president of policy and communications Elliot Schrage explained some of the kinds of Russian-bought ads it was handing over to Congress.

Schrage said that the ads contained “divisive social and political messages across the ideological spectrum” and focused on “topics from LGBT matters to race issues to immigration to gun rights.” Schrage did not cite specific ads, of which 44% were seen prior to the U.S. presidential election on Nov. 8, with the remainder appearing after.

Get Data Sheet, Fortune’s technology newsletter.

Schrage defended Facebook’s automated advertising service, writing that it was “designed to show people ads they might find useful, instead of showing everyone ads that they might find irrelevant or annoying.” However, he acknowledged that Facebook (fb) now recognizes the potential to misuse the ad-targeting system and he reiterated Facebook’s earlier statement that it would hire an additional 1,000 people to review online ads and determine if they violate the company’s policies.

Schrage also said that Facebook is working with Google (goog), Twitter (twtr), and other unspecified technology companies to combat the spread of misleading online ads, although he didn’t reveal any specifics of their efforts. U.S. lawmakers have called on Google, Facebook, and Twitter to testify before Congress on the possible role their services played in distributing fake and deceptive information.

Facebook also “would have caught these malicious actors faster and prevented more improper ads from running” if it knew then what it knows now about how people can misuse the service. He also spoke about how the company’s improved tools to identify bogus ads would have helped.

“The ad transparency tool we’re building will be accessible to anyone, including industry and political watchdog groups,” Schrage said. “And our improved enforcement and more restrictive content standards for ads would have rejected more of the ads when submitted.”

Tech

It seems like every business discussion today is just counting the seconds before the term “platform” comes up. Books and articles are written, pundits swoon and conference audiences nod and exchange glances in knowing agreement. Everyone, it seems, wants to transform their business into a platform.

Yet take the argument to its logical conclusion and the message becomes problematic. Platforms, as many have observed, function as multi-sided markets and therefore must connect value to value. So if everybody becomes a platform, who actually creates the value to make a vibrant marketplace?

The truth is that, while some amazing platform businesses have been created, there is also a considerable amount of survivorship bias going on. We notice the Amazons, Ubers and AirBnB’s, but forget about the thousands of platform startups that failed. Make no mistake, even in an increasingly networked world, you still need to create, deliver and capture value.

Creating Value

When Elance was founded in 1999, it seemed like a really good idea. Taking its name from a Harvard Business Review article titled The Dawn of the E-Lance Economy the founders sought to match freelance contractors and firms much like Monster.com did for full-time recruiting. Unfortunately, the business really never gained any traction.

So the investors decided to hire a new CEO and take the company in a new direction. Instead of matching firms to freelancers, it would help companies manage relationships. This idea met with much greater success and Elance became a pioneer in vendor management software. In fact, it became so successful that it attracted stiff competition from the likes of SAP and Oracle.

So Elance sold the software business and return to the original idea. This time though, it applied what it had learned about making relationships successful rather than just making matches. It partnered with training firms to help freelancers build and certify skills, created private talent clouds for customers and developed algorithms to create better engagements.

The strategy was a resounding success and the company later merged with oDesk to form Upwork, the world’s largest freelance platform. Elance is no exception either. From Netflix to Amazon to just about everything in between, it seems that eventually platform businesses eventually need to go beyond merely making matches and create a product or service.

Delivering Value

One industry that’s been absolutely ravaged by the platform economy is retail. With digital commerce platforms offering better prices, selection and convenience, how can brick and mortar retailers ever compete? After all, who goes into a store anymore?

Apparently, just about everybody. According to the most recent data from the US Census more than 90% of sales still go to traditional retail outlets. While clearly automation and e-commerce have depressed margins and sent many companies reeling, there’s still a lot to be said for an in-person, in-store experience.

Take a closer look and you’ll find that digital commerce has its limitations. For example, AirBnB’s estimated revenues of $ 2.8 billion are impressive, but represent less than 1% of the $ 500 billion hospitality industry. That shouldn’t be surprising. Many travelers — especially business travelers — are looking for more than just a hotel room, but also the service that comes with it.

That’s why Apple has invested so much time, effort and money into its stores and why successful platforms like Amazon and and Warby Parker are opening up retail locations. An online purchase is only a mere transaction, but in a retail environment, well trained salespeople can build relationships, cater to a particular customer’s needs, service purchases and upsell.

Capturing Value

One of the most attractive aspects of platform businesses is how easy they are to start. Pretty much anyone can build a website, aggregate disparate information and offer it as a solution to customers. You don’t need to rent expensive commercial space or even develop particularly sophisticated software. You just collect data, make it accessible and you’re in business.

Yet that is also the platform model’s achilles heel. Low barriers to entry lead to aggressive competition, which makes an expensive marketing war almost inevitable. That’s why becoming a successful platform so often depends on how much of a war chest you can attract from venture capital funds and, because platforms tend to be “winner take all” propositions, there are far more losers than winners.

To see how this often plays out, take the time to read Timothy B. Lee’s profile of Uber. Yes, Uber has driven down the cost of taxis, but it has lost billions in the process. It’s not at all clear whether it has actually built a sustainable business model or is just in a predatory race to drive competitors out of business so that it can use its monopoly power to drive up prices.

Sure, it’s possible that Uber may eventually become profitable, but is it really the paragon of the new economy that its advocates make it out to be, or a throwback to the robber baron days of Vanderbilt, Rockefeller and Carnegie?

Harnessing The True Power of Platforms

None of this to say that platforms are all hype and no substance. As I’ve written before, platforms allow us to access ecosystems of talent, technology and information in an incredibly powerful way. That, in turn, is changing how we need to compete, shifting the basis of competition from optimizing efficiency to widening and deepening connections.

To understand why, let’s return to Elance. As a freelance matching service, it offered little benefit. Most companies have their own networks of contractors they like to work with. It was only when it started to create value above and beyond a simple search function that it became a profitable business. There’s no free lunch. Value creation is simply not something you can run a successful business without.

In a similar vein, Amazon allows you to access ecosystems of retailers and that’s incredibly helpful and powerful, but its competitive advantage is its distribution system. If all it was doing was showing you offers, anybody could compete with it on an even playing field and that would make it very hard for the company to make money.

So don’t be fooled. Leveraging the power of platforms can be an excellent way to extend and improve a strong business model, but it cannot replace one.

Tech

You will soon be able to ride home from your local car dealership in a car that finds its way there unassisted while you nap or read. That reality came a whole lot closer this week, with bipartisan agreement in the Senate on legislation allowing self-driving cars to take the the roads. The law is expected to come up for vote in the near future, and pass.

The House passed similar legislation, also with bipartisan support, several weeks ago. That legislation allows car manufacturers to sell up to 25,000 autonomous vehicles the first year they offer them. That will go up to 100,000 cars a year if the self-driving cars prove as safe as human-driven ones. And that’s not all. The Trump administration also helped out recently by issuing voluntary safety guidelines for autonomous cars and at the same time requesting that states avoid writing laws or regulations governing self-driving cars and possibly hampering their introduction.

