in 2019, Microsoft spent $17bn on research and development (R&D). Google spent $26bn. Amazon spent $36bn. Facebook spent $14bn. Apple spent $16bn. In total that’s $109bn.
How big are those numbers? In 2019 the UK government spent £12bn ($15bn) on R&D, about the same as Facebook, the smallest spender. The French government spent €18bn ($20bn), about the same as Microsoft. The German government spent €30bn ($34bn), about the same as Amazon.
The five US tech giants together spent more on R&D last year than all of the current EU’s governments combined.
Huge spending on R&D produces advanced technology. For most people in most of the world, video calls with dozens of people are nearly free. They keep working without interruption even as use goes well beyond predicted demand. Detailed maps centred on our exact location with directions and guides to every city in the world are free. Setting up a shop is nearly instant, you can buy any of millions of books in a minute. You can have them delivered in a few days or listen to them instantly. There are almost no barriers to the broadcasting and sharing of ideas and culture. Banking and payments in the real world are fast, cashless, and secured by our fingerprints or faces. We can take photos on a device that fits in our pocket, at a quality we couldn’t have imagined a decade ago, then share those photos around the world from almost anywhere on the planet, within seconds, while listening to any song ever recorded. When we search for that photo of a swan years in the future we can find it by asking our phone to “find that picture of a swan I took”.
Arthur C. Clarke famously wrote that “any sufficiently advanced technology is indistinguishable from magic”. R&D at this scale creates magic.
There is always dark magic too. I could list the many drawbacks, costs, and risks of our technological advances. We could discuss those who are left behind and excluded.
We discuss negatives a lot in the UK and I both join in and host those debates. I help to include those who are excluded. But I feel like we talk about the negatives of technology too much given how little we are involved in creating it. My favourite example is in smart cities. The UK hosts conference after conference and forum after forum on smart cities and ethics in smart cities. Meanwhile in our real cities, most people still pay for the bus with coins. We talk so much, we seem to do so little.
Today I’m mostly writing about Google in the UK and their COVID-19 Community Mobility Reports. I have no inside information about Google, so the following is all public fact or my guesses. As with all my writing during the Coronavirus pandemic, if you can’t disagree politely, please stop reading.
You can read Google UK’s annual accounts online. They spent £462m on R&D in the UK in 2019 and they paid £44m on £226m of profit in the UK. At 19% that’s about the same as a pretty normal and well-liked business such as Timpson (they paid 24% of their profit in tax in 2018 and 13% in 2017). Perhaps I’ve misread Google’s accounts, and I imagine that they have some exotic tax arrangements internationally via their parent company Alphabet that reduces their tax bill, but Google look to me like a company that pays their taxes, at least in the UK. And given the very high salaries that they pay to their staff, and the difficulty of avoiding income tax, their activities are likely to generate a very significant contribution to the running of the UK state above the tax that they pay on their profits.
But that’s not the main reason why I think it’s important to read Google UK’s accounts. In them they list the risks to the future profitability on which they will pay their staff and their taxes. Three of the six primary risks are about privacy concerns of users and governments and the potential for changes in the law that would limit the company’s ability to operate. That’s important.
On the timescale of technology and innovation, Covid-19 happened suddenly. There has been almost no time to develop new technologies. Existing technologies can be adapted to new purposes, but in a panicked mess of pain and emotion it is hard to know what any person, any company, or any technology can do. There is always a risk that in trying to help, more damage than good will be done.
And for Google there is an extra risk, that whatever it does to help will trigger its biggest fear, a perception that it is “creepy”, over-powerful, and requires greater regulation, perhaps to the extent of being forced to split up.
A quick search for “Google”, “mobility”, and “creepy” on twitter, Facebook, or (I appreciate the irony) Google News will turn up what Google fear. Comment after comment and article after article about how creepy it is that they’re tracking us all the time to produce these reports.
