Category: UNCATEGORIZED

Does quantum computing really exist? It’s fitting that for decades this field has been haunted by the fundamental uncertainty of whether it would, eventually, prove to be a wild goose chase. But Google has collapsed this nagging superposition with research not just demonstrating what’s called “quantum supremacy,” but more importantly showing that this also is only the very beginning of what quantum computers will eventually be capable of.

This is by all indications an important point in computing, but it is also very esoteric and technical in many ways. Consider, however, that in the 60s, the decision to build computers with electronic transistors must have seemed rather an esoteric point as well. Yet that was in a way the catalyst for the entire Information Age.

Most of us were not lucky enough to be involved with that decision or to understand why it was important at the time. We are lucky enough to be here now — but understanding takes a bit of explanation. The best place to start is perhaps with computing and physics pioneers Alan Turing and Richard Feynman.

‘Because nature isn’t classical, dammit’

The universal computing machine envisioned by Turing and others of his generation was brought to fruition during and after World War II, progressing from vacuum tubes to hand-built transistors to the densely packed chips we have today. With it evolved an idea of computing that essentially said: If it can be represented by numbers, we can simulate it.

That meant that cloud formation, object recognition, voice synthesis, 3D geometry, complex mathematics — all that and more could, with enough computing power, be accomplished on the standard processor-RAM-storage machines that had become the standard.

But there were exceptions. And although some were obscure things like mathematical paradoxes, it became clear as the field of quantum physics evolved that it may be one of them. It was Feynman who proposed in the early 80s that if you want to simulate a quantum system, you’ll need a quantum system to do it with.

“I’m not happy with all the analyses that go with just the classical theory, because nature isn’t classical, dammit, and if you want to make a simulation of nature, you’d better make it quantum mechanical,” he concluded, in his inimitable way. Classical computers, as he deemed what everyone else just called computers, were insufficient to the task.

The problem? There was no such thing as a quantum computer, and no one had the slightest idea how to build one. But the gauntlet had been thrown, and it was like catnip to theorists and computer scientists, who since then have vied over the idea.

Could it be that with enough ordinary computing power, power on a scale Feynman could hardly imagine — data centers with yottabytes of storage and exaflops of processing — we can in fact simulate nature down to its smallest, spookiest levels?

Or could it be that with some types of problems you hit a wall, and that you can put every computer on Earth to a task and the progress bar will only tick forward a percentage point in a million years, if that?

And, if that’s the case, is it even possible to create a working computer that can solve that problem in a reasonable amount of time?

In order to prove Feynman correct, you would have to answer all of these questions. You’d have to show that there exists a problem that is not merely difficult for ordinary computers, but that is effectively impossible for them to solve even at incredible levels of power. And you would have to not just theorize but create a new computer that not just can but does solve that same problem.

By doing so you would not just prove a theory, you would open up an entirely new class of problem-solving, of theories that can be tested. It would be a moment when an entirely new field of computing first successfully printed “hello world” and was opened up for everyone in the world to use. And that is what the researchers at Google and NASA claim to have accomplished.

In which we skip over how it all actually works

Much has already been written on how quantum computing differs from traditional computing, and I’ll be publishing another story soon detailing Google’s approach. But some basics bear mentioning here.

Classical computers are built around transistors that, by holding or vacating a charge, signify either a 1 or a 0. By linking these transistors together into more complex formations they can represent data, or transform and combine it through logic gates like AND and NOR. With a complex language specific to digital computers that has evolved for decades, we can make them do all kinds of interesting things.

Quantum computers are actually quite similar in that they have a base unit that they perform logic on to perform various tasks. The difference is that the unit is more complex: a qubit, which represents a much more complex mathematical space than simply 0 or 1. Instead you may think of their state may be thought of as a location on a sphere, a point in 3D space. The logic is also more complicated, but still relatively basic (and helpfully still called gates): That point can be adjusted, flipped, and so on. Yet the qubit when observed is also digital, providing what amounts to either a 0 or 1 value.

By virtue of representing a value in a richer mathematical space, these qubits and manipulations thereof can perform new and interesting tasks, including some which, as Google shows, we had no ability to do before.

A quantum of contrivance

In order to accomplish the tripartite task summarized above, first the team had to find a task that classical computers found difficult but that should be relatively easy for a quantum computer to do. The problem they settled on is in a way laughably contrived: Being a quantum computer.

In a way it makes you want to just stop reading, right? Of course a quantum computer is going to be better at being itself than an ordinary computer will be. But it’s not actually that simple.

Think of a cool old piece of electronics — an Atari 800. Sure, it’s very good at being itself and running its programs and so on. But any modern computer can simulate an Atari 800 so well that it could run those programs in orders of magnitude less time. For that matter, a modern computer can be simulated by a supercomputer in much the same way.

Furthermore, there are already ways of simulating quantum computers — they were developed in tandem with real quantum hardware so performance could be compared to theory. These simulators and the hardware they simulate differ widely, and have been greatly improved in recent years as quantum computing became more than a hobby for major companies and research institutions.

To be specific, the problem was simulating the output of a random sequence of gates and qubits in a quantum computer. Briefly stated, when a circuit of qubits does something, the result is, like other computers, a sequence of 0s and 1s. If it isn’t calculating something in particular, those numbers will be random — but crucially, they are “random” in a very specific, predictable way.