The senators who arrived at the self-driving deal note that autonomous cars appear to be safer than human-driven ones. “Ultimately, we expect adoption of self-driving vehicle technologies will save lives, improve mobility for people with disabilities, and create new jobs,” said Senators John Thune (R-S.D.) and Gary Peters (D-Mich.) in a joint statement. They may be right: When a Tesla owner died while his car was in Autopilot mode last summer, company founder Elon Musk pointed out that it was the first known Autopilot fatality in 130 million miles of driving, whereas there’s a human fatality for every 89 million miles of traditional driving.

But if cars with no one at the wheel will soon become a common sight, the same won’t be true of semi trucks. The Teamsters successfully lobbied for the House version of the bill to limit self-driving vehicles to 10,000 pounds or less. That could be a problem for the U.S. trucking industry, which was short an estimated 48,000 drivers at the end of 2015, a shortage that’s expected to grow to 175,000 over the next seven years. That will create enormous pressure to replace hard-to-find long-haul truck drivers with no-muss, no-fuss AI.

Tech

Apple’s iPhones are critical to its business and its cash hoard. But now speculation is that the money machine may not be humming as well as expected.

The upcoming iPhone X smartphone, available on Nov. 3, may be having manufacturing problems that could lead to a more limited availability at launch than investors had hoped. There’s also talk of whether slower-than-expected iPhone 8 sales are due to shoppers preferring to wait for the iPhone X or something more troubling—like disappointed customers.

But it wasn’t all bad news for Apple. For instance, the company is believed to be making a healthy profit on the sale of each iPhone, further bolstering arguments by industry watchers who believe the iPhone X and iPhone 8 will help Apple reach new financial heights.

This is Fortune’s latest weekly roundup of the biggest Apple news. Here’s last week’s roundup.

Quick Hits

Get Data Sheet, Fortune’s technology newsletter

One more thing…Apple’s Beats division is releasing a new, high-end Studio3 Wireless Headphones next month. I’ve been testing them for the last few weeks. Here’s my review. The short version? They’re really, really, really good.

Tech

I take a lot of pride in the fact that my portfolio has never experienced a dividend cut. I came close once with KMI, but I managed to sell the position before the cut was announced. I spend a lot of my time during the due diligence process focusing on dividend-related metrics with a specific focus on sustainability and dividend growth prospects. Well, I just put that perfect record at risk with a purchase of Uniti Group (UNIT) shares at $ 15.01, or a very hefty 15.98% yield.

This ~16% yield is nearly double my previously high yield, which was Omega Healthcare Inc (OHI) at just a tad bit more than 8%. Typically, when I see yields in the double digits, I get nervous. Yields that high mean the asset is distressed. When looking at stocks like UNIT investors are receiving a very high potential reward for exposing themselves to a very high perceived risk. I’m not a huge fan of making these risky bets. But, I’ve spent a lot of time reading articles and commentary about UNIT published over the last couple of weeks focused on the company’s enormous ~40% fall since the start of August. I’ve read enough bullish commentary to get me interested in the stock, especially from contributors here at Seeking Alpha that I’ve come to respect over the years.

Honestly, I think this company’s recent drama has been exhausted by the Seeking Alpha community and I don’t have anything new to add to the conversation other than the fact that I am now long the stock. I like to keep followers up to date on my recent portfolio maneuvers though, so I wanted to write this piece. However, instead of re-hash the pros and cons of UNIT ownership here, I will link you to some of my favorite articles recently published regarding UNIT.

My absolute favorite REIT contributor here at Seeking Alpha is Brad Thomas. Mr. Thomas has led me to highly profitable investment decisions on several occasions. I respect his opinion in the REIT space above all others. In late August/early September, he published two bullish pieces on UNIT (when shares were trading at levels much higher than they are today). One of them remains behind SA’s Marketplace paywall, but another is free to the public. Here’s a link to Mr. Thomas’s most recent UNIT piece which includes an informative interview with UNIT’s CEO Kenny Gunderson and a reiteration of Mr. Thomas’ “BUY” rating on shares post Q2 results.

Another UNIT piece that really caught my eye was Dividend Sensei’s recent article explaining why he’s adding to his UNIT stake, making it his largest individual position. I really like Dividend Sensei’s work here at SA. He puts together a very in-depth analysis that is also easy to understand. I admit that I am much more risk-averse than he seems to be. He trades with margin and oftentimes seeks much higher yields than I do. I would never allow a company like UNIT to become my largest holding. Actually, I don’t imagine a future where UNIT ever makes up more than 1% of my overall portfolio (right now, it’s weighting is ~0.375%). Even so, I oftentimes find is opinions to be more than reasonable and while our portfolio management strategies aren’t the same (which is to be expected because no two people are in the same situation when it comes to personal finance and long-term financial goals), I still respect his opinion immensely.

I’ll talk more about this piece in a bit, but Ian Bezek’s recent article on the matter was valuable to me as well, especially in terms of trying to put this company’s potential risks into perspective against what seems to be an overly bullish consensus amongst SA contributors and readers, mainly, I think, because of UNIT’s incredibly high yield. Ian is long UNIT, although as of his latest piece, he hadn’t added to his position on more recent weakness. I think Ian has a keen eye for value and the fact that he too was long, played a role in my decision-making.

Alpha Gen Capital wrote a particularly bullish piece, hinting at the fact that UNIT could be one of the year’s best opportunities due to recent overreactions in the share price movement. This piece really breaks down the issues that UNIT is facing with WIN, some of the potential fallouts of legal/bankruptcy scenarios. All of this is very confusing and remains highly speculative, though my main takeaway is that it appears likely that, regardless of a WIN bankruptcy, UNIT will still be in a position of strength due to the Master Lease arrangement it has with WIN. Lease re-negotiation still appears to be a possible scenario here, which would change the landscape that UNIT operates in the present, but for the time being, I’m willing to trust in the payments from the Master Lease deal and rely on the strength of UNIT’s infrastructure, which should remain in demand moving forward.

And most recently, Beyond Saving and Dane Bowler have written pieces regarding the breaking news that broke this week surrounding more legal/head fund issues regarding WIN bonds defaulting. The comment streams following all of these pieces have been enlightening. There are bulls and bears on either side of the aisle, but I was pleasantly surprised to see that another one of my favorite SA REIT contributors, Bill Stoller, recently went long UNIT as well. As far as I know, Mr. Stroller hasn’t published an article focused on UNIT, but I’ve seen him make enough solid calls in the past to give weight to his recent purchase in my own decision-making process.

So, there you have it. This is a unique situation for me, investing in a speculative income play like this. I may not like to take big risks like this, but I have always said that I like to buy things when they’re cheap. At this point, I admit that UNIT could just as easily turn out to be a value trap as it could a tremendous value. Looking at the value of the company’s assets and its cash flow potential, I see validity in calls that have price targets in the $ 35-40 range. That would imply massive upside at today’s prices. Due to issues that UNIT faces with its over-reliance on distressed Windstream (WIN), I don’t foresee UNIT selling anywhere near the fair value of its parts anytime soon though, so their estimates really amount to a hill of beans.

There are so many rumors and potentially headwinds swirling around this stock that I think it’s nearly impossible to predict its future share price movements. I wouldn’t be surprised to see a short squeeze that sends the stock rocketing up to $ 20 or more tomorrow. I also wouldn’t be surprised to continued pressure on shares, sending them down into the single digits. I won’t attempt to signal any sort of direction of these shares; simply put, I acknowledge that I am speculating here.

