Posts by randomwalker@mastodon.social
(DIR) Post #APL6oNsUDsUPEAKl3A by randomwalker@mastodon.social
2022-11-06T17:00:56Z
0 likes, 3 repeats
Mastodon deliberately has no quote-toot feature because it leads to toxicity https://mastodon.social/@Gargron/99662106175542726 đź’ŻBut can't people achieve the same (toxic) effect by linking to toots instead of quoting? I think not. The couple extra steps involved in posting a link are enough to get you to stop and think about what you're doing. A more subtle reason is that pile-ons work by making the quoted tweet the butt of a joke, and jokes don't work if you have to click a link to see the punchline.
(DIR) Post #APL6oPk3ICvn0aP3BI by randomwalker@mastodon.social
2022-11-06T18:01:59Z
0 likes, 0 repeats
As many people have said, quote tweeting does have benefits, like providing context to your followers to explain why they might want to read the quoted thread.I wonder if there's a way to have the best of both worlds: allow quote toots, but **require approval by the original poster**. QTs sit in the original poster's notifications and aren't visible to anyone else until approved. It's a bit like a comments section where only approved comments are posted. Just thinking out loud!
(DIR) Post #APLS8l4fmoTi01afC4 by randomwalker@mastodon.social
2022-11-04T21:23:11Z
0 likes, 1 repeats
Algorithms aren't the enemy. Chronological feeds don't scale and the signal-to-noise ratio will plummet if this ever gets popular. The real problems with today's algorithmic feeds are non-transparency, lack of choice, and optimizing for engagement instead of healthy discourse.Open-source is a perfect opportunity to fix all this. Have there been any efforts to create a Mastodon instance with a (community governed) ranking algorithm? Is that technically feasible? Or is the idea simply anathema?
(DIR) Post #APLS8mfDsPJCvfhMno by randomwalker@mastodon.social
2022-11-05T11:58:43Z
0 likes, 0 repeats
Update: it turns out that lots of people have similar views and Simon willison is exploring building something along these lines.https://fedi.simonwillison.net/@simon/109289663684761988
(DIR) Post #APqFnqEnh8oqa8yQ64 by randomwalker@mastodon.social
2022-11-21T19:48:40Z
0 likes, 1 repeats
There's a claim going around that Mastodon is "centralizing around a few instances" because 1% of instances account for 84% of users. https://mathstodon.xyz/@manlius/109383027753990134That's misleading. 1% of instances is ~360. That's a lot more than I expected! Impressively decentralized.Measuring centralization as a % doesn't make sense. Hypothetically, if 100K new people each started their own instance, centralization as measured by this stat would greatly *increase*, even though that's not what actually happened.
(DIR) Post #APqFnvt2gwyo6Oq812 by randomwalker@mastodon.social
2022-11-21T19:57:51Z
0 likes, 0 repeats
The goal of decentralization shouldn't be a uniform distribution of instance sizes. That's unnecessary and unrealistic.Decentralization prevents any one entity from exerting too much power over the whole ecosystem, allows cultural differences between communities, and enables semi-independent technical experimentation by instances so that successful innovations can percolate. These are the things that made the web successful. Mastodon has all of these, and isn't in danger of losing them.
(DIR) Post #AQoOIFK1SBxABV1L4i by randomwalker@mastodon.social
2022-12-20T22:15:53Z
0 likes, 1 repeats
Mastodon has roughly 1% as many active users as Twitter. But it feels like much more than 1%. Why is that?Because we're not a random 1%. People in some communities are much more likely to find Mastodon appealing. For example, at least 30% of the people I follow on Twitter are on here, based on their Twitter bios.This "homophily" makes it easier for communities to reach critical mass. The downside is that it makes Mastodon more of an echo chamber (on top of the other reasons that make it so).