Think of a pachinko ball falling through its gauntlet of pins, holes and ramps. The path it takes is random in a way, but if you drop 10,000 balls from the exact same position into the exact same maze, there will be patterns in where they come out at the bottom — a spread of probabilities, perhaps more at the center and less at the edges. If you were to simulate that pachinko machine on a computer, you could test whether your simulation is accurate by comparing the output of 10,000 virtual drops with 10,000 real ones.

It’s the same with simulating a quantum computer, though of course rather more complex. Ultimately however the computer is doing the same thing: simulating a physical process and predicting the results. And like the pachinko simulator, its accuracy can be tested by running the real thing and comparing those results.

But just as it is easier to simulate a simple pachinko machine than a complex one, it’s easier to simulate a handful of qubits than a lot of them. After all, qubits are already complex. And when you get into questions of interference, slight errors and which direction they’d go, etc. — there are, in fact, so many factors that Feynman decided at some point you wouldn’t be able to account for them all. And at that point you would have entered the realm where only a quantum computer can do so — the realm of “quantum supremacy.”

Exponential please, and make it a double

After 1,400 words, there’s the phrase everyone else put right in the headline. Why? Because quantum supremacy may sound grand, but it’s only a small part of what was accomplished, and in fact this result in particular may not last forever as an example of having reached those lofty heights. But to continue.

Google’s setup, then, was simple. Set up randomly created circuits of qubits, both in its quantum computer and in the simulator. Start simple with a few qubits doing a handful of operational cycles and compare the time it takes to produce results.

Bear in mind that the simulator is not running on a laptop next to the fridge-sized quantum computer, but on Summit — a supercomputer at Oak Ridge National Lab currently rated as the most powerful single processing system in the world, and not by a little. It has 2.4 million processing cores, a little under 3 petabytes of memory, and hits about 150 petaflops.

At these early stages, the simulator and the quantum computer happily agreed — the numbers they spat out, the probability spreads, were the same, over and over.

But as more qubits and more complexity got added to the system, the time the simulator took to produce its prediction increased. That’s to be expected, just like a bigger pachinko machine. At first the times for actually executing the calculation and simulating it may have been comparable — a matter of seconds or minutes. But those numbers soon grew hour by hour as they worked their way up to 54 qubits.

When it got to the point where it took the simulator five hours to verify the quantum computer’s result, Google changed its tack. Because more qubits isn’t the only way quantum computing gets more complex (and besides, they couldn’t add any more to their current hardware). Instead, they started performing more rounds of operations with a given circuit, which adds all kinds of complexity to the simulation for a lot of reasons that I couldn’t possibly explain.

For the quantum computer, doing another round of calculations takes a fraction of a second, and even multiplied by thousands of times to get the required number of runs to produce usable probability numbers, it only ended up taking the machine several extra seconds.

For the simulator, verifying these results took a week — a week, on the most powerful computer in the world.

At that point the team had to stop doing the actual simulator testing, since it was so time-consuming and expensive. Yet even so, no one really claimed that they had achieved “quantum supremacy.” After all, it may have taken the biggest classical computer ever created thousands of times longer, but it was still getting done.

So they cranked the dial up another couple notches. 54 qubits, doing 25 cycles, took Google’s Sycamore system 200 seconds. Extrapolating from its earlier results, the team estimated that it would take Summit 10,000 years.

What happened is what the team called double exponential increase. It turns out that adding qubits and cycles to a quantum computer adds a few microseconds or seconds every time — a linear increase. But every qubit you add to a simulated system makes that simulation exponentially more costly to run, and it’s the same story with cycles.

Imagine if you had to do whatever number of push-ups I did, squared, then squared again. If I did 1, you would do 1. If I did 2, you’d do 16. So far no problem. But by the time I get to 10, I’d be waiting for weeks while you finish your 10,000 push-ups. It’s not exactly analogous to Sycamore and Summit, since adding qubits and cycles had different and varying exponential difficulty increases, but you get the idea. At some point you can have to call it. And Google called it when the most powerful computer in the world would still be working on something when in all likelihood this planet will be a smoking ruin.

It’s worth mentioning here that this result does in a way depend on the current state of supercomputers and simulation techniques, which could very well improve. In fact IBM published a paper just before Google’s announcement suggesting that theoretically it could reduce the time necessary for the task described significantly. But it seems unlikely that they’re going to improve by multiple orders of magnitude and threaten quantum supremacy again. After all, if you add a few more qubits or cycles, it gets multiple orders of magnitude harder again. Even so, advances on the classical front are both welcome and necessary for further quantum development.

‘Sputnik didn’t do much, either’

So the quantum computer beat the classical one soundly on the most contrived, lopsided task imaginable, like pitting an apple versus an orange in a “best citrus” competition. So what?

Well, as founder of Google’s Quantum AI lab Hartmut Neven pointed out, “Sputnik didn’t do much either. It just circled the Earth and beeped.” And yet we always talk about an industry having its “Sputnik moment” — because that was when something went from theory to reality, and began the long march from reality to banality.

That seemed to be the attitude of the others on the team I talked with at Google’s quantum computing ground zero near Santa Barbara. Quantum superiority is nice, they said, but it’s what they learned in the process that mattered, by confirming that what they were doing wasn’t pointless.

Basically it’s possible that a result like theirs could be achieved whether or not quantum computing really has a future. Pointing to one of the dozens of nearly incomprehensible graphs and diagrams I was treated to that day, hardware lead and longtime quantum theorist John Martinez explained one crucial result: The quantum computer wasn’t doing anything weird and unexpected.