This is why I bought a relatively small, ¼ position. I bought these shares because of the combined upside potential of the shares in a turnaround as well as the very high ~16% dividend yield. Right now, it appears that UNIT’s dividend is covered by AFFO, which management expects to come in somewhere in the $ 2.50 range in 2017. This is a good thing. However, as discussed at length in this article by Ian Bezek, a dividend cut may still be in the cards because without one, it will be very difficult for UNIT to raise cash.

UNIT needs to raise cash over time to continue to diversify itself away from WIN. Right now, WIN makes up ~70% of the company’s business. Management has stated plans to get this ratio down to ~50% in the short-term; however, this transformation will require additional acquisitions and I think it’s ludicrous to think that UNIT management will be able to find investments with cap rates that exceed its current dividend yield.

Because of this scenario, one could argue that a dividend cut for UNIT would actually be a good thing for the long-term. It might enable it to continue to diversify away from WIN exposure and grow its asset base. Michael Boyd wrote an article focused on this possibility today. This general point was that a distribution cut for UNIT is the right move for management to make. Once again, in the comment section, there are members on both sides of the fence of this issue. There are many question marks when it comes to UNIT in the present, but the one thing that is clear is that the company’s 16% has surely caught the eye on SA’s dividend and income community.

My portfolio’s rule regarding dividend cuts is cut and dry. A cut equals a sell, without exception. Well, being that an investment in UNIT breaks just about half of my stock screening rules anyway, I will be in wait and see mode if UNIT should slash its dividend. This is a small enough position for me that in the event of sudden weakness, it won’t do significant damage to my portfolio’s overall returns. On the flip side of this coin, UNIT’s yield is high enough to move the needle a bit in terms of my annual income expectations. Due to its extremely high yield, this ¼ position in UNIT is currently scheduled to generate the same amount of income as a typical full position with a “normal” yield for my portfolio would over a year in just a couple of quarters (my portfolio’s overall yield is just a tad above 2%).

Investing in distressed assets has led to riches for investors throughout the history. It has also lead to ruins. I’m not saying that I’m smart enough to pick and choose the winners, but I have seen enough bullish opinions from well-respected analysts/contributors to inspire me to make a small bet on UNIT. I don’t think these shares are for the faint of heart. There are so many rumors flying around regarding WIN that attempting to trade in and out of UNIT on a daily basis seems to be a fool’s errand as well. I plan on stashing the small position of UNIT shares that I bought at $ 15 away and accepting the income that they generate for my portfolio, whatever that may be. I bought one day before UNIT went ex-dividend, meaning that I’ve already captured one $ 0.60 payment. I don’t know how many more investors can expect at this level, but if management is able to maintain the dividend, I expect to do quite well here.

Although I realize that I may end up having to stay in this name for awhile depending on what happens moving forward, I don’t think this is a buy and forget type of stock. It’s both a speculative income play as well as a turnaround play. If it turns around, I think it will turn around quickly. I will continue to monitor the business and management’s attempt at diversifying its customer base. If management cuts the dividend I’m sure the share price will suffer and at that point I’ll be in it for the long-haul, hoping for a Kinder Morgan-like recovery post dividend cut. If the market receives better than expected news out of WIN and UNIT bounces drastically, I will be happy to sell my shares, taking my profits large short-term profits (these shares are held in a tax-advantaged account so that I don’t pay taxes on that hefty dividend).

Disclosure: I am/we are long UNIT, OHI.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Tech

A brief history of the EV industry

The birth of the EV industry is usually pegged to the summer of 2008 when crude oil prices peaked above $ 140 per barrel. Tesla (TSLA) was a few months from launching its electric Roadster and I started blogging on battery investing for Seeking Alpha.

At the time, the DOE and everybody in the auto industry pegged their EV hopes and dreams on lithium manganese oxide, or LMO, and lithium iron phosphate, or LFP, batteries. The reasons were simple. Both chemistries had great performance profiles for EVs and both chemistries were made using cheap and abundant raw materials – lithium, manganese and iron.

The sole renegade was Tesla, which planned to use consumer grade cells and a nickel-cobalt chemistry instead of more costly automotive grade LMO and LFP cells. Tesla’s theory, which had more than a touch of genius, was that using consumer grade cells would allow it to over-build its battery packs to improve safety and slow cell degradation while pitching a 300-mile range with neck snapping acceleration as major advantages, even though most Tesla owners would crawl through city traffic with the rest of us and average less than 35 miles of daily driving.

While Tesla’s electric muscle cars have always been energy, emissions and economic nightmares, the sales pitch resonated with one percenters who were drawn to Tesla’s richly subsidized eco-bling like moths to a flame.

The consumer response was so strong that most players in the EV industry are moving away from LMO and LFP batteries and embracing high-energy nickel manganese cobalt, or NMC, batteries.

The EV industry’s transition away from cheap and abundant raw materials in favor of costly nickel and cobalt will not end well.

The cobalt cliff

Over the last 18 months I’ve repeatedly cautioned readers that intrinsic and unavoidable constraints on cobalt production would create insurmountable obstacles to the widespread deployment of long-range EVs with high energy nickel-cobalt batteries.

While the problems will be shared by all automakers that have launched or plan to launch EV products with nickel-cobalt batteries, most will be able to roll with the punches and adjust their business models to accommodate the vulgar exigencies of cobalt production dynamics.

Tesla stands alone as an EV manufacturer that cannot implement its business plan, or for that matter continue in business, without easy access to unlimited cobalt supplies. In my view, Tesla’s failure to secure a robust and reliable cobalt supply chain before starting construction for its Nevada Gigafactory is the biggest OOOPs in the history of supply chain management.

Frankly, Tesla’s cobalt predicament reminds me of the cliff sequence in the 1991 movie classic Thelma and Louise:

In March 2016, the first of my five prior articles on the Cobalt Cliff explained:

• Cobalt is an essential raw material for all high-energy lithium-ion cells and the battery industry accounted for 44% of cobalt consumption in 2015.
• Cobalt is also an essential raw material for superalloys, machine tools, catalysts, pigments and other high value products that accounted for 56% of cobalt consumption in 2015.
• Roughly 90% of cobalt supplies come from copper (60%) and nickel mines (30%) that produce cobalt as a minor byproduct.
• The other 10% comes from primary cobalt mines (2%) and “artisanal” cobalt mines in the DRC (8%) that reportedly rely on slave and child labor.
• Reliance on a byproduct is incredibly risky because availability always is tied to demand for the primary product; in this case nickel and copper.
• Without rock solid supply contracts, Gigafactory owners will find themselves between a rock and a hard place as they try to outbid other companies that need cobalt for higher value products.
• The likely price of supply chain failure will be business failure because you cannot make EVs without batteries and battery Gigafactories cannot operate profitably without robust and reliable raw material supply chains.

In the last 18 months, several events, developments and reports have corroborated, ratified and reinforced my original thesis, including:

Until recently, I wasn’t completely convinced that leading automakers were truly committed to their EV initiatives. While the Chinese are deadly serious about promoting new energy vehicles as a mean of reducing pollution in their mega-cities, Nissan (OTCPK:NSANY), Tesla, GM (NYSE:GM) and BMW were the only western automakers with credible EV programs. That dynamic changed after dieselgate when VW came to the EV party with a vengeance, partly as penance for past sins and partly in response to Germany’s increasing political support for electric drive.