The safest thing for Google to do would have been to copy Apple. Apple collect anonymised location data the same as Google, but they have not released anything similar. I suspect that this is for two reasons. First, that they have not developed the technologies needed to do the analysis that Google has; they’ve skimped on R&D. Second, they want to protect their strong brand around privacy even if that comes at the cost of not helping governments understand how well their lockdowns are working.
Apple’s calculation seems to be working well for them. I can see no online criticism of their failure to provide location information to governments and I suspect that among those in positions of power criticising Google the use of Apple products is disproportionately high.
Given the obvious potential costs to Google’s reputation, why did they produce and publish their global mobility reports?
Mostly, I guess that their employees felt that it was the right thing to do. But also, the UK government asked them to.
A few days after the UK entered lockdown I was asked if I knew how the UK government could understand how well it was working. In London, tube trains were still busy at peak times. There was significant concern that more drastic interventions would be required. CityMapper’s excellent data visualisation tool was showing that UK cities were lagging well behind the rest of Europe in reducing public transport use. But it was glaringly obvious on the UK government’s daily briefing slides that it had no data.
I explained in these calls what I have explained for the past decade, that outside of the regulated public transport systems of London we have almost no idea how many people use public transport on any given day in the UK. While every city in France, The Netherlands, Germany, and pretty much every other country in the world knew in real time how public transport use was changing, the UK outside of London had almost no data. We still have almost no data.
Why we have no data is contested and political. I have had the argument many times in the past and I will have it with part of my frustration converted into anger once the worst of this crisis has passed. Our lack of data stems from some mixture of UK underfunding of R&D, UK centralisation of government competence and responsibility around digital government and data, and our economic model. We can figure it out later.
Fortunately for us, Google could provide the data we needed. Thanks to Google’s huge spending on R&D in recent years, nearly double in 2019 the total spending on R&D by the public sector in the UK, spending that was underwritten by the potential for future profit, the UK would have far less data on how its lockdown is working and a significantly lower chance of timing its emergence from lockdown correctly.
And yet by sharing this data Google has risked further regulation and threat to the profits that fund its R&D today.
Managing that risk is why I think Google publishes its mobility reports as PDFs.
The PDF is a format hated by many advocates of open data and the better use of data in society. By locking up the data that Google calculated in graphs within PDFs, Google are deliberately making it harder for others to re-use. It would be easy for them to release the raw data behind the graphs, but they chose not to.
So I asked a poll on twitter about why people think they chose to do that. I don’t think my followers picked the best answer.
I think that PDFs best meet what people call “user needs”. The people preparing briefing notes for Cabinet are typically not data scientists. Even if they were, they should probably not be doing any original data analysis. These people have heavily secured and locked down laptops. They can definitely open a PDF, but they cannot definitely open much else. A briefing pack for Cabinet is some slides and often a big stack of paper. Documents, not data. So Google published documents.
But why not publish data as well? I think it’s also because of user needs, but defined more deeply. Our society is better because of the huge amounts that Google and others spend on developing the technologies that we love to use and that we love to complain about. From failing to invest in an emergency warning system, to failing to create systems that could monitor public transport use, the UK’s governments have proven themselves less good at creating the digital infrastructure we need to handle pandemics best. The public sector has failed to create the structures of power, or invest the sums required. Google and others have made up the gaps in this and many ways by raising money to invest in R&D on the promise of future profits. It is a user need of the mobility reports that Google can continue to make the profits that underwrite its ability to exist to produce them.
But the risk to those profits of the company’s work being labelled “creepy” is too high for Google. Remember, three out of six risks that Google UK imagines to its own future existence are associated with it. Anything similar to Uber’s God View, anything even hinting at something similar, would threaten the ability of Google to invest in the R&D that has proven so valuable.
Google will have known that other organisations would extract the data from their PDFs. They will have decided in advance to turn a blind eye to re-uses whose legality is unclear.
Innovative, speculative, even creepy analysis of Google’s data is in the interests of Google. But only if they are associated with it as little as possible.