This is very important when doing something completely new. It was entirely possible that in the process of connecting dozens of qubits and forcing them to dance to the tune of the control systems, flipping, entangling, disengaging, and so on — well, something might happen.

Maybe it would turn out that systems with more than 14 entangled qubits in the circuit produce a large amount of interference that breaks the operation. Maybe some unknown force would cause sequential qubit photons to affect one another. Maybe sequential gates of certain types would cause the qubit to decohere and break the circuit. It’s these unknown unknowns that have caused so much doubt over whether, as asked at the beginning, quantum computing really exists as anything more than a parlor trick.

Imagine if they discovered that in digital computers, if you linked too many transistors together, they all spontaneously lost their charge and went to 0. That would put a huge limitation on what a transistor-based digital computer was capable of doing. Until now, no one knew if such a limitation existed for quantum computers.

“There’s no new physics out there that will cause this to fail. That’s a big takeaway,” said Martinez. “We see the same errors whether we have a simple circuit or complex one, meaning the errors are not dependent on computational complexity or entanglement — which means the complex quantum computing going on doesn’t have fragility to it because you’re doing a complex computation.”

They operated a quantum computer at complexities higher than ever before, and nothing weird happens. And based on their observations and tests, they found that there’s no reason to believe they can’t take this same scheme up to, say, a thousand qubits and even greater complexity.

Hello world

That is the true accomplishment of the work the research team did. They found out, in the process of achieving the rather overhyped milestone of quantum superiority, that quantum computers are something that can continue to get better and to achieve more than simply an interesting experimental results.

This was by no means a given — like everything else in the world, quantum or classical, it’s all theoretical until you test it.

It means that sometime soonish, though no one can really say when, quantum computers will be something people will use to accomplish real tasks. From here on out, it’s a matter of getting better, not proving the possibility; of writing code, not theorizing whether code can be executed.

It’s going from Feynman’s proposal that a quantum computer will be needed to using a quantum computer for whatever you need it for. It’s the “hello world” moment for quantum computing.

Feynman, by the way, would probably not be surprised. He knew he was right.

Google’s paper describing their work was published in the journal Nature. You can read it here.

Hello and welcome back to Startups Weekly, a weekend newsletter that dives into the week’s startups and venture capital news. Before I jump into today’s topic, let’s catch up a bit. Last week, I wrote about All Raise’s expansion, Uber the TV show and the unicorn from down under.

Remember, you can send me tips, suggestions and feedback to kate.clark@techcrunch.com or on Twitter @KateClarkTweets. If you don’t subscribe to Startups Weekly yet, you can do that here.

---

The SoftBank saga

According to a new report from The Wall Street Journal, SoftBank plans to take a more conservative approach as it begins deploying capital from Vision Fund II. Why? The Japanese mega-fund’s track record is less than stellar. Not only has it lost billions on WeWork, but several of its other portfolio companies are suffering through layoffs, mismanagement and more.

Fair.com, a startup building a flexible car ownership business that is valued at $1.2 billion — backed by some $500 million in equity from SoftBank and others, plus billions more dollars in debt funding — said this week that it will be laying off 40% of its staff. On top of this, it’s removing its CFO, Tyler Painter, the brother of the CEO and co-founder (and car business veteran) Scott Painter. He’s being replaced in the interim by Kirk Shryoc. Read more from TechCrunch’s Ingrid Lunden.

WeWork, the co-working giant we’re all too familiar with at this point, has benefited from $18.5 billion from SoftBank, according to Marcelo Claure, a SoftBank executive who’s speech to WeWork employees was leaked to Recode this week. “We have guaranteed the future of WeWork, but more importantly is we’re putting the future back into our hands. There’s no more days needed to go fundraising …The size of the commitment that SoftBank has made to this company in the past and now is $18.5 billion. To put the things in context, that is bigger than the GDP of my country where I came from. That’s a country where there’s 11 million people.” Now nearly every single WeWork investor, particularly SoftBank, is entirely under water.

There’s more where that came from. Brandless, another … star of SoftBank’s portfolio, has struggled greatly with layoffs and a CEO shake-up, according to The Information. The dog-walking startup Wag raised $300 million from SoftBank, has also endured layoffs and management changes, and has failed to protect the safety of its pets, per this great report from CNN. And Compass, a real estate unicorn, has lost its CFO, CMO and CTO in what’s been labeled “Another SoftBank-Fueled Real Estate Exodus.”

---

Meet me in Berlin

The TechCrunch team is heading to Berlin again this year for our annual event, TechCrunch Disrupt Berlin, which brings together entrepreneurs and investors from across the globe. We announced the agenda this week, with leading founders including Away’s Jen Rubio and UiPath’s Daniel Dines on tap for great talks. Take a look at the full agenda.

I will be there to interview a bunch of venture capitalists, who will give tips on how to raise your first euros. Buy tickets to the event here.

---

VC deals

• Logistics management business Shipwell raises $35M
• Very Good Security gets $35M
• Grafana Labs raises $24M from Lightspeed
• Mobile banking startup Current secures $20M
• Modern Animal raises $13.5M to make going to the vet suck less
• Pet health tech company Gallant nabs $11M
• Bespoke Financial lands $7M to fund cannabis entrepreneurs
• YC grad Taskade nabs $5M to take on Notion
• Workplace learning platform HowNow gets $3M
• Vendr, already profitable, nabs $2M to replace your enterprise sales team

---

Equity

This week, Alex was remote and I was in studio to chat about a new angel fund, the WeWork saga and Lime’s losses. Listen to the latest episode here.

