I was reading Thomas Baekdal’s latest newsletter…
…and in it I came across this passage in a note about how risky automated systems make modern online life.
” the idea that I could potentially just lose access and have everything deleted simply because an automated system thought I had posted too many emojis one day … that’s scary as hell.”
Basically, a Youtuber asked his viewers to do something that sounded like fun and that accidentally triggered an automated process that perma-banned many of his viewers.
A corollary to that is that this represents an immense risk to Google.
Unlike Twitter or Facebook, Google has a constellation of businesses that all make money and are strategically important to it as a tech organisation. Youtube represents a huge systemic risk to every other part of Google:
- All forms of moderation and spam control cost Google more than they do its competitors: when Google bans or limits a user, it’s also removing a customer from its other businesses.
- All experiments with abuse control have a huge secondary risk because all of Google’s other properties use Youtube heavily for promotion. Imagine if the Youtuber above had been a Google Cloud tech evangelist talking about new platform features. A mistake like this, with the wrong concentration of viewers, could have taken a big chunk out of Google Cloud’s business.
A lot of people want Google broken up. What Google is missing is that breaking it up would probably be good for its various businesses. Google Cloud, for example, would benefit from not being saddled with the privacy violations of the ad business or the crowd control risks that come with Youtube.