In its “Global EV Outlook 2017” the International Energy Agency summarized the electric car ambitions of the world’s automakers as follows:

With an average cobalt content of 8 kg per car, the 41,200 tonnes available for EV batteries in 2025 will only support the manufacture of 5.15 million EVs, a little over half of the aspirational totals set forth above.

At this point I have to believe the EV revolution has entered a melee phase where there won’t be enough cobalt to satisfy everybody’s needs and anyone who wants to play the game will have to stand toe-to-toe and compete for cobalt supplies with China Inc. and six of the world’s ten largest automakers.

I don’t think Tesla is up to the business challenge establishing robust and reliable supply chains in the face of relentless competition from the big boys.

I know it lacks the financial strength to stand toe-to-toe with the big boys in a bidding war.

I can almost hear Elon Musk and JB Straubel reprising a Tesla version Thelma’s last conversation with Louise.

Conclusion

Benjamin Graham, the father of value investing, once explained that in the short run, the market acts like a voting machine – tallying up which firms are popular and unpopular. But in the long run, the market acts like a weighing machine – assessing the substance of a company. Tesla is clearly a voting machine stock that will maintain an irrational value until the market starts to act like a weighing machine. Since I expected Tesla’s stock to crumble when it was trading in the $ 30s, I haven’t a clue when that will happen.

This article focuses on a third reason why I think Tesla’s business model is fatally flawed and its inherent investment value is zero. I will discuss others in future articles. The biggest challenge with macro issues like the cobalt supply and dynamics discussed in this article is that it’s almost impossible to predict when an unavoidable outcome will occur. In my experience, being right about an unavoidable outcome too early isn’t much better than being wrong.

Since Elon Musk is the most talented stock promoter I’ve ever seen and I’ve never shorted a stock, I have no advice to offer Tesla bears. That being said, I wouldn’t own an un-hedged long position in Tesla because the downside risk is enormous.

From my perspective the only safe place to watch this circus is the sidelines.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Tech

The Power Factors System is the backbone of my research service, The Data Driven Investor. It’s essentially a quantitative ranking system that selects stocks based on three powerful and time-proven return drivers: financial quality, valuation, and momentum.

Multiple academic studies have proven that companies exhibiting strong numbers in these three areas tend to beat the market in the long term, and my own backtesting work confirms that the Power Factors Systems can generate impressive performance over time.

The specific details behind the system are not particularly important, the main idea is using a combination of indicators and ratios to select companies with strong metrics in these main areas. Among others, the Power Factors System includes the following metrics:

• Financial quality: the system looks for companies with superior profitability on sales, considering ratios such as gross profit margin and free cash flow margin. In addition, financial quality includes metrics based on return on capital, such as return on investment and return on assets.
• Valuation: this covers classical valuation ratios like price to earnings, and price to free cash flow, among several other metrics based on similar concepts.
  
• Momentum: the system picks companies that are outperforming expectations, and it also looks for stocks that are doing better than the broad market. In a nutshell, we want companies that are delivering performance numbers above Wall Street forecasts, and we also want the stock price to be reflecting such outperformance.

An equally-weighted portfolio comprised of the 50 best-ranking companies in the system produced an impressive annual return of 26.39% since 1999. By comparison, the S&P 500 produced a far more modest return of 3.77% per year over the same period.

In other words, a $ 100,000 position in the S&P 500 back in 1999 would currently be worth nearly $ 199,100, while the same amount of money invested in the Power Factors portfolio would be worth an exponentially larger sum of $ 7.8 million.

Data and chart are from Portfolio123, and the full list of companies in the system is available to subscribers in The Data Driven Investor.

The ranking system is based on a stock universe that excludes over-the-counter stocks in order to guarantee a minimum size and liquidity levels. Nevertheless, most stocks in the system are relatively smaller than those in the S&P 500, and in many cases far more volatile.

Interestingly, Gilead (GILD) is a noteworthy exception. The company has a market capitalization value of more than $ 109.6 billion, and it ranks remarkably well across the three dimensions in the Power Factors System. These particularities make of Gilead a particularly intriguing name among the stocks selected by the quantitative model.

Case Study: Gilead

Gilead is a leading player in the biotech space. The company is focused on life-threatening infectious diseases, with a big presence in treatments for HIV, hepatitis B, and hepatitis C. Gilead has made a series of acquisitions to expand its portfolio in cardiovascular diseases and Cancer over the past several years. More recently, the company made a big move with the acquisition of Kite Pharma (KITE) for $ 11.9 billion in cash. This deal could provide a big boost to Gilead in cell therapy and oncology treatments.

The business is under pressure due to lower sales and increasing competition in Hepatitis C (HVC) products.

On the other hand, Gilead has a promising pipeline of new developments across different areas, and this should drive increased revenue growth over the years ahead.

Importantly, the company has an impressive track record of financial performance over the long term, and profitability levels are considerably above-average. The following table compares key financial metrics for Gilead vs. other big biotech companies, such as Amgen (OTC:AMGM), Celgene (CELG), and Biogen (IBB).

The numbers are quite clear, Gilead ranks above the competition across all of the five indicators: sales growth over the past five years, return on assets, return on investment, operating margin, and net margin.

Financial performance over the years ahead will depend on variables such as demand for Gilead’s new products, and this is always a source of uncertainty. Nevertheless, the company’s track-record and current performance are a positive reflection on its management team and its ability to deliver attractive returns for shareholders.

In terms of valuation, Gilead stock is fairly conveniently priced, if not downright undervalued. The stock trades at a price to earnings ratio around 9.15 times earnings over the past year. This is a huge discount versus the average company in the S&P 500, which trades at a price to earnings ratio around 21.5.

Looking at valuation ratios in comparison to industry peers, Gilead also looks quite cheap in terms of price to earnings, forward price to earnings, price to free cash flow, and price to sales.

Offering a similar perspective, the following chart shows how Gilead’s valuation has evolved over the past several years, and current entry price looks quite compelling by historical standards in terms of price to earnings, price to free cash flow, and enterprise value to EBITDA.

GILD PE Ratio (ttm) data by YCharts

The bottom line is that Gilead stock is substantially cheap, be it in comparison to the broad market, when compared to industry peers, or by the company’s own historical standards.

Momentum is favoring the bulls. Both revenue and earnings came in above Wall Street expectations last quarter, and analysts are adjusting their earnings forecasts to the upside. The average earnings estimate for Gilead in 2017 was $ 8.35 per share 90 days ago, and it has steadily increased towards $ 8.78 currently.

Stock prices don’t just reflect fundamentals, expectations about those fundamentals are tremendously important. When expectations are on the rise, this generally means that stock prices are rising too. On the back of increasing earnings forecasts, Gilead stock has substantially outperformed the S&P 500 index over the past several months.

GILD data by YCharts

Past performance does not guarantee future returns. However, profitability metrics, valuation, and momentum are all positive forces for investors in Gilead on a forward-looking basis.

Disclosure: I am/we are long GILD.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Tech

One of the most popular jobs on TaskRabbit, a service that lets you hire workers for quick gigs, is assembling Ikea furniture. So perhaps it’s no surprise that the Swedish retail giant has reportedly acquired the startup for an undisclosed price.

