I talk sometimes about an algorithm I built once. It's potential for harm is so high that I don't even talk publicly about what it does.
-
-
Replying to @EmilyGorcenski
This is the reality we have to face *every day* doing our mundane work. I _never_ expected this algorithm to do what it does.
1 reply 10 retweets 19 likes -
Replying to @EmilyGorcenski
When we're pulling together data on our users, we need to ask ourselves "what could this be used for? What harm can come?"
1 reply 38 retweets 55 likes -
Replying to @EmilyGorcenski
In reliability engineering, there is a process called FMEA -- Failure Mode Effect Analysis. It's a harm evaluation process.
1 reply 11 retweets 26 likes -
Replying to @EmilyGorcenski
You think of a failure that can happen, and score from 1-10 the likelihood of failure, the severity of harm given failure, and detectability
1 reply 6 retweets 18 likes -
Replying to @EmilyGorcenski
Data Scientists probably need to start doing this. "What is the harm that happens from classifying this object incorrectly?"
1 reply 35 retweets 66 likes -
Replying to @EmilyGorcenski
It's well-known among data scientists that if you work hard enough, you can find a pattern in the data that matches whatever you want it to.
1 reply 12 retweets 47 likes -
Replying to @EmilyGorcenski
And we now have an administration that will use its reach to classify people as they see fit. What is the harm in what we build?
2 replies 8 retweets 26 likes -
Replying to @EmilyGorcenski
I read yesterday about librarians destroying patron records after the Patriot Act was passed. Are we willing to destroy our algorithms?
7 replies 89 retweets 157 likes -
Replying to @EmilyGorcenski
Would you be able to direct me to where you read that? It sounds very relevant to something I'm researching!
1 reply 0 retweets 1 like
here is one source:http://mobile.nytimes.com/2003/04/07/us/some-librarians-use-shredder-to-show-opposition-to-new-fbi-powers.html …
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.