Discover Weekly quickly became a habit for people, Newett reports, giving them something to look forward to on Monday mornings. Then, a few months after the mid-2015 launch, he says, the team had its "first production incident—it didn't update." Some users "went into blind rage or existential crisis."
The exact formula for how search engines like Bing and Google rank results is governed by secret algorithms mere mortals aren't allowed to know. But two factors dominate: Pages linked by other pages are ranked higher, as are pages with keywords matching the search terms. That's what puts Wikipedia pages at the top of most Google searches—they cite, and are cited by, numerous other lesser sources (such as blogs). But sex traffickers don't want to be found via Web search. To throw off the index, they advertise through one-off ads, unlinked to others. They hide deep in chat rooms or uncrawlable social-media posts. They avoid search-engine optimization. Instead of keywords, they use photos and code words. At this moment, there are likely hundreds of thousands of active ads for sex for sale on the Internet. Detectives using regular search engines have an extremely difficult time finding these or making cases against criminals who don't play by Google's rules.
Chris White was given the chance to change the rules.
As machine learning increasingly affects domains protected by anti-discrimination law, there is much interest in the problem of algorithmically measuring and ensuring fairness in machine learning. Across academia and industry, experts are finally embracing this important research direction that has long been marred by sensationalist clickbait overshadowing scientific efforts.
This sequence of posts is a sober take on the subtleties and difficulties in engaging productively with the issue of fairness in machine learning. Prudence is necessary, since a poor regulatory proposal could easily do more harm than doing nothing at all.