TaskRabbit has only a few dozen full-time employees, but it is a platform for a large number of independent contractors who help customers with all sorts of errands, handymen tasks and, of course, furniture assembly.

According to tech news site Recode, Ikea will treat TaskRabbit, which is reportedly profitable, as an independent subsidiary and keep on its CEO Stacy Brown-Philpot. Recode sees the deal as a strategic acquisition at a time of rapid change in the world of retail and home delivery:

TaskRabbit has received investments from a number of prominent venture capital firms, including Shasta Ventures, Lightspeed Venture Partners and Founders Fund.

Currently, customers are able to hire “rabbits” in around 40 U.S. cities.

TaskRabbit is one of the most high profile of the so-called “gig economy” companies, which connect customers with workers on an independent contractor basis. Other such companies include home cleaning service Handy, and the car-hailing services Uber and Lyft.

The “gig” business model is popular with investors because it can grow quickly, and allows companies to try to avoid the costs and legal entanglements of hiring staff. In recent years, however, workers on such services have won several court challenges claiming they are not contractors, but are instead employees.

Ikea did not immediately respond to a request for comment about the acquisition.

Tech

Stocks discussed on the in-depth session of Jim Cramer’s Mad Money TV Program, Tuesday, September 26.

Many investors are wondering what’s wrong with the high flying tech stocks. “Let me give you my take on the dramatic see-saw-like action in the leaders of this market. First, let’s tackle the real cause of the weakness that started last week: technology. Come on, you all know the epicenter here. It’s Apple (NASDAQ:AAPL).” The talk about tepid iPhone 8 sales along with iPhone X coming in November, Apple and its suppliers stocks fell on this news.

Cramer reiterated, “Buy Apple, don’t trade it.” Betting against Apple means getting odds of losing in your favor. Apple makes the best consumer products and is cheap trading at 14 times 2018 earnings compared to consumer product companies and the average stock of the S&P500. It’s worth noting that they have $ 260B in cash as well.

“So, the proximate sell-off cause, at least number one? Apple. Now, I am saying it’s been pretty neutralized. Sure, the stock can go lower, but at these levels, it’s beginning to reflect the failure of the iPhone 9, and that product doesn’t exist,” said Cramer.

The second concern of the tech stocks is the cloud. After Adobe (NASDAQ:ADBE) reported a surprise disappointing quarter, CEO Shantanu Narayen said on the conference call that cloud growth is slowing temporarily. This led to investor panic. Which led to a sell-off in everything cloud-related.

“Remember, the cloud lives in the data center. So if the cloud’s slowing, the data center should be slowing, too. Yet last night Nvidia (NASDAQ:NVDA) announced that its new line of chips won business from three of the largest Chinese data center operators out there: Alibaba, TenCent and Baidu, which are growing like mad. I’ll match those orders against a defeat at Tesla any day of the week,” added Cramer.

“Bottom line? We’re at the end of a good quarter in a good year, and we’re seeing profit-taking and forced selling of the winners while some money’s going into cheaper stocks and, when you put it that way, there’s a kernel of rationality to the entire move,” concluded Cramer. This does not mean selling is over, but the weakness should be used as a buying opportunity.

CFO interview – Red Hat (NYSE:RHT)

The stock of open source software solutions provider Red Hat went up on better than expected earnings and guidance. Cramer interviewed CFO Eric Shander to find out more about the quarter.

Shander said that the cloud world is changing. “Everybody’s talking about it being a hybrid cloud world; I can tell you, whether it’s on-premise, off-premise, they’re looking for optionality, they’re looking for multiple cloud providers, they’re looking for flexibility,” he added.

The company has been investing in application development and emerging technologies which made them $ 150M in the last quarter. The result of their investments can be seen in the strong growth. “Every single industry is transforming. And not only that – there is an IT implication related to that. They’re looking for agile development and that’s where a lot of our technologies are helping enable that,” he added.

Their sales to the Federal government have been strong and not capped as thought by many. Apart from that, Europe is a growing category for the company’s business. Bottom line is that Red Hat has all the components to enable companies to run successfully in the cloud.

Carnival Corp (NYSE:CCL)

The stock of cruise-liner Carnival went down on fears of hurricane damage. However, after the solid earnings and hurricane-adjusted guidance, the stock rallied.

“The most valuable currency in this business isn’t dollars. It’s not gold. It’s not even bitcoin. It’s the benefit of the doubt. When the market realizes that a CEO deserves it, their stock tends to catch fire. If there’s one thing we can bank on, it’s that he’s not going to be tripped up by some storms, even really bad ones,” said Cramer referring to Carnival CEO Arnold Donald.

Donald held the Costa Concordia catastrophe, which resulted in 32 deaths, really well. Apart from that, they got through the Ebola and Zika crisis as well. “Donald seems ready for any disaster and he always seems to have a back-up plan,” said Cramer.

Donald raised the lower end of the guidance despite the hurricane and said that there is only 1% cancellation rate. Cramer calls the weakness in the stock a chance to buy it.

COO interview – VMware (NYSE:VMW)

VMware is the virtualization infrastructure solutions provider which reported a good last quarter and its stock is down just $ 4 from its recent highs. Cramer interviewed COO Sanjay Poonen to hear what lies ahead.

Poonen said that more companies turn to cloud-based solutions in order to cut costs and connect with the digital world. Many companies are adopting to the hybrid approach to the cloud. “As you think about companies and their future, they have to decide how much data center capacity they want to use now. Some people feel they run a data center very well. We can help them modernize it with software, and that’s what we do very well. Some of the companies say, ‘Listen, I don’t want to expand a lot of my data centers. I want to use the new hardware economy.’ The new hardware economy is not just the traditonal players, but Amazon Web Services, Microsoft Azure, Google, IBM,” he added.

VMware is an important name in data center business and they have saved enough energy with their virtualization to power 40% of US homes for a year. They have partnered with AWS to bring scalable solutions to the public and private cloud.

“If you could get the benefit of both worlds, the same tools that you’ve known at VMware for managing and automating that, but get the data center capacity on the fly, that’s what we’ve brilliantly innovated here,” said Poonen. Their aim is to help high-profile customers on-premise and private cloud operations, pair those functions with public cloud etc.

“We think there’s going to be a good amount of spending in the private cloud on-premise and into the public cloud. And we feel we’re one of those quintessential companies that can bridge both sides of that chasm. I think there’s innovation to be had in tech,” concluded Poonen.

Viewer calls taken by Cramer

Emerge Energy Services (NYSE:EMES): Cramer thinks the entire fracking group is a sell.

Supernus Pharma (NASDAQ:SUPN): The migraine business is very competitive.

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Jim Cramer’s Action Alerts PLUS: Check out Cramer’s multi-million dollar charitable trust portfolio and uncover the stocks he thinks could be HUGE winners. Start your FREE 14-day trial now!

Get Cramer’s Picks by email – it’s free and takes only a few seconds to sign up.

Tech

Ford Motor (f) has struck a partnership with Lyft to develop and test self-driving vehicles on the ride-hailing company’s growing network of passengers.

Ford, which announced the partnership in a blog post early Wednesday, said that the goal is to put self-driving vehicles onto Lyft’s ride-hailing network. Just don’t expect to see self-driving Ford vehicles shuttling around Lyft customers anytime soon.