Equity drops every Friday at 6:00 am PT, so subscribe to us on iTunes, Overcast and all the casts.

---

Startup Spotlight

Bespoke Financial wants to provide cannabis businesses with the same kind of financial services that other businesses get, but that dispensaries and growers can’t yet access.

The regulations around cannabis operations are so stringent at the local level — and so nebulous at the federal level — that national banks won’t give businesses in the cannabis industry the same basic services (like short-term loans).

That’s why one former Goldman Sachs banker has partnered with two entrepreneurs from the traditional agriculture industry to create Bespoke Financial. And it’s why the company has raised $7 million in financing led by Casa Verde Capital — the investment firm launched by legendary cannabis aficionado, Calvin Broadus (AKA Snoop Dogg). Read more by TechCrunch’s Jon Shieber here.

Uber is selling foodie experiences such as cooking classes and multi-course fine dining in its on-demand food delivery app, Uber Eats, under a new Moments tab, per Forbes, which reports on a small-scale test currently running in San Francisco.

It says Uber Eats users in the city have received an email saying they can book Uber Moments for the next month, until November 17, with initial bookable experiences being a $75 class on making Chinese dumplings and a $55 five-course Nigerian dinner.

“We’re always thinking about new ways to enhance the Eats experience,” an Uber spokesperson told Forbes when asked about the pilot.

The test sounds similar in concept to the experiences which Airbnb has baked into its on-demand accommodation platform over the past three years — with tens of thousands of experiences now being offered in its case, running the gamut from (similar) foodie offerings, to pretty much anything you can think of wanting to do; guided hiking tours, glamping, animal petting, performing arts classes and so on.

It might seem odd for Uber, a ride-hailing giant, to try to blend its on-demand food delivery arm — with its raison d’être of quickly filling a lunch-hole in your stomach — with aspiration culinary experiences like lessons on preparing elaborate dinner.

But the company has designs on building what CEO Dara Khosrowshahi described last month as the “operating system for your everyday life”.

In June we also reported that Uber had begun testing folding Eats into its main app, ahead of publicly laying out its plan to roll multiple services into a single app to rule users’ daily decisions. (Work is another area of current focus for Uber: Earlier this month it launched a shift-finder app in Chicago, partnering with local staffing agencies but saying it would expand the offering to more areas “soon”.)

So, ultimately, Uber Moments looks intended to sit alongside a range of Uber-powered services, from ride-hailing to micromobility, employment and on-demand delivery.

As well as — most likely — new services Uber hasn’t launched yet but needs to given its ongoing quest to hail a profitable business model.

It’s pretty easy to envisage Uber Moments rubbing shoulders with other branded tabs in Uber’s ‘uber app’ — say Uber Stays, Uber Trips and Uber Cover, for example, if the company were to pivot towards travel; or Uber Clean, Uber Care and Uber Fix, if it decided to get in on on-demand home services.

Last month it launched an incubator to develop new services to plug into its planned ‘everything app’. Khosrowshahi added the app would be “a one-click gateway to everything that Uber can offer you” — though what else it can offer which people will want to buy remains to be seen.

The company’s reputation has taken a battering in recent years. And whether Uber has traveled far enough down the road of reforming its culture and detoxifying its brand for consumers to want to lean in and deepen their relationship, after earlier years of scandals plus ongoing question marks over issues like passenger safety, is not yet clear.

If Uber’s ride-hailing business still can’t weed out problematic drivers then consumers will have little reason to trust it to delivery a wider range of everyday services.

Sequoia Capital founder Don Valentine passed way at his home in Woodside, Ca., today at age 87 of natural causes.

Sequoia posted a tribute to Valentine shortly afterward, calling him “one of a generation of leaders who forged Silicon Valley.”

A native of New York, Valentine majored in chemistry at Fordham University before joining Raytheon in South California, then moving north to the Bay Area to work at Fairchild Semiconductor, where over the years, Valentine began investing his own small checks into technology companies that he was meeting. According to Sequoia Capital, he soon attracted the attention of an early mutual fund group, Capital Group, which staked Valentine, allowing him to form a $3 million venture fund in 1974. Among his first bets from that pool of capital: Atari and Apple. He later led the firm into numerous other high-flyers, including Cisco Systems.

Valentine continue to lead Sequoia until handing over the reins well before retirement age to Doug Leone and Michael Moritz, though he continued attending partner meetings for another 10 years. The partners have said they were happy for his continued advice and guidance — not that they always agreed with him.

In 2017, in keeping with the firm’s focus on succession and ensuring smooth transitions, partners Roelof Botha and Alfred Lin were made U.S. co-heads of the firm working under Leone, who oversees the firm’s global operations with Neil Shen, the founder and managing partner of Sequoia Capital China. (Moritz stepped away for health reasons in 2012, though he has continued to remain actively involved in the firm.)

Leone issued a statement this afternoon about Valentine’s passing, writing: “We are deeply saddened to share that Don Valentine passed away on October 25, 2019. Don’s life is woven into the fabric of Silicon Valley. He shaped Sequoia and left his imprint not just on those of us who had the privilege to work with him or the many philanthropic institutions that invested with Sequoia, but also on the founders and leaders of some of the most significant technology companies of the later part of the twentieth century. Our thoughts are with Don’s wife, Rachel, with his family, and with all those inspired by his pioneering vision and indelible impact.”

