Causal inference in human social systems is hard. Your "data" is limited and biased. Need depth of understanding. Being smart isn't enough.
-
-
Replying to @zeynep
If you want to deal with bias in digital platforms, you gotta bring in people who think a lot about stratification, inequality, culture...
1 reply 50 retweets 85 likes -
Replying to @zeynep
Engineers, you know how you are befuddled by very smart people who can't figure out email? Sometimes, that's you, with human social systems.
22 replies 300 retweets 472 likes -
Replying to @zeynep
(Yes, Smart people can be confused by computers. I know one of world's top physicists: can barely email. Domain expertise. Really).
4 replies 5 retweets 36 likes -
Replying to @zeynep
The culture clash is real; methodological gap between engineers & trad social scientists is real but silos won't let you solve the problems.
4 replies 12 retweets 36 likes -
Replying to @zeynep
There is a lot of social science expertise in "unintended" consequences of "easy-seeming" fixes from available data. A lot of it is subtle.
1 reply 7 retweets 16 likes -
Replying to @zeynep
You can't just hire a few social scientists to Silicon Valley (often mostly economists anyway). Would need real teams; different thinking.
3 replies 5 retweets 29 likes -
Replying to @zeynep
Also, AirBnB's race problem was apparent from day one among social scientists BECAUSE IT WAS TERRIBLY EASY TO PREDICT. Why did they not see?
9 replies 30 retweets 105 likes -
Replying to @zeynep
We see this often: a terribly easy to see complication to business model is ignored for years, then a team of engineers tasked after outcry.
1 reply 12 retweets 34 likes -
Replying to @zeynep
Truth is, there *are* a lot of things companies CAN do easily to help solve this, and then there are a lot of hard problems in this space.
2 replies 3 retweets 7 likes
You'd need—at a minimum—a huge dose of humility; willingness to hire & *empower* social scientists of stratification, culture & inequality.
-
-
Replying to @zeynep
Meanwhile, academia & funders should empower people studying these systems from the outside. Too important to leave to "we'll try". /end
2 replies 5 retweets 9 likes -
Replying to @zeynep
zeynep tufekci Retweeted SonicTheHegemon
Also this. Hiring economists to design your ad auction system is fine; these are not such issues.https://twitter.com/DontMakeMoi/status/773906172822839296 …
zeynep tufekci added,
0 replies 2 retweets 13 likes
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.