The initial aim is to combine the strengths of each company. For Ford, that’s large-scale manufacturing and development of autonomous vehicles technology, which its partner Argo AI is currently working on. Lyft, meanwhile, has a vast network of customers across the United States that has given the startup greater insight in how people move within cities. Both companies have fleet management and Big Data experience, according to Ford’s blog post written by Sherif Marakby, Ford Vice President of autonomous vehicles and electrification.

Ford, which is now being led by CEO Jim Hackett, hopes to learn how to create self-driving cars that can easily connect with a platform like Lyft’s so they can be quickly dispatched to pick up customers. The automaker also wants to use Lyft’s data (and its own) to determine which cities would be worth launching a self-driving vehicle service and what kind of infrastructure would be needed to properly service and maintain a fleet of self-driving vehicles.

Get Data Sheet, Fortune’s technology newsletter.

Lyft is taking a more collaborative approach to self-driving cars, unlike rival Uber. Earlier this year, Lyft launched an open platform designed to give automakers and tech companies working on self-driving cars access to its ride-sharing network of nearly 1 million rides per day.

And even before the open platform began, Lyft has been locking in partnerships. The company landed its first major partnership in January 2016 with GM, which like Ford also wants to eventually deploy self-driving cars with Lyft’s network.

Lyft has made at least three other partnerships in 2017, including startups Drive.ai and nuTonomy, and Waymo, the Google self-driving car project that spun out to become a business under Alphabet (googl).

Tech

A divided Federal Communications Commission on Tuesday approved a report that found for the first time since 2009 there is “effective competition” in the wireless market, a finding that could help Sprint and T-Mobile to merge.

Reuters reported last week that the wireless carriers are close to agreeing on tentative terms on a deal to merge, a major breakthrough in efforts to merge the third and fourth largest U.S. wireless carriers.

The transaction would significantly consolidate the U.S. telecommunications market and represent the first transformative U.S. merger with significant antitrust risk since President Donald Trump’s inauguration.

FCC Chairman Ajit Pai said “most reasonable people see a fiercely competitive marketplace” citing intense price competition carriers. “This is strong, incontrovertible evidence,” he added.

The FCC approved the report by a 3-2 vote.

A decade ago there were seven major U.S. wireless carriers and today the largest four carriers led by Verizon Communications and AT&T control 98.8% of the U.S. market, according to the FCC.

The FCC would need to approve any merger as would the Justice Department. In 2014, the FCC and Justice Department told the carriers they would not back a merger and the companies abandoned merger talks.

FCC Commissioner Jessica Rosenworcel referenced the potential looming merger.

Get Data Sheet, Fortune’s technology newsletter.

“While this report celebrates the presence of four nationwide wireless providers, let’s be mindful that a transaction may soon be announced that seeks to combine two of these four,” Rosenworcel said.

“For my part, any transaction before us will require someone to explain how consumers will benefit, how prices will not rise, and how innovation will not dissipate in the face of so much more industry concentration.”

She added: “Someone will also need to explain how having fewer potential big bidders in upcoming spectrum auctions will not render our most potent distribution mechanism substantially less powerful.”

FCC Commissioner Mignon Clyburn said the competition report “takes a decidedly myopic view of the ecosystem, and instead focuses only on ‘competition in the provision of mobile wireless services.’ This is like a doctor looking at one organ and pronouncing a patient fit as a fiddle.”

FCC Republican Commissioner Mike O’Rielly praised the report and noted intense wireless competition has led to price cuts, despite carriers investing $ 200 billion in networks over the last six years.

AT&T said in a statement the report shows “with an array of providers, pricing plans and service offerings to choose from, there’s no question that consumers are reaping the benefits of a competitive industry.”

Tech

Last night, CBS finally took the wraps off its oft-delayed new show, Star Trek: Discovery. The two-part debut (“The Vulcan Hello” and “Battle at the Binary Stars”) gave fans the first new TV Trek since Star Trek: Enterprise ceased subspace transmission in 2005. And they were ready for it—last night’s premiere set a single-day record for new signups for CBS’ All Access streaming service. Those who ponied up for an account got both parts of the debut; those who didn’t only got the first episode, which aired on broadcast like something from the 1990s.

Is it worth the money? Was it worth the wait? WIRED writers Adam Rogers and Brendan Nystedt, two life-long Trek fans, boldly agreed to discuss the newest venture. Shields up! Red alert! Spoilers ahead!

Adam Rogers: All right, Brendan. My mind to your mind; my thoughts to your thoughts. How are we feeling? On the one hand, I am glad to have some Star Trek to watch, and while I got all aflutter watching a Klingon armada get #disruptive on the United Federation of Planets, some part of my brain was definitely spinning on the story problems and possible solutions I wrote about last week. Like, it spent two hours on the kind of character deployment and story set-up that Deep Space Nine could’ve knocked out in a single cold open. And this definitely wasn’t explore-strange-new-worlds-seek-out-new-life Trek. This was dark Starfleet at war, with a captain who meets her Kobayashi Maru and a promising officer who ends up sentenced to life in the stockade for mutiny.

Hey, relatedly, there is no piece of culture I can embrace wholly where Michelle Yeoh dies an ignominious death. I didn’t like it in Sunshine and I don’t like it here. She was in Heroic Trio, for pete’s sake. She coulda taken that Klingon.

Brendan Nystedt: I saw it coming from a lightyear away but still, ouch. What I didn’t see coming was that they’d also kill the Klingon cult leader, T’Kuvma.

Overall, even with my quibbles over the first two episodes, I’m still holding off judging it entirely. I think the two-parter backstory was an intriguing way to open this new show but it also worries me that the CBS All Access “first taste is free” thing means we don’t have a sense of what the bulk of the story will involve because the network is putting a lot of window dressing in the episodes people can watch for free. Maybe fans who didn’t like the introduction would learn to love where it pivots to next.

Was the opening dark? Yeah, it was. I appreciated that Michael Burnham wanted to get out there and check out the OUO (object of unknown origin) and that they bring up the balance between exploration and war. I think that darkness is something this time period can exploit more since we’re firmly in the “cowboy diplomacy” days of Kirk.

Even with my Star Trek brain fully engaged (heh), I was surprised that the drama worked on me. When Burnham disables the captain with a perfectly-executed Vulcan Nerve Pinch, I was stunned. When that Klingon ship took out the USS Europa while de-cloaking, I recoiled. As much as I was scrutinizing the uniforms and the Klingon makeup, the show worked for me on a basic level.

Rogers: Yes, yes. Me too.

But be honest. If the set up for this show was that it is set 100 years after Voyager rather than a decade before TOS, would it be any different? In what sense is this a prequel? The Klingons do not look like Klingons we have seen. The instrumentation on the ships is new. The uniforms are new. The pew-pew of the phasers and photon torpedoes are new. The Federation starships don’t look anything like the ships of the era—were no Constitution-class starships deployed at that point? Why don’t the Bussard collectors have the light-up pinwheel spinny effect? Why don’t the warp nacelles have to be as far from the crewed parts of the ships? What are these Star Wars communications holograms doing here?