Valentine chose the name Sequoia because it “conveyed the longevity and strength of the tallest of redwoods,” according to the firm’s tribute today to Valentine. They note, too, the “humility of someone who refrained from putting his own name on our business.”

Valentine is survived by his wife; three children; and seven grandchildren, according to Sequoia.

Valentine joined TechCrunch at a Disrupt event back in 2013. He appeared along with another pioneer of the venture industry, Kleiner Perkins Caufield & Byers cofounder Tom Perkins . Perkins passed away in June 2016 at age 84.

Ben Horowitz, the co-founder of the venture firm Andreessen Horowitz, has a new book coming out this coming Monday titled “What You Do is Who You Are,” that takes a look at how to create “culture” at a company.

It’s a word that’s thrown around a lot but very hard grasp, let alone implement in a sustainable way. Horowitz learned firsthand as a CEO how elusive it can be when he took stock of his company, only to discover it was made up of “screamers who intimidated their people,” others who “neglected to give any feedback,” and at least one compulsive liar who excelled at sucking up to Horowitz and also making up stories from whole cloth.

Horowitz says creating culture was a missing part of his education, and in this new book — a follow-up to his best-selling “The Hard Thing About Hard Things” — he does his best to fill that gap for other CEOs, using his own experience, as well as lessons gleaned from historical figures Toussaint Louverture and Genghis Khan, along with Shaka Senghor, a contemporary who served time for murder and today is a criminal justice reform advocate.

It’s an instructive and novel combination, and we suggest picking up the book, especially if you love history. In the meantime, we sat down recently with Horowitz to talk about its timing and whether some of the biggest cultural blow-ups in the startup industry — Uber and WeWork — could have been avoided. These excerpts have been edited for length and clarity. Note that we’ll have more of the conversation — including Horowitz’s thoughts about dual-class shares —  for readers of Extra Crunch on Monday.

TC: You’ve just written a book about culture that’s coming out just as a lot of questions are being raised about culture because of WeWork. What happened there?

BH: [Cofounder Adam Neumann] had a certain kind of culture there. He had some holes — some great strengths and great holes. And sometimes that happens. When you’re really good at part of it, you can delude yourself into thinking that you’ve got everything you need, when you have some massive incompleteness.

Adam is so amazing. Like, the way they got all the money and everything. And the vision was so spectacular. And everybody there believed it, and they recruited some phenomenal talent. But when you’re that optimistic, it does help to have something in the culture that says [allows] people to bring you the bad news, like, if the accounting is all over the place or what have you.

TC: As with Uber’s Travis Kalanick, whose culture also came under fire, Neumann operated in very plain sight. He wasn’t hiding who he was or what he was spending. 

BH: Right, everybody knew how Travis was running the company. Everyone in Silicon Valley knew, let alone everyone on the board. The culture was published. You can look up Uber’s values [from that period].

Travis designed, I think, a really compelling culture, and believed in it, and published it. And the consequences of what he was missing were also super well-known. It’s only when board members think people are coming after them that [they take an interest in these things].

TC: What are the biggest lessons in these two cases?

BH: I obviously know more about Uber [as a Lyft investor who follows the space]. In Uber’s case, it’s a very subtle thing. Travis had a really good code. But he had a bug in it.

I think it was reported that, like, Travis encouraged bad behavior. I don’t think he did at all. I just think he didn’t make it clear that legal and ethical [considerations] were more important than competitiveness. As a result, when left to their own devices, in a distributed organization where there was a lot of distributed power, that combination had people doing things that were out of bounds.

And he was making everybody so much money. And the company was growing so fast that, for the board members, I suspect they were like, ‘As long as it’s making money, I’m not going to worry about what happens next.’

To me, the unfair part is, like, they shouldn’t get any credit at the end. Whatever you’re blaming Travis for [you should blame them, too] because they didn’t see it, either. I think that’s a charitable way of putting it.

The reason I wanted to go through [how to create business culture] in the book is so if you’re a new CEO, you can see, look, this thing looks like a small problem, but it’s going to become a big problem. Ethics are a bit like security issues. They’re not a problem at all until they’re a problem. Then they’re existential.

TC: Why was this issue on your mind?

BH: A few things. First, it was the thing that I had the most difficult time with as a CEO. People would say, ‘Ben, pay attention to culture, it really is the key.’ But when you were like, ‘Okay, great, how do i do that?’ it was like, ‘Um, maybe you should have a meeting about it.’ Nobody could convey: what it was, how you dealt with it, how you designed it. So I felt like I was missing a piece of my own education.

Also, when I look at the work I do now, it’s the most important thing. What I say to people at the firm is that nobody 10 or 20 or 30 years from now is going to remember what deals we’ve won or lost or what the returns were on this or that. You’re going to remember what it felt like to work here and to do business with us and what kind of imprint we put on the world. And that’s our culture. That’s our behavior. We can’t have any drift from that. And I think that’s true for every company.

On top of that, the companies in Silicon Valley have grown so fast and become so powerful that they’re getting a lot of criticism about their culture now, which, some of it is fair enough. But the proposed solutions are wacky  . . . so it kind of felt like somebody had to make a positive contribution and not just a critique about, like, okay, here’s what you ought to do.

TC: They’re also distributed, as you noted with Uber, but I don’t think you talk about remote workforces in this book. Do you have thoughts about establishing a culture where individuals are scattered here and there? 

