Shreeharsh Kelkar (@scritic) header
Shreeharsh Kelkar (@scritic) profile picture

Lecturer at @UCBerkeley in ISF. Research on AI, algorithms, organizations, work, labor, and expertise. Writing at https://nitter.cf/t.co/HoZfMecSz6.

Berkeley, CA
Joined May 2008
Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)Photo by Shreeharsh Kelkar (@scritic)
Shreeharsh Kelkar retweeted
The environmental impacts of data centers is a complicated topic, and there are reasonable people on all sides of it, but this NYTs article has the dumbest f-ing frame you could possibly imagine. You should always ask yourself what the comparator is. In this case, why are American data centers being compared either to very poor nations or to very small ones? (Example: Switzerland is only about 40% the size of Virginia. Per the article: the data centers in Virginia put out 7 million tons of carbon dioxide while Switzerland puts out 32 million. But we don't get that comparison. And, in fact, really . . . wtf are we even comparing here?) And why are we not getting comparisons to other American activities with technologies like driving cars (~1 billion tons), flying in airplanes (~200 million), or producing metals (~80 million), to name a few? It's because the headline is motivated, and it's designed to stir people up. This kind of "reporting" really does a disservice to our media landscape, our collective understandings, and our democratic processes. We can and should do better. One thing I totally agree with in the article: Far too much industrial activity in the United States is secret, and we could use a lot more sunshine, openness, and public investigation. This applies not only to the environmental impacts of data centers but also to things like what social media companies actually *know* about the social effects of their platforms. We cannot reason well without knowing more, and we need legislators that will bust company books open.
1
5
1
24
1,176
New post on what using AI for peer review or replications might mean. I argue that it depends on whether peer review takes place in adversarial or cooperative contexts. (Partially inspired by the fight between @tylercowen and @gabriel_zucman on this site.) See comments for link.
1
54
Shreeharsh Kelkar retweeted
This is the tip of the iceberg in online education. Both tuition-paying students and institutions benefit from nobody paying attention. Students want the degree. Faculty want the cash. The people who sign the diplomas aren't held accountable. calmatters.org/education/202…
16
41
12
234
43,434
I like Canvas, find it quite useful, and no one forces me to use it. Most of the points listed in this post aren't about Canvas or specific problems teachers face reconciling Canvas features with their learning goals or assessments, but rather a set of political preferences.
In which I come out swinging against Canvas. Canvas is why students prefer AI to their professors, and I don't blame them one bit. hollisrobbinsanecdotal.subst…
3
2
361
Which is fine! I encourage people with these political preferences to start a company and create a product. I'd be happy to be on a focus group, talk about my requirements and learning goals, and be a beta-tester. Making a product is hard! In the meantime, I'll stick to Canvas.
1
2
43
New post defending (slightly) Dario Amodei and his cost-benefit analysis: computingandsociety.substack…
Readers added context they thought people might want to know
The excerpt is from Dario Amodei's 2010 GiveWell guest post as a Princeton physics grad student; he donated the $10k to VillageReach (child vaccines) and described the critical analysis as emotionally taxing. blog.givewell.org/2010/06/03/my-… nytimes.com/2026/11/05/tec…
1
84
In the post, I use the NHS and GiveWell as two examples of (quantitative) CBA done right and the p-doom stuff as CBA done all wrong.
1
23
*Spoilers*: Agree with this take. Holmes and Evans should have stuck to their case for taking a percent of the profits so that we, the audience, can feel guilty from being entertained by the reality TV shenanigans in this movie. My @letterboxd review: boxd.it/h22HV3
Replying to @tracewoodgrains
“You Can See Everything: In which Nathan Fielder does A24 an enormous favor by guilting and playing mind games with the stars of his movie until they give up their rights to millions of dollars for no good reason.” Seriously, that was the entire second half. Holmes’s reputation was already shot, and the movie was never, ever going to rehabilitate it. The only things her partner will likely ever be known for are being with Elizabeth Holmes and now showing his cock, literal and figurative, on TV. The one thing they were actually going to get out of this movie was cash—and a lot of it, given its enormous success. And if there were even a shred of plausible deniability, they could have gotten away with it. Billy does the dirty work, Holmes goes wide-eyed, sure they get an embarrassing documentary but they get money. But Holmes’s entire cover story relied on being scrupulously honest and not caring about money. As soon as the money comes up, Nathan obsesses about it. It’s the only topic worth mentioning, really, after she goes to jail—beyond his addition of a third season to The Rehearsal, of course. The whole movie has been building up a bit of Holmes going Nathan Fielder on Nathan Fielder, putting him at least ostensibly in a position where he never knows how much is serious. He doesn’t want to be beaten. If he lets her maintain her ambiguity throughout, he’s beaten. So he spends the whole back half orchestrating the prison conversation, where he forces her to be visibly aware that she’s visibly profiting from the film, with a chance to back out, while her entire story relies on her being a mission-driven person who doesn’t care about the money. Checkmate. Nathan won, Holmes lost. But only because she blinked and let Billy give up the money. Hope A24 sends him a nice gift basket.
2
78
Shreeharsh Kelkar retweeted
🙏🙏
Great episode from @WSITYpod and a model for how contentious issues like the tradeoffs of non-pharmaceutical interventions should be conducted. Wish they had gotten to vaccines. This is going into my course readings. Watch it here: youtube.com/watch?v=ZCWQ-qJT…
1
1
1
240
Great episode from @WSITYpod and a model for how contentious issues like the tradeoffs of non-pharmaceutical interventions should be conducted. Wish they had gotten to vaccines. This is going into my course readings. Watch it here: youtube.com/watch?v=ZCWQ-qJT…
We're recording our first live show at @WashU on October 27th 4 CDT. We'll be joined by @BhadeliaMD , @davidzweig, Katherine Wells, and Jeff Hutt to unpack what we should do when the next pandemic hits. Register to view the conversation live here: washu.qualtrics.com/jfe/form…
1
1
476
Thanks to @RBMD1982 for first introducing me to this podcast!
1
2
62
What can the open-weight AI movement learn from the history of the open-source? I cover open-source from its informal beginnings at the dawn of UNIX to Stallman's "free software" and LINUX to its christening as "open source" and to "cloud computing" which changed everything.
1
3
79
The short answer is that ends matter and advocates of open weight AI are interested in it for different reasons.
1
24
Link: computingandsociety.substack… Of possible interest to: @natolambert @binarybits @STS_News and of course, thanks to @beenwrekt whose original blogpost got me started.
1
34
Oct 15 would have been Michel Foucault’s 100th birthday. I thought I would re-up my post on him. While the post is critical of him to some extent, I hope it also conveys my belief that he gave the best answer to the question of “what is modernity?” computingandsociety.substack…
37
I finally listened to the first season of “Sold A Story” and loved everything but the framing. computingandsociety.substack…. It is NOT a story about science but about the long-running conflict in American education about standardization and teacher autonomy. Cc: @bjfr @PhilOnEdTech
1
1
140