Flip side, I loved seeing the rethought handheld phasers and communicators; they really are elegant. And I liked the bridge noises being the TOS bridge noises. But other than knowing Sarek is Spock’s dad, what’s prequel-y about this?

This isn’t necessarily a complaint. I like the story so far. I just feel about it the same way I did about the reboot movies’ alternate timeline. As a fan, I don’t need it. I dunno; maybe it’s all a set up for the last shot of the season being a TOS-authentic Constitution-class Enterprise heading off on Captain Pike’s five-year mission.

Nystedt: The look of the show is something I’m still reconciling, but Trek has been here before. Whether it’s the retcons of Star Trek: Enterprise or especially the 1979 Motion Picture, designs and tech change a bit to suit the time. I’ve never believed that the movie Enterprise was the same Enterprise as the TV show, and yet it’s known to be a “refit.” Unless it’s the ship of Theseus, the movie Enterprise just can’t logically be the same as the one from the show!

I digress … I’m also unsure why they decided to make this a prequel, especially if it’s going to try to do its own thing. I’d understand if it were for the sake of fanservice but as of yet there have been a limited number of references to the universe. I’m hoping they’ll at least hint at conforming to a style closer to what we know from TOS, but I’d be OK if that were a reinterpretation, too.

The Klingons were another sticking point: I was cool with showing this rogue gang of Kahless-worshippers but when the rest of the house leadership Skypes in to T’Kuvma’s sarcophagus ship I was disappointed. Enterprise tried to give the makeup and characterization changes of past Klingons some kind of sense, but I felt like even that explanation couldn’t remotely cover for why all these Klingons looked different. At least they were completely subtitled—that I really dug.

Rogers: I liked hearing all that Klingon, too—and then seeing the universal translator kick in when T’Kuvma called the Shenzhou. Cool starship names all around, actually. I loved the references to a USS Yeager.

Speaking of, though, you make a good point that we don’t even really know what the story of this show will be. We haven’t yet discovered the Discovery, presumably the ship we’ll spend the bulk of the season on. And despite my gripes, I’m psyched for it. I don’t know if a lot of people will spring for this show, but I’m glad it exists, and I’m looking forward to the season. Glad there’s more Star Trek in my life.

Can I tell you a separate story? I know that the first ep did great in the ratings, 9.6 million people. And CBS is staying that it got a big pulse of new subscribers to All Access, but not giving out numbers. So OK. Last night I watched the first ep on my DVR and then went to subscribe to All Access to watch the second.

First I tried to do that via Apple TV, but apparently my Apple TV is so old that it doesn’t really know how to do two-factor authentication. You have to get a verification code from your phone and then type it in along with your password, apparently, but timed with the deftness of a longtime gamer, which I am not. I gave up.

I went to the app on my iPad, which seemed ready to let me sign up, until it decided that my zip code was invalid. Several times. (Narrator: It was not invalid.) Finally I logged all the way out and then logged in with Google, which somehow convinced the app my zip code was real. At which point I learned that payment was going via my Apple account, which makes me wonder why I’m paying CBS instead of just Apple. This all took about 45 minutes to figure out, by the way.

This is not an onboarding process anyone should be proud of, is my point.

Nystedt: That’s a huge bummer! I had signed up for my All Access account earlier in the day through the website, but that wasn’t ideal either. Like, I’m happy to give my money over to Trek, but I wish it were more streamlined. One disappointment for me is that even though Discovery is a launch title for the service, the rest of the Star Trek offerings are a mix of HD and non-HD. Voyager and DS9 haven’t been restored (and they might not ever be) but only about a half of The Next Generation and none of the original show appear in their HD incarnations.

And so, even though we get a new show, Trek continues to get dissed. It’s a treasure of global popular culture but it just doesn’t get the respect it deserves, whether it’s from CBS or Paramount. I’d go so far as to say that MGM is giving Stargate better treatment with its upcoming streaming service, offering up just that show’s back catalog (and movies!) for a flat rate plus the promise of new content to come.

And when you have to look to Stargate to find a decent single-franchise online service, you know it’s gonna be a long road… But, I got faith of the hearrrttttt!!!

Rogers: No Sto’vo’Kor for you, pal. Eesh. Anyway, no matter what these streaming services show or don’t show, they can’t take the sky from me.

Nystedt: Awww man I was looking forward to seeing my ancestors in the afterlife. One thing I enjoyed seeing yesterday was how fandom reacted to the show. Definitely check Twitter for #OnFleet—some excellent, funny commentary there, particularly from nerds of color and women fans.

So, next Sunday, we finally get to see the titular ship. Somehow, Burnham gets out of her life sentence, and I guess we’ll get more Saru, too. I’m on the Discovery train, are you feeling optimistic?

Rogers: I’m with Ambassador Spock. There are always possibilities.

Nystedt: LLAP and tune in next week.

Tech

Global accountancy firm Deloitte has been hit by a sophisticated hack that resulted in a breach of confidential information and plans from some of its biggest clients, Britain’s Guardian newspaper said on Monday.

Deloitte—one of the big four professional services providers—confirmed to the newspaper it had been hit by a hack, but it said only a small number of its clients had been impacted.

The firm discovered the hack in March, according to the Guardian, but the cyber attackers could have had breached its systems as long ago as October or November 2016.

The attack was believed to have been focused on the U.S. operations of the company, which provides auditing, tax advice, and consultancy to multinationals and governments worldwide.

“In response to a cyber incident, Deloitte implemented its comprehensive security protocol and began an intensive and thorough review including mobilizing a team of cybersecurity and confidentiality experts inside and outside of Deloitte,” a spokesman told the newspaper. “As part of the review, Deloitte has been in contact with the very few clients impacted and notified governmental authorities and regulators.”

A Deloitte spokeswoman declined immediate comment, saying that the firm would issue a statement shortly.

Tech

In recent years, Silicon Valley has struggled to respond to a bruising series of sex scandals, to tone down what’s sometimes seen as a childish “tech bro” culture, and to rebalance a workforce in which as few as one-fourth of technical jobs are held by women.

But some workers in the Valley think the real victim here is men.

According to a new report in the The New York Times, more and more men in Silicon Valley are joining online and offline “men’s rights” groups that regard gender diversity efforts as an attack on them. Some believe that the standards for what qualifies as harassment are too stringent, and that gender equality in the industry is an unreasonable goal.

The rise of such beliefs, or at least a rising willingness to express them, may be best embodied by James Damore, the engineer whose memo deriding the abilities of women got him fired from Google last month. While many tech leaders were scornful of Damore and his worldview, expressions of support did come from heavyweights including Y Combinator founder Paul Graham. Since his firing, Damore has asserted his own victimhood at the hands of Google leadership.

Get Data Sheet, Fortune’s technology newsletter.

Nvidia engineer James Altizer, leader of a group called Bay Area Fathers’ Rights, told the Times that heightened awareness of discrimination against women is actually a “witch hunt” targeting men, spearheaded by “dangerous” human resources departments motivated by “zealotry.” According to the Times, Altizer’s views were shaped by a divorce. Another group that the Times says is growing is known as Mgtow, or “Men Going Their Own Way,” a “men’s separatist” movement that resists commitment and children.

Altizer may have missed the irony of his critique. In the Middle Ages, an estimated 60,000 accused witches were executed throughout Europe, and many thousands more were tortured. More than 70% of victims were women, and at least one historian has described the witch hunts as “genderized mass murder.”