BH: I didn’t talk about remote workforces and that one is interesting because it’s evolving because the tool sets are changing. It used to be nearly impossible for an engineering organization to be distributed and to be effective, because the information flow wasn’t good enough and the build systems weren’t good enough. And so for years, Microsoft would only buy companies that they could move to Redmond.

Lately, due to things like Slack and Tandem, people are getting better results with it. And I think a lot of the cultural techniques intersect with the tools quite a bit. But then you have to set the culture through electronic media more than you would walking around, catching somebody doing thing in a meeting.

We just did a thing on email the other day. We have this cultural value, which is: we don’t like to criticize entrepreneurs. I don’t care if we think your idea is stupid or whatever. You’re trying to create something from nothing. You’re trying to chase your dream. We support that, period. So if you get on Twitter and do what Bill Gurley does and say, ‘That company is a stupid piece of shit. It’ll never made a $1 and blah blah blah,’ like, you get fired for that. [Similarly] on our podcast, we have a news segment and I didn’t want us to do a story on ‘WeWork, the cautionary tale.’ That’s not us. And it’s a cultural statement. There are a million people who are going to write that story; we were like, let them write that story.

So you don’t have to be in person to set the tone, but you do have to be thoughtful about the way you do it, and who all hears it.

TC: Why weave in the cultural figures that you have? There are so many people you could have included, and you focused on these three individuals.

BH: It’s a weird origin story, but Prince years ago put out an album called 3121, and he opened this club in Vegas called the 3121, and he would perform there, like, every weekend. And the show would start at 10 and he would show up at midnight or 1 a.m., but during that time in between, he would show these old films with these really interesting dancers in these elaborate clothes. And you’d just be watching these old guys, and then Prince would start to splice in [his own movies, including] “Under the Cherry Moon” and “Purple Rain,” and you’d go, well those are the dance moves from those guys [in the older films] and that’s a quote from those guys And you realize: that was what he was trying to express. And I thought, you know, I finally really understand him. And I thought, you know, [these three] have really influenced my views on culture [for a variety of reasons] and it would be a good way to tell this story.

More on Monday . . .

The U.S. Department of Defense today announced that Microsoft has won its Joint Enterprise Defense Infrastructure (JEDI) cloud contract, worth up to $10B over a period of ten years. With this, Microsoft will provide infrastructure and platform services for both the Pentagon’s business and mission operations.

“The National Defense Strategy dictates that we must improve the speed and effectiveness with which we develop and deploy modernized technical capabilities to our women and men in uniform,” DOD Chief Information Officer Dana Deasy said in a related announcement. “The DOD Digital Modernization Strategy was created to support this imperative. This award is an important step in execution of the Digital Modernization Strategy.”

Microsoft beat out Amazon in the final round for this lucrative contract after the two cloud giants beat out other competitors like IBM and Oracle in an earlier round. Most pundits considered Amazon to be the frontrunner to win the deal.

The process to get to this point has been anything but uncomplicated, though, with various lawsuits, last-minute recusals and other controversies, with even the President getting involved at one point.

It’ll remain to be seen how Microsoft’s employees will react to this news. Last year, a number of Microsoft employees posted an open letter, asking the company not to bid for the contract. More recently, its employees also protested against GitHub’s relatively small $200,000 contract with the U.S. Immigration and Customs Enforcement agency. Against this backdrop, chances are we’ll see similar protests now that the company has won this deal with the Pentagon.

NASA is looking for liquid gold on the Moon — not oil, but plain-old water. If we’re going to have a permanent presence there, we’ll need it, so learning as much as we can about it is crucial. That’s why the agency is sending a rover called VIPER to the Moon’s south pole — its first long-term surface mission since 1972.

VIPER, or the Volatiles Investigating Polar Exploration Rover, will touch down in December 2022 if all goes according to plan. Its mission: directly observe and quantify the presence of water in the permanently shadowed polar regions.

These perennially dark areas of the Moon have been collecting water ice for millions of years, since there’s no sunlight to melt or vaporize it. NASA already confirmed the presence of water ice by crashing a probe into the general area, but that’s a bit crude, isn’t it? Better to send a robot in to take some precise measurements.

VIPER will be about the size of a golf cart, and will be equipped with what amounts to prospecting gear. Its Neutron Spectrometer System (mentioned yesterday by NASA Administrator Jim Bridenstine ahead of this announcement) will let the rover detect water beneath the surface.

When it’s over a water deposit, VIPER will deploy… The Regolith and Ice Drill for Exploring New Terrain, or TRIDENT. Definitely the best acronym I’ve encountered this week. TRIDENT is a meter-long drill that will bring up samples for analysis by the rover’s two other instruments, a pair of spectrometers that will evaluate the contents of the soil.

By doing this systematically over a large area, the team hopes to create a map of water deposits below the surface that can be analyzed for larger patterns — perhaps leading to a more systematic understanding of our favorite substance’s presence on the Moon.

The rover is currently in development, as you can see from the pictures at the top — the right image is its “mobility testbed,” which as you might guess lets the team test out how it will get around.

VIPER is a limited-time mission; operating at the poles means there’s no sunlight to harvest with solar panels, so the rover will carry all the power it needs to last about a hundred days there. That’s longer than the U.S. has spent on the Moon’s surface in a long time — although China has for the last few years been actively deploying rovers all over the place.

