On Privacy and Service Offering Quality

John Gruber reacts to Tim Cook’s privacy-oriented speech from Tuesday, and Thomas Ricker’s response:

Apple needs to provide best-of-breed services and privacy, not second-best-but-more-private services. Many people will and do choose convenience and reliability over privacy. Apple’s superior position on privacy needs to be the icing on the cake, not their primary selling point.

The argument being that Apple, or any tech company, should not have to choose between offering great services and protecting user privacy.

I’ve alluded to this before, but I’m not sure Apple can provide a Google-equivalent quality of cloud service while keeping things private. Take, for example, one of the most impressive features of Google Photos: its ability to catalogue and tag the contents of your photos automatically, without requiring any intervention on the user’s part. They search “cat”, and they get all the pictures they’ve taken of cats. Magic.

To do this, they need a lot of information on what cats look like from every conceivable angle, in a wide range of lighting conditions. The good news is that Google has billions upon billions of users’ image searches of cats. They’ve crawled the web for the past fifteen years and found a whole bunch of pictures of cats based on the alt and title tags of the images, their captions, and the content around them on the originating page.

While they haven’t released details on exactly how their image recognition tech works, it’s reasonable to guess that Google has been tracking a user’s image search for “cat” to the images they click on. The results that get more clicks — in aggregate — are probably going to be the most accurate representation of the search term. Google can therefore take the knowledge garnered from billions of searches a day on myriad search terms and create a way for their Photos product to detect the subject matter of user images.

We have, in simple terms, been providing Google with keywords and an indication of their accuracy for years, which they can use across all of their services.

Apple struggles with this kind of machine learning prowess because they don’t operate services in the same way that Google does. They don’t analyze user-provided data in aggregate; they often even keep a single user’s information siloed across multiple services. Google, by contrast, governs most of their services under a single privacy policy that allows them to blend all data collected under this policy. This allows them to offer a breezy first-run experience and services that require multiple kinds of data cross-referenced and associated with one another.

Google Now is the perfect example of such a service. It mines your email, calendar, location, todos, Google’s collected search data, and other information to try to guess what you will need to see at a given moment. If you are at the airport at 14:39 and you have a plane ticket in your Gmail for 15:48, it can identify a flight number and show you whether your flight is on time or not. Word on the street is that Apple is going to compete with Now, but without the kind of deep integration and cross-comparison that Google’s consolidated privacy policy allows, I’m curious to see how Apple could implement something in a privacy-friendly way.

Maps is another example. Google’s data is, generally, better than Apple’s¹ because it’s been around for a while, it’s free to use, and they’re able to check their data against pages they’ve crawled. They also sent out a bunch of cars to verify data and take Street View pictures. Apple partnered with a bunch of other companies to build their data index, but it’s hard to compete against Google crawling the web every single day for all kinds of information that could be used to bolster map results.

I’ve gotten this far without talking about the money aspect. It is in Google’s best interests to be accurate and create an index that is as robust as possible which is cross-referenced on a per-user basis, and in aggregate. The more they understand, the more relevant ads they can sell.

Apple doesn’t have such an incentive, and that’s a value I cherish. But this is not an excuse for creating worse products. I want the very best, of course, and Google is the high watermark for almost every web service. But I fear that it is unrealistic for Apple to improve their machine learning capabilities and web service quality without relaxing their firm stance on privacy. I’m not sure what I worry about more: that Apple’s service quality will suffer, or that I will have to give up a little bit of my privacy to prevent it from doing so.