google

Two stories have me thinking about Google, today.

Timothy B. Lee wrote about Waymo, the self-driving car company owned by Google, which has been operating a fully driverless taxi service in Arizona for months now.

(Technically Waymo is part-owned by Alphabet, the holding company that also owns Google, but I refuse to use the name Alphabet except when I'm forced to because it 1. is trading under the stock ticker GOOG 2. has the same CEO as Google 3. is dominated financially by Google's finances 4. is based out of the same office as Google and 5. no-one knows what the fuck Alphabet is and the entire thing is almost deliberately designed to confuse and obfuscate)

Lee's piece is a good overview of the current state of the art in self-driving cars. Waymo has cracked it – for a certain definition of "it".

The most striking thing about these videos is how boring they are. In nearly five hours of driving, I didn't see Waymo's vehicles make a single significant mistake. That contrasts with the "full self-driving" software Tesla released in beta back in October. I watched three hours of videos of customers testing out Tesla's technology. Drivers intervened more than a dozen times—including two cases where a crash seemed imminent.

Waymo manages to offer an actually-safe self-driving system by building something that only works in a very specific place. Arizona is perfect for trialling these cars: wide streets, gridded neighbourhoods, lights controlling most complex intersections, and bright clear days year round. On top of that, Waymo's system is closely integrated with the maps the company has of the area, which are closely linked to Google's own maps but not identical to them. The company has to map new territory whenever it expands.

Even so, asks Lee, the obvious question is why hasn't Waymo expanded? Even within Arizona, there are obvious gaps: Scottsdale, for instance, could be covered fairly quickly, it seems. And yet Waymo's service is resolutely local.

Lee proposes a few possibilities. One "is that the back-end costs of operating a Waymo car is higher than the revenue. Perhaps because they have a lot of well-paid people overseeing every ride."

Another option "is that they're just risk-averse. They want to maintain their near-perfect safety record and they're not yet 100 percent confident its cars won't make a serious mistake."

I was reminded of Lee's question by a second story about Google, from a very different arm of the company. From today, users of Google's Stadia videogame streaming service will be able to livestream the games they're playing directly to Google's YouTube video-sharing site. 

Stadia was released in November 2019. YouTube integration was a headline feature, discussed heavily before launch, and flagged as one of the ways in which Stadia would be able to offer more than conventional consoles, rather than just replicating their features over streaming.

So why on earth did it take over a year to bring to Stadia a feature which was advertised before launch, which is already on offer on other consoles, and which only requires an integration with a different company that Google also owns?

Like with Lee's questions about Waymo, I don't have any direct answers. My guess is that Google correctly realised that Stadia's problems are way deeper than simple lack of integration with YouTube, and focused its efforts on fixing those issues instead. Now, it's not done a great job doing so – the service still has baffling inconsistencies, a poor selection of games, and a fundamentally unanswered question about who it's for – but it's got better, at least, over the past year.

But I do think there's an overarching answer to both my and Lee's questions that is important to note. It's rarely brought up because it's considered common knowledge inside tech, and never directly relevant to any particular news item:

Google is just a really badly-run company.

This shouldn't be surprising. "Corporate culture" can be overstated, but it is a real thing: companies replicate their attitudes internally amongst new joiners, and even with concerted effort, change is hard. And Google's corporate culture was set in its first decade, a period when the company's experience was "if build an incredibly technically proficient service and monetise it in the obvious way, it will print money, steal customers from the competition, and turn its founders into billionaires without any significant management effort".

I exaggerate, of course. But the early years of Google.com are not exactly tales of management derring-do. Eric Schmidt joined in 2001, three years after the company was founded; before then, it was basically just nerds building a nerdy website, with the actual business an afterthought. And even after Schmidt joined, the Google way of building products continued largely unchanged, and to this day, Google's advertising division brings in $37.1bn of its $46.2bn growth, overwhelmingly driven by the search and display ad businesses that the company developed in that first decade.

And so Google never really had to learn basic management skills. How do you take an underperforming service and bring it around to profitability? When is the right time to kill a project that's not working? What should a minimum viable product be, and how far in advance should a roadmap be set? These are hard questions – hard enough for a startup, where some of the answers are, at least, set by the fact that you need to make enough money to keep the lights on. For a company like Google, where mistakes can persist for decades without anyone really noticing their effect on the bottom line, it's almost impossible to come to an answer without concerted effort.

Effort that Google just hasn't put in. The company's culture remains, to this day, focused on the same things it was back in 1998: building incredible technology, and sort of just expecting everything else to fall into place around it.

And so the answer to why Waymo hasn't expanded, and why Stadia took a year to get a basic feature promised before launch, is at heart the same response, I think: Google just doesn't really care. These aren't interesting engineering challenges, or passion projects for staff members involved. And so they're forgotten down the back of the sofa, ready to join the Google graveyard when whichever employee is pushing them forward decides it's time for an early retirement.

But those robot taxis are cool though