Every year Google IO gets less interesting and sees Google heading in a direction that ultimately threatens their long term future. The core of every part of the presentation outside of pure search, was assistant. It drove the Android section and the devices section.

Assistant was a great step forward for Google, and is at least as good as Amazon’s Echo as a voice assistant. There are 2 fundamental problems with it that are starting to become non-negotiable with me though.

Hostile to privacy

You cannot use many of the privacy tools if you are using Google Home. If I tell Google not to keep track of my web search history then Home will stop working.


Most of what I use Google Home for is setting an alarm, basic web searches, timers, weather and to tell me the time. I also use it to interact with home automation. None of that requires any knowledge of what I’ve been searching for.

I’m still willing to believe this is apathy to privacy on Google’s part, rather than an attempt at locking users into storing history. That willingness is being eroded by other privacy hostile actions Google has taken.

Another announcement just before IO was the Google will allow user to control how long location history is stored. I use location history for only one purpose - sharing my location with my wife in real time. This doesn’t require any history to be stored; the user only ever gets to see the last data point anyway.

You can disable location history, but then that disables realtime sharing as well, which makes no sense other than Google trying to lock you into sharing data. The new controls only allow you to automatically remove data 3 months ago. For the purposes of ad targeting that’s about as much history as you need, and data any longer than that is much less economic to monetise. So privacy is fine as long as it’s not impacting revenue.

Google monolith

Google is turning into a monolith that is hostile to 3rd parties.

For nearly all of the demos shown at IO, they only work properly if you are using Gmail, Google Calendar, Google Photos, Google Maps and an Android phone. You have to be fully invested in the ecosystem to get the advantages. I don’t necessarily want all my photos in Google Photos, where they’re doing facial recognition, or my location recorded in Google Maps, which is often handed over to police forces.

The other problem is the general degradation of these services over time. Gmail’s latest update was a disaster at launch. It was incredibly slow and missing key functionality in Firefox. It’s still nowhere near as good as it used to be, but if I want assistant to be properly useful I can’t migrate away to another email service.

I made a conscious decision to move away from many Google services over the past year because individually there are better services from other providers. The ultimate result is that I step away from one non-revenue generating part of Google like Calendar and it lowers the quality of the revenue generating portion - Search. The Google monolith has made Google much less sticky for me.

One encouraging sign

The big encouraging sign for me was the move to on device ML. For all of Google’s smart online services there are others that perform just as well without sending data to the cloud. The entity detection in Synology’s Moments application performs at least as well as Google Photos for me.

It makes sense to do classification on device, if only to save on server costs. My question is whether Google can turn that into something that works well across 3rd parties, and can monetise it in a non-privacy invasive way then there is potentially a bright way forward. My expectation is that isn’t what will happen.