(DIR) Post #ARJQUpuNTZwrIdL68m by randomwalker@mastodon.social
2023-01-04T19:57:03Z
1 likes, 6 repeats
The number of active Mastodon users — those who've logged in in the last 30 days — has fallen to 1.8 million from a high of 2.6 million. This reflects the fact that while there's a wave of new users after each Musk tantrum, the majority aren't sticking around. Even the 1.8M figure is inflated by new accounts, so I expect a further drop. Still, the number of active users is over 4 times what it used to be before the Twitter takeover.Source: https://api.joinmastodon.org/statistics (and Wayback machine).
(DIR) Post #AS4zdT3AzL7AoNfX6W by randomwalker@mastodon.social
2023-01-27T16:19:23Z
0 likes, 1 repeats
I recently came across a fascinating theory that traces the crisis of American democracy to the fact that everyday people used to participate in local governance, but they don't anymore (I forget why... increasing urbanization? The nationalization of politics?) The distance from governance leads to distrust in government (and institutions in general).Unfortunately I can't remember the source. If anyone can point me to it, I'd be eternally grateful.
(DIR) Post #ASuWPNSDKh4m5iyMpE by randomwalker@mastodon.social
2023-02-21T15:31:18Z
0 likes, 1 repeats
Governments, banks, employers, and many other institutions make life-altering decisions about us by using machine learning to predict our future behavior. In a new paper, we challenge the legitimacy of this type of decision making, which we call “predictive optimization”. It is sold as accurate, fair, and efficient. But we show that it fails *on its own terms* and suffers from seven recurring flaws.By @ang3linawang, @sayashk, Solon Barocas, and me.https://predictive-optimization.cs.princeton.edu/
(DIR) Post #AU8On6a28Ju2Y28aYK by randomwalker@mastodon.social
2023-03-29T21:22:31Z
0 likes, 7 repeats
The AI moratorium letter only fuels AI hype. It repeatedly presents speculative, futuristic risks, ignoring the version of the problems that are already harming people. It distracts from the real issues and makes it harder to address them. The letter has a containment mindset analogous to nuclear risk, but that’s a poor fit for AI. It plays right into the hands of the companies it seeks to regulate. By @sayashk and me. https://aisnakeoil.substack.com/p/a-misleading-open-letter-about-sci
(DIR) Post #AULZnqET3ScbUiQRPM by randomwalker@mastodon.social
2023-04-05T13:23:03Z
0 likes, 1 repeats
I recently published an essay with @knightcolumbia explaining the concepts behind social media recommendation algorithms: https://knightcolumbia.org/events/optimizing-for-what-algorithmic-amplification-and-societyI'm grateful the response has been so positive. Several people said they're planning to assign it in their courses. 🙏 So I drafted up a few discussion questions: https://knightcolumbia.org/blog/understanding-social-media-recommendation-algorithms-a-discussion-guideIf you're interested in the topic, sign up for our symposium Apr 28/29 (NYC & online): https://knightcolumbia.org/events/optimizing-for-what-algorithmic-amplification-and-society[Filling up quickly — over 400 registrations already]
(DIR) Post #AV2nzCbyS5M5hjW9sO by randomwalker@mastodon.social
2023-04-26T12:42:58Z
0 likes, 2 repeats
People have been posting glaring examples of ChatGPT’s gender bias, like arguing that attorneys can't be pregnant. So @sayashk and I tested ChatGPT on WinoBias, a standard gender bias benchmark. Both GPT-3.5 and GPT-4 are about 3 times as likely to answer incorrectly if the correct answer defies gender stereotypes — despite the benchmark dataset likely being included in the training data. https://aisnakeoil.substack.com/p/quantifying-chatgpts-gender-bias
(DIR) Post #AV3geL53GRo5MtBP04 by randomwalker@mastodon.social
2023-04-26T13:15:45Z
0 likes, 1 repeats
OpenAI mitigates ChatGPT’s biases using fine tuning and reinforcement learning. These methods affect only the model’s output, not its implicit biases (the stereotyped correlations that it's learned). Since implicit biases can manifest in countless ways, OpenAI is left playing whack-a-mole, reacting to examples posted on social media.