The sale of Boston Dynamics shows Alphabet is waking up to reality

By Graham Templeton 

Looking at reports of the internal conversation that led to the surprise sale of Boston Dynamics last week, Google seems to be facing up to an increasingly evident fact: The Google model of old is not nearly as robust as previously believed. The company, from Nest to Calico to Search itself, seems like it’s about to get a lot more conservative in its spending.

From a purely economic perspective — that is, leaving aside any incentive for the pure advancement of mankind, which I’m sure exists — Google’s premise has always been that the advantages of owning a market will far outweigh the costs of buying it. You can come to own a market by literally purchasing all the important parts, but for Google the more viable option has been to simply make a new market from scratch. Through technological and social engineering, the strategy is to both create a product for the world and change the world so it better fits that product.

Acer Chromebook 15

As the company sets its sights on ever-more-ambitious ideas about what it could do with perfect insight into our lives, it’s having to cast a much wider net in terms of companion projects. If it wants to know how we live, it’ll need to be in our homes early, and build often — Nest. If it wants to know everything we do with our computers (nearly approaching every single keystroke) it’ll need an always-online platform — Chromebooks. If it wants these and other upcoming services to be possible, it’ll need a fundamentally better internet infrastructure — Fiber. It doesn’t want these businesses necessarily, but it wants the markets they create. Nest is treated like a necessary but burdensome annoyance, Chromebook manufacturing has been almost entirely offloaded to other companies, and I wouldn’t be surprised if Alphabet sold some or all of Fiber once it’s stimulated the market a bit further.

That sort of hardcore interventionism doesn’t come cheap. As some of these sorts of projects start to mature and others don’t, Google is getting a harsh lesson in just why mankind didn’t invent the Google model until well after the invention of the Internet. As revolutionary as that model has been, it’s also fundamentally rooted in the freedom allowed by Internet technology. When you want to revolutionize webmail, and read everyone’s messages, you simply invest a bunch in storage and release Gmail in near-perpetual beta; when you try to release a self-driving car according to such a model, things get a whole lot more complicated.

The simple fact is that when it comes to physical products, Google’s only successful moves and initiatives have not shot for the moon, but climbed. They always have a way of making money before achieving 100% of the ultimate goal. The least successful ventures are those where, due to either difficulty or bad decision-making, such half-measures cannot be made profitable.

An example where it’s due to bad decision-making is the company’s self-driving car division. Try telling the National Highway Traffic Safety Administration that the first self-driving cars should be sold before the moonshot goal of a completely safe self-driving car has been achieved. Other car manufacturers, like Tesla and BWM, have a chain of more achievable products moving quickly toward approval. These more modest technologies apply self-driving tech to some aspects of driving, like freeway driving or parallel parking.

If Google had taken such an approach from the beginning, their genuine foresight in this area could possibly have gotten their tech to market before any competitor had the chance to catch up. That could have allowed an implementation of the Android Gambit against car manufacturers: Run our software and compete, or bet on your own janky in-house alternative and get driven from the market. But now, hamstrung by their decision to lock their product launch to the goal of total safety in a self-driving vehicle, competitors can develop their own solutions at a leisurely pace.

The point is, through the income, brand loyalty, and bug-finding advantages of having real products in the real world, it’s possible the car companies will successfully use their early foothold to scoop Google on its own idea and beat it to market with a fully self-driving vehicle. If that happens, it won’t be due to lack of vision or ambition on Google’s part, but in a sense due to an overabundance of both.

The humanoid robot (“Replicant”) division faced similar problems, but they were far less avoidable. According to Bloomberg’s expose of the planned sale, it comes back to how the company is spending large fractions of the overall budget on ideas with no hope of commercialization for at least 10 years. It is quite simply difficult to release a real, physical product — what the software world lovingly calls a patch, real industry calls a “recall.”

Still, Boston Dynamics did have a couple of ideas with real interested investors. One of their most promising projects, the LS3 heavy-lifting version of BigDog, was recently shelved by the military due to apparently unfixable design flaws. But realistically, how many robot pack mules was the company ever really going to be able to sell? A few thousand?

If Google’s assessment of Boston Dynamics is to be believed, perhaps not even that many.

That’s why I wonder who (but the military) will see a profit potential in Boston Dynamics. If Google X found it didn’t have the funds and emotional wherewithal to see the Replicant project through, or to cow the company into meaningfully changing its approach, then who could possibly hope to succeed in their place? Amazon’s Jeff Bezos is extremely near-term focused in terms of large investments, so I’m skeptical of reports that Amazon would be looking to purchase Boston Dynamics. Amazon uses streamlined, utilitarian robots designed specifically to interact with their warehouses, so what use would they have for a pack mule or general purpose man-bot?

Fiber has been expanding fast.

Google isn’t necessarily setting its sights lower when it comes to robotics, though this move will certainly be read that way by many. It’s just that Google seems to be developing some doubt about the idea that the moonshot model is the best way to make it to the moon. The more realistic, and perhaps even the faster path, is to forecast a chain of profitable small-scale innovations that collectively lead you in the direction of your goal.

We’ll know just how far the culture has shifted based on the currently ongoing projects that have yet to truly show their hand. Calico comes to immediately mind, though the hush-hush health startup could still start with a more traditional, Google-y software technology for all we know. They could easily try to push some sort of Google Health Reborn as a stable and potentially quite successful software platform, perhaps getting Calico a source of steady income so it can self-fund its hardware and wetware ambitions, whatever those may specifically be.

Some will see this as a sign of Google’s “decline,” but as mentioned, this doesn’t need to be a lessening of ultimate ambitions, only proximate ones. For now, DARPA is still the only major research body in the world that gets to super-not-care about money. Google X would like to get there, and it may do it someday, but this week’s announcement shows that the moonshot lab is indeed mortal.

The question is whether this is the extent of the cull, or just the beginning.