Interestingly, the rover is planned for deployment via a Commercial Lunar Payload Services contract, meaning one of these companies may be building the lander that takes it from orbit to the surface. Expect to hear more as we get closer to launch.

Tesla has launched the third iteration of its solar roof tile for residential home use, which it officially detailed in a blog post on Friday and in a call with media. Tesla CEO Elon Musk kicked off the call with some explanatory remarks on the V3 Solar Roof, and then took a number of questions. The company says it’ll begin installations in the coming weeks (Musk says some installations have already begun) and that it hopes to ramp production to as many as 1,000 new roofs per week.

Tesla’s solar roof tiles — which are designed to look just like normal roof tiles when installed on a house, while doubling as solar panels to generate power — are something of a work-in-progress. The company is still tinkering with the product three years after announcing the concept, having done trial installations with two different iterations so far. “Versions one and two we were still figuring things out,” said Elon Musk on an earnings call earlier this week, adding that he thinks “version three is finally ready for the big time.”

Tesla’s Solar Roof website now includes a pricing estimator, which lists $42,500 as the total price for the average 2,000 square-foot home, with 10kW solar panels. It also lists $33,950 as the price after an $8,550 federal tax incentive. You can also enter your address and get an updated estimate that takes into account local costs and incentives, and add on any Powerwalls (with 3 as the default for a 2,000 square-foot roof).

“The solarglass roof is is not going to make financial sense for somebody who has a relatively new roof, because this is itself a roof, that has integrated solar power generation,” Musk explained, but went on to note that Tesla has managed with this version three product to achieve a price point which is “less than what the average roof costs, plus the solar panels” that you would add on top of said roof.

“Figuring out how to install it effectively is very non-trivial. And we’re actually going to have […] kind of ‘installations,'” Musk said, which will pit two teams against each other to see who can roof one of two similar sized/designed roofs faster. Musk reiterated later that there’s “quite a bit of R&D just in the installation process itself.”

Musk also said that while it’s hiring and training specialized installers at first, the plan is to ultimately expand installations to any third-party contractors as well. On the call, he and the Tesla team discussed how they focused on getting the installation time down to where it’s faster than installing traditional shingles, plus solar panels on top of that. Musk added that his ultimate goal is to install the solar glass tiles even faster than comparative shingles. This is a significant change from V2 of the solar roof, Musk later said.

“We’re doing installations as fast as we possibly can, starting in the next few weeks,” Musk said about availability, adding that the goal is to “get to 1,000 roofs per week” sometime in “the next several months.”

A report from CNBC from September 2018 found that Tesla still hadn’t performed many actual installations of its solar roof tile, despite the two-year gap between announcement and the date of their investigation, and a January announcement about the initiation of solar roof tile production at Tesla’s Buffalo-based Gigafactory. During the company’s annual general shareholder meeting in June, Musk said that the third iteration of the tile was being worked on, and while he didn’t detail the actual number of installations, he did say that they were in progress in eight different states across the U.S. at that point.

Musk addressed some of the production delays to date, addressing the installation complexity of previous generations, but also citing the Tesla Model 3 production ramp, which he said “really stripped resources from solar for a year or a year and a half.” Now that Model 3 production is in a good place, Musk said that that has unblocked significantly some of the company’s ability to focus on this challenge.

The total addressable market that Musk sees for this product is somewhere on the order of 100 million houses worldwide, and Musk stressed that the company does indeed intend to make this available worldwide.

While at launch, there will be only one available look for the Solar Roof, the Tesla CEO also said that the company will roll out additional variants as quickly as it can, including tiles that resemble clay and other alternatives.

The tiles and roof installation carry a warranty of 25 years, which includes their protective weatherization (including 130 MPH wind resistance) and their power generation capability. On balance, the Solar Roof provides more energy generation than a similarly-sized roof retrofitted with traditional tiles, though individually, the tile’s power-gathering cells themselves are less energy efficient than a traditional solar cell. The Solar Roof is better performing, however, because it covers more surface area of a home.

While Facebook CEO Mark Zuckerberg seemed cheerful and even jokey when he took the stage today in front of journalists and media executives (at one point, he referred to the event as “by far the best thing” he’d done this week), he acknowledged that there are reasons for the news industry to be skeptical.

Facebook, after all, has been one of the main forces creating a difficult economic reality for the industry over the past decade. And there are plenty of people (including our own Josh Constine) who think it would be foolish for publishers to trust the company again.

For one thing, there’s the question of how Facebook’s algorithm prioritizes different types of content, and how changes to the algorithm can be enormously damaging to publishers.

“We can do a better job of working with partners to have more transparency and also lead time about what we see in the pipeline,” Zuckerberg said, adding, “I think stability is a big theme.” So Facebook might be trying something out as an “experiment,” but “if it kind of just causes a spike, it can be hard for your business to plan for that.”

At the same time, Zuckerberg argued that Facebook’s algorithms are “one of the least understood things about what we do.” Specifically, he noted that many people accuse the company of simply optimizing the feed to keep users on the service for as long as possible.

“That’s actually not true,” he said. “For many years now, I’ve prohibited any of our feed teams … from optimizing the systems to encourage the maximum amount of time to be spent. We actually optimize the system for facilitating as many meaningful interactions as possible.”

For example, he said that when Facebook changed the algorithm to prioritize friends and family content over other types of content (like news), it effectively eliminated 50 million hours of viral video viewing each day. After the company reported its subsequent earnings, Facebook had the biggest drop in market capitalization in U.S. history.

