A word about the META team build by @ruchowdh @quicola and others --

This was probably one of the last teams at a big tech/social company that had the ear of product and policy, that wasn't dismantled at the whim of a whiny white tech boy. 1/ They put out important self-critical research, including work that showed the conservative amplification bias of the platform, and took a user-driven concern around cropping bias and made a contest out of it. 2/

I appreciate a lot about this piece by @emilymbender, and I'm glad she was gracious enough to spend time debunking what seemed like fluff from the NYT. Read it in full, but I wanted to point out a few thoughts:

https://twitter.com/emilymbender/status/1515847090542317569

Emily does a great job shifting the framing of this piece and fundamentally challenging the technodeterminist terrain that it's on. The author boxes her critique into one of "we need to teach machines ethics" rather than the broader critique of organizational power and reach.

I followed Lilly's lead on this and stepped down from participating in this conference. I fully echo her sentiment and well-thought out thread, with a few notes of my own. 1/

https://twitter.com/gleemie/status/1455774049523568641

First, _funding matters_ in academic venues. Even though it may be a small act, withholding labor can be akin to withholding legitimacy to those funders. In the "AI ethics" space, as with much of AI, money is pouring out right and left to "solve" ethics. 2/

It turns out the Ethical AI team was the last to know about a massive reorganization, which was prompted by our advocacy. This was not communicated with us at all, despite promises that it would be.

bloomberg.com/news/articles/… Nothing about what we asked for has been addressed here.

* Samy Bengio is no longer in our reporting chain.
* An apology has not been offered to Timnit by Jeff Dean or Megan Kacholia.
* Our input for such an organization was solicited, but then decided upon behind closed doors.

A thing to pay attention to in @sundarpichai's non-apology is this bit:

"One of the best aspects of Google’s engineering culture is our sincere desire to understand where things go wrong and how we can improve." Google has the notion of a "blameless" postmortem, the idea that if a system breaks, then folks sit down and write up what went wrong, without anyone to blame.

This was brought up by a higher up in a prior meeting as well.

watching The Christmas Gay movie with @Stakselrod and this is now a Jane fan account these two aren't going to go very fast in this ice skating race without doing crossovers

The time has come.

That's right, I'm going to sit down and give The Social Dilemma (2020) the Cats (2019) livetweeting treatment. Note, if you want a well-thought out critique of this movie, Tristan Harris, and CHT more broadly, I refer you to these sources by @libshipwreck librarianshipwreck.wordpress.com/2020/09/17/fla… and @gleemie and @ruchowdh wired.com/story/tech-nee…

But this? This is going to be purely affective and annoyed.

I'm annoyed that it took an angry mind parasite to take Kes to dump Neelix. I like angry mind parasite Kes much more #VOY Oh, are they gonna use this body swap as an excuse for a fake gay plot line #VOY

The episode has finally arrived...

TUVIX

#VOY Honestly, where can I get this outfit. This is the by far the best StarFleet uniform

Every Q episode is the worst episode #VOY Remember when Sisko just punched Q in the mouth

I should surely binge Supermarket Sweep, correct? Supermarket Sweet taught me, at a very young age, the exorbitant cost of diapers

Excited to chat about the beginnings of our genealogies of data project at the Participatory ML workshop (participatoryml.github.io) at @icmlconf. In it, we approach benchmark datasets in machine learning and think through the labor and work practices embedded in them. 1/4

https://twitter.com/andrewthesmart/status/1283806960186322944

We start from @DMulliganUCB's conceptualization of contestability, asking where designers can intervene in contingent decision in developing benchmark datasets, and pose four research questions for the analysis of benchmark ML datasets. 2/4

"The whitewashing of roller skating's online revival" by @jessjoho is very good (h/t @BrownJefe). A few thoughts: 1/4

#derbytwitter She lays out how the segregation of Black and white skating communities has been transmuted into the online -- Tik-Tok's selection algorithms further deepening the separation between the two and Planet Roller Skate's censoring of Black skaters 2/4

Public comment beginning for this Oakland Police Commission meeting, about 2.5 hours after the start of the meeting. 33 comments submitted beforehand and 140+ people now in queue

These few past days have given me such a strong #Jan25 energy. Municipal cops, Egyptian state security -- same gassed up, unbridled state-sanctioned violence. If there are counter protesting camel attacks this week I'm going to go buy a lotto ticket

I thought this would happen: a bimodal distribution of productivity during COVID. People who are unencumbered or more productive in isolation can produce more, whereas those with kids, who are working through mental health, etc would not be. This sort of supports that hypothesis.

https://twitter.com/KathyDMKE/status/1246169379760455682

I'm already hearing about this kind of bimodality at Google: single (mostly male) tech workers loving the absence of co-present meetings and their ability to chug away at code, while...