Asterisk

> One thing is clear: Screening works. If you’re of the appropriate age, please get screened. If your tubes are acting funny, please get screened without delay. The best method and the level of benefit are debatable, but we know it helps.

That's unfortunate, because it's a poor summary of the actual paper the article is discussing. If you bother to read the rest of the article, you'll see the actual summary of the trial, and it's pretty darned easy to interpret for yourself:

https://asteriskmag.com/media/pages/issues/04/you-re-invited-to-a-colonoscopy/426b760313-1696013385/dynomight_temp01.png

to wit:

> The 18% reduction in colorectal cancer incidence was statistically significant, while the 10% reduction in colorectal cancer mortality and 1% reduction in overall mortality were not.

(Those reductions are relative risks, which are on very small baseline numbers.)

Then the article goes on for many pages about how American gastroenterologists do not agree...which isn't terribly shocking, considering the source of the argument. But a number of rebuttals are made, some of which are reasonable (i.e. colonoscopies are one of the few screening tests to actually do something to prevent disease, in the form of clipping polyps), and others of which are quite simply lies ("other studies have estimated larger benefits for colonoscopies" -- yeah, but those studies were bad, and this one was good, which is why we're talking about it!)

Note that American gastroenterologists do a lot…

My bad, it was an article the previous issue, the inaugural one:

https://asteriskmag.com/issues/01/making-sense-of-moral-change

> C: It doesn’t require a great deal of moral or ethical insight to see that treating a human being as a thing, not a person, is wrong. Just about every sort of spiritual, ethical or moral system has described slavery in one form or another as “against nature.”

> But historically, people also thought of slavery as a product of civilization and therefore found ways to justify it. They created ways to explain, for example, why specific groups of people should be enslaved, or why a particular system of slavery is justified, even a form of progress over a previous system.

> I’ve always thought that the best analogy is how we regard eating meat. [...] We also know that eating animals is unethical. We’re doing something cruel and unnecessary. But we do it because that’s just what we do. It wouldn’t be surprising to me if 30, 50 or 120 years from now, people look back on this time period and ask, “What was wrong with these people? They must have been like moral infants. They must not have realized this is a cruel way to treat animals.” But we know it perfectly well. It’s just what we do.

> A: I don’t eat pork or chicken for animal welfare reasons. And the line of thinking that convinced me to actually do this was that if I was alive in the 18th century, I’d want to be the kind of person who’s boycotting sugar.

That line of thinking convinced me too.

> I mostly agree with you, but would ask: Why are you here, reading this thread? This isn't, like, a thread about something interesting that is being tragically drowned out by all this gossip. It's just an entirely bad thread that we should (and probably do) all feel bad about getting sucked into.

I was just reading the comments after reading the article to see if anything new came up, and was pretty appalled at the quality of commentary here. I'm not participating in the thread more because it's not worth it.

> But the tiny sliver of disagreement I have with "this is a bad thing to discuss here and we should all feel bad" is that some people who frequent this site are sometimes some of the people involved in making decisions that might lead to threads like this. And it might be nice for those people to have read some comments here that push back on the narrative that it's actually fine to do stuff like this, especially if it's legal (but maybe also if it isn't, sometimes?).

I've been involved in big decisions in other Big Tech companies. I'm proud of having fought to preserve Tor access to our offerings because I believe in Tor despite the spam and attacks it brings. I don't know about other folks in these positions, but if I were to read discussion like this, I'd roll my eyes and close the thread. If a random incoherent drunk ranter told me something was wrong with my ideas, I'd dismiss them without much hesitation.

> The way I see this particular discussion is:…

> your network will not have a jump start on everyone else

I think this is not the most useful frame. You don't necessarily need to outcompete people for limited spots (I think a lot of people's definition of success or greatness is very narrow, and mostly informed by what others are vying for)

I think the bar for doing something great is shockingly low, most people aren't even trying. That's the impression I get every time I read a patio11 article (most recently the super long but worth it "story of vaccinate CA" [1]. This is not patio but same vibe, "Lies, Damned Lies, and Manometer Readings" [2]). My takeaway from these is that there are big problems in society that nobody is working on.

This is kind of a good thing for ambitious folks: everywhere you look there are problems everywhere, the bar for making something better than what is there is low. I think the network & support is critical but it isn't always given. I think great things happen when we start to move towards what we are capable of fixing, what is within our control, finding others who have that attitude and supporting each other.

I think for me now, the point isn't to outcompete everyone/make it to the top 1% or whatnot, it's to make it as far as is possible with what I have. Those two approaches are identical if you have the support to reach the 1%. But if not, why worry about what is outside of my control? I have plenty within my control in the menatime.

[1]…

There are a few things working in favor of community notes there

1. Viral tweets have a longer-than-average time window between the time they start to go viral and the time the median viewer sees them, so a community note can get there before the median viewer. 2. Users who interacted with a tweet before it got a community note will get a notification when the community note is added. 3. Community note writers can leave a note on a piece of media. If a tweet with a video gets a community note, and that note is about the video rather than about the tweet, that note will show on all other tweets that show that video.

Source: this excellent interview with the Community Notes team (https://asteriskmag.com/issues/08/the-making-of-community-notes), in the section that starts with "Asterisk: Another thing I wanted to talk about is speed". Really that whole interview is great, highlights how deliberate and thoughtful the Community Notes team was regarding everything about the feature. Which is, I think, why community notes have succeeded where a lot of previous fact-checking attempts have failed.

I think you could see this coming as soon as they implemented chain-of-thought in their previous models.

Chain-of-thought is well understood to be the way to squeeze out performance out of these models. Slow down, go step by step, use more tokens to get a little bit better output. This is so useful, that often in the comparision graphs you see between different models, the best one (the one I want to sell) is using chain of thought while others are not. Not always, but companies have been caught using this technique to hype up their models.

We’ve already seen that CoT (usually!) improves a model’s performance. So does prompting the model with examples of correct question-answer pairs (called in-context examples). But in reports, some models are evaluated with CoT, while others aren’t. The number of in-context examples is often different, and the prompts are almost always different.

https://asteriskmag.com/issues/07/can-you-trust-an-ai-press-release

As soon as OpenAI was desparate for performance upgrade, they implemented this as an actual feature.