Zuckerberg was onstage in New York with News Corp CEO Robert Thomson to discuss the launch of Facebook News, a new tab within the larger Facebook product that’s focused entirely on news. Thomson began the conversation with a simple question: “What took you so long?”

The Facebook CEO took this in stride, responding that the question was “one of the nicest things he could have said — that actually means he thinks we did something good.”

Zuckerberg went on to suggest that the company has had a long interest in supporting journalism (“I just think that every internet platform has a responsibility to try to fund and form partnerships to help news”), but that its efforts were initially focused on the News Feed, where the “fundamental architecture” made it hard to find much room for news stories — particularly when most users are more interested in that content from friends and family.

So Facebook News could serve as a more natural home for this news (to be clear, the company says news content will continue to appear in the main feed as well). Zuckerberg also said that since past experiments have created such “thrash in the ecosystem,” Facebook wanted to make sure it got this right before launching it.

In particular, he said the company needed to show that tabs within Facebook, like Facebook Marketplace and Facebook Watch, could attract a meaningful audience. Zuckerberg acknowledged that the majority of Facebook users aren’t interested in these other tabs, but when you’ve got such an enormous user base, even a small percentage can be meaningful.

“I think we can probably get to maybe 20 or 30 million people [visiting Facebook News] over a few years,” he said. “That by itself would be very meaningful.”

Facebook is also paying some of the publishers who are participating in Facebook News. Zuckerberg described this as “the first time we’re forming long-term, stable relationships and partnerships with a lot of publishers.”

Several journalists asked for more details about how Facebook decided which publishers to pay, and how much to pay them. Zuckerberg said it’s based on a number of factors, like ensuring a wide range of content in Facebook News, including from publishers who hadn’t been publishing much on the site previously. The company also had to compensate publishers who are taking some of their content out from behind their paywalls.

“This is not an exact formula — maybe we’ll get to that over time — but it’s all within a band,” he said.

Zuckerberg was also asked about how Facebook will deal with accuracy and quality, particularly given the recent controversy over its unwillingness to fact check political ads.

He sidestepped the political ads question, arguing that it’s unrelated to the day’s topics, then said, “This is a different kind of thing.” In other words, he argued that the company has much more leeway here to determine what is and isn’t included — both by requiring any participating publishers to abide by Facebook’s publisher guidelines, and by hiring a team of journalists to curate the headlines that show up in the Top Stories section.

“People have a different expectation in a space dedicated to high-quality news than they do in a space where the goal is to make sure everyone can have a voice and can share their opinion,” he said.

As for whether Facebook News will include negative stories about Facebook, Zuckerberg seemed delighted to learn that Bloomberg (mostly) doesn’t cover Bloomberg.

“I didn’t know that was a thing a person could do,” he joked. More seriously, he said, “For better or worse, we’re a prominent part of a lot of the news cycles. I don’t think it would be reasonable to try to have a news tab that didn’t cover the stuff that Facebook is doing. In order to make this a trusted source over time, they have to be covered objectively.”

A fleet of electric, autonomous Hyundai Kona crossovers — equipped with a self-driving system from Chinese autonomous startup Pony .ai and Via’s ride-hailing platform, will start shuttling customers on public roads next week.

The robotaxi service called BotRide will operate on public roads in Irvine, California, beginning November 4. This isn’t a driverless service; there will be a human safety driver behind the wheel at all times. But it is one of the few ride-hailing pilots on California roads. Only four companies, AutoX, Pony.ai, Waymo and Zoox have permission to operate a ride-hailing service using autonomous vehicles in the state of the California.

Customers will be able to order rides through a smartphone app, which will direct passengers to nearby stops for pick up and drop off. Via’s expertise is on shared rides, and this platform aims for the same multiple rider goal. Via’s platform handles the on-demand ride-hailing features such as booking, passenger and vehicle assignment and vehicle identification (QR code). Via has two sides to its business. The company operates consumer-facing shuttles in Chicago, Washington, D.C. and New York. It also partners with cities and transportation authorities — and now automakers launching robotaxi services — giving clients access to their platform to deploy their own shuttles.

Hyundai said BotRide is “validating its user experience in preparation for a fully driverless future.” Hyundai didn’t explain when this driverless future might arrive. Whatever this driverless future ends up looking like, Hyundai sees this pilot as a critical marker along the way.

Hyundai said it is using BotRide to study consumer behavior in an autonomous ride-sharing environment, according to Christopher Chang, head of business development, strategy and technology division, Hyundai Motor Company .

“The BotRide pilot represents an important step in the deployment and eventual commercialization of a growing new mobility business,” said Daniel Han, manager, Advanced Product Strategy, Hyundai Motor America.

Hyundai might be the household name behind BotRide, but Pony.ai and Via are doing much of the heavy lifting. Pony.ai is a relative newcomer to the AV world, but it has already raised $300 million on a $1.7 billion valuation and locked in partnerships with Toyota and Hyundai.

The company, which has operations in China and California and about 500 employees globally, was founded in late 2016 with backing from Sequoia Capital China, IDG Capital and Legend Capital.

It’s also one of the few autonomous vehicle companies to have both a permit with the California Department of Motor Vehicles to test AVs on public roads and permission from the California Public Utilities Commission to use these vehicles in a ride-hailing service. Under rules established by the CPUC, Pony.ai cannot charge for rides.