Public
- Public
- Network
- Groups
- Featured
- Popular
- People

Conversation

Notices

Embed this notice
Erin 💽✨ (erincandescent@akko.erincandescent.net)'s status on Friday, 01-Mar-2024 03:10:30 JST Erin 💽✨
- Hrefna (DHC)
It alarms me how many people have been banging the "Machine Learning is just Money Laundering for bias" drum for years, yet claim that algorithmically curated block lists are a way of avoiding it

RE: https://hachyderm.io/users/hrefna/statuses/111988051617673256
In conversation about 9 months ago from akko.erincandescent.net permalink
Attachments
1. No result found on File_thumbnail lookup.
  
  Hrefna (DHC) (@hrefna@hachyderm.io)
  
  from Hrefna (DHC)
  
  To reiterate: my problem with consensus #blocklist approaches is that they present a "false consensus" of the world. Whatever biases are in the source lists? Are present in the output. Often amplified. Whatever priorities are in the source lists? Are present in the output. Often amplified. The idea is to remove biases, but that only works if biases in your source lists are _normally distributed_ Which they aren't. Thus: amplified But this is obscured and it is presented as "more fair." 1/

Feeds