The reactionary dynamic of the men’s rights movement is reminiscent of the rising visibility of white supremacists as members of racial minorities gain firmer social and economic footing. A few men in the Valley have even filed employment discrimination lawsuits linked to gender, directly paralleling a long string of lawsuits targeting affirmative action policies at universities.

Fortune has reached out to Nvidia seeking a response to Altizer’s comments.

Tech

Ryan Sweeney is a content strategist at WorkWise, the provider of OnContact customer relationship management (CRM) software. We asked Ryan why small business owners should consider implementing a CRM software solution. Here’s what he shared:

Running a small business means facing an ongoing series of ever-evolving challenges. We’re fortunate that entrepreneurs have more tools at their disposal than ever before to meet such challenges efficiently and effectively. One particularly powerful tool is CRM software, which adds value by improving sales, marketing and customer service efforts.

Why Small Business CRM Rocks

When small business owners ask if the time is right to implement a CRM solution in their company, here’s what I share with them.

Flexibility of the Cloud: CRM software was once a hefty investment within reach of only large enterprises with deep pockets. That’s no longer the case. Today, small businesses can leverage the great perks CRM offers through a cloud or “hosted” software deployment. This slick solution eliminates the necessity of having an extensive, in-house IT department. Small business owners with cloud deployment also won’t have to worry about software maintenance issues including backup, updates and other expenses commonly associated with an on-premise deployment. And in some cases, CRM solutions are even available on a month-to-month basis without a long-term commitment.

Helps Get—and Keep—You Organized: Staying organized while trying to grow your business is a never-ending effort. You might even be relying on a low-tech system, such as Excel spreadsheets, to carry your business forward. If I just described your methodology, it’s time to reconsider. Excel spreadsheets don’t offer the foolproof safety and flexibility that CRM software inherently provides. You can’t take them with you everywhere on a mobile device or tablet, and keeping data updated and de-duplicated is a feat in and of itself. CRM software does all of that for you and more. And you get mobility, which is pretty awesome, too.

CRMs Offer Features Galore: Conduct some research and you’ll likely identify a CRM software vendor with the ideal solution for your business. There’s a strong likelihood that it will come with all the features your business could ever want because modern CRM solutions are packed with diverse and incredibly useful productivity tools. Some specific features to look for include a centralized database, email designer tools, email integrations, accounting and finance functionality and integrations, call center software, and workflow automation. And that’s just the tip of the productivity iceberg.

CRM Tools You Can Leverage

Now that I’ve exposed the benefits of CRM software, let’s take a closer look at some specific tools you can leverage to stay in control of your growing business.

Sales Automation: This is what you think of when you think CRM software: It provides small businesses with the visibility they need to visualize their pipeline, create influential forecasts for future sales and grow their business for the long term. It can also track business leads and new prospects that arrive through your company’s website, create new tasks for you (and your sales team, if applicable) and remind you when it’s time to connect with existing customers and prospects.

Email Marketing Functionality: If it’s available, it’s often wise to invest in a CRM solution that’s fully integrated with a marketing automation tool. This gives you email marketing functionality, which can be a formidable resource for small businesses. With email marketing at your disposal, you benefit from features like an email designer that enables you to build quality, unique email designs to dazzle prospects and customers. In addition, you’ll have the capability to segment email automation efforts based on target audiences, locations or other parameters which you set. Utilizing this feature saves time and amps up your cool factor.

Landing Page Integration: You might not expect such an advanced feature from a traditional CRM tool, but the best ones now include landing page integration. Here’s how it works: As prospects land on your website, you can track their actions and see which pages they linger on the longest. If they submit a form through your website, you’ll capture all of their relevant customer information, which is then automatically stored in your CRM database. This data will be accessible to your entire team, from sales to marketing. Your lead generation efforts just got way more refined, benefitting your business’ bottom line.

Tech

Most online brands we know and love have been around for long, have likely lived through their fair share of growing pains when it comes to providing a streamlined customer experience (CX) that keeps people coming back.

Many online businesses make the detrimental mistake of reducing CX to a supplementary factor in their brand image. In reality, it should be a cornerstone of the values they represent – take Zappos for example. To nearly everyone familiar with the digital landscape, a superior CX is practically synonymous with their reputation.

For the companies still trying to find the right formula that works for them, here are three ways to make sure your brand delivers outstanding CX.

1. Optimize Customer Journeys

Fine-tuning your customer journey is an ongoing process that requires constant attention. This involves knowing exactly how people go from being unaware of what you offer, to becoming loyal customers. Your analytics reveal ‘what’ your customers are doing, and ‘which’ of your digital channels are working to bring in new customers.

However, this data will only reveal so much. When it comes to enhancing the customer journey, the most effective advice actually comes straight from the source – the voice of the customer.

For this purpose, Medallia, a global CX management software company, released The Digital Voice of Customer Toolkit. This top-to-bottom resource focuses on the best ways for measuring and improving the customer experience across websites, mobile web, and mobile applications. A digital VoC program allows you to systematically engage with your customers, and gain comprehensive insights in real-time, on users’ interactions with your digital touchpoints.

From here, you will be able to identify the most effective methods to engage with the audience and gain meaningful feedback, which can take your business to the next level. Ultimately, a VoC program pinpoints the deeper meaning of the ‘why’ and uses feedback straight from the source. This feedback compliments the data that you collect from your analytics provider to improve customer satisfaction and experience.

2. Emphasize Persona

Your brand persona is how people see you and what impression you leave them with, after an interaction.

The harsh truth of conducting business digitally is that your competition is no longer confined to a specific geographic location. Brands are now faced with competition across their entire industry. That being said, brand loyalty these days is commonly built and developed based on the persona you project.

These unique characteristics should be present throughout your whole strategy – from your marketing messages to the actual products. Even more, it needs to speak to the existing and prospective customers in a manner they can easily relate to.

A great place to start with is archetypes. If your brand was a human being:

· What do they sound like?

· What do they look like?

· What is their general outlook on life?

· What is the most important thing to them?

· How do they interact with others?

For these answers, social media and web monitoring tools come in very handy. Tools like Brandwatch allow you to track relevant keywords, brand names, industries, and more to give you a better idea of who your ideal customers are, and how to speak their language.

3. Always A/B Test

The digital world is one that is constantly changing. This is due to the rapid advancement of the internet, e-commerce technology, and user mindsets. As a result, you cannot expect a formula for a good CX to work forever. As time goes on, you will need to refine the smaller details of how people use your digital channels.

This process involves personalizing website engagement at all stages, from knowing customer preferences all the way to the checkout. The more you experiment and gauge results, you’ll quickly learn that even the tiniest tweaks can make a world of difference. Perhaps the most frequently-tested aspects in strategies to increase website conversion rate are tweaking copy and calls to action (CTA) buttons. Doing something as simple as altering the color, text, or placement can boost your conversions.

Tools like Optimizely allow you to set up different versions of your website and landing pages to run customized tests that determine which approaches work well, and which ones can be scrapped. The program is equipped with heat maps, behavioral targeting, usability testing, and more to ensure your CX is constantly evolving alongside user preferences.

By consistently A/B testing variations across your website, you can increase online sales by up to 20%!

Tech