The point of a model is to be validly predictive of something. Fitting your exponential is validly predictive of local behaviour more often than not. Often, insanely so.^[1] You can directly use the numerical model to make precise and relevant predictions.

Your exponential doesn't tell you when the trend stops, but it's not trying to, for one because it's incapable of modelling that. That's ok, because that's not its job.

Fitting a sigmoid doesn't do this. The majority of times, the only additional thing the result of a sigmoid fit tells you is how an arbitrarily chosen dampening model fits to the arbitrary noise in your data. There's nothing you can do with that, because it's not predictive of anything of value.

This doesn't mean you shouldn't care about limiting behaviour, or dampening factors. It just means this particular tool, fitting a numerical model to numerical data, isn't the right tool for reasoning about it.

1. ^{^}

   “I answered that the Gods Of Straight Lines are more powerful than the Gods Of The Copybook Headings, so if you try to use common sense on this problem you will fail.” — Is Science Slowing Down?, Slate Star Codex, https://slatestarcodex.com/2018/11/26/is-science-slowing-down-2/

It's only an argument against fitting curves to noise. Rather than explain, it turns out there's already a post that puts this better than I could hope to. I endorse it fully.

https://www.lesswrong.com/posts/6tErqpd2tDcpiBrX9/why-sigmoids-are-so-hard-to-predict

Low effort, low nuance hot take:

Sigmoids are fake. Real curves aren't symmetrical, and research progress curves in particular tend to look like exponentials that run into brick walls. Adoption curves can be relatively balanced but I still don't buy it.

And given you already can't fit a sigmoid that isn't already dying and probably shouldn't try, fitting the thing that actually happens in reality is doomed. Fit your exponential and intuit the rest. You don't actually have quantified evidence for when it ends, so use the evidence you do have instead.

But I (and almost everyone else who didn't call it as obvious in advance), should pay attention to the correct prediction, and ignore the assertion that it was obvious. 

I think this is wrong. The scenarios where this outcome was easily predicted given the right heuristics and the scenarios where this was surprising to every side of the debate are quite different. Knowing who had predictors that worked in this scenario is useful evidence, especially when the debate was about which frames for thinking about things and selecting heuristics were useful.

Or, to put this in simpler but somewhat imprecise terms: This was not obvious to you because you were thinking about things the wrong way. You didn't know which way to think about things at the time because you lacked information about which predicted things better. You now have evidence about which ways work better, and can copy heuristics from people who were less surprised.

My first reaction is that this is bad decision theory.

It makes sense to actualize on strikes when the party it's against would not otherwise be aware of or willing to act on the preferences of people whose product they're utilizing. It can also make sense if you believe the other party is vulnerable to coercion and you want to extort them. If you do want fair trade and credibly believe the other party is knowing and willing, the meta strategy is to simply threaten your quorum, and never actually have to strike.

We don't seem to be in the case where an early strike makes sense. The major reaction to this post is not of an unheard or silenced opposition, but various flavours of support. In order for the moderators to cede to your demand, they have to explicitly overrule a greater weight of other people's preferences on the basis that those people will be less mean about it. But we're on LessWrong, people here are not broadly open to coercion.

Additively, we also don't seem to be in a world where your preferences have been marginalized beyond the degree that they're the minority preference. The moderators clearly spent a huge personal cost and took a huge time delay precisely because preferences of your kind are being weighed heavily.

Given the moderators are presumably not going to act on this, and would seemingly be wrong to do so, this comment reads as someone hurting themselves and others to make moderation incentives worse. Harming people to encourage bad outcomes is not something LessWrong should endorse.

I respect the integrity and strength of person needed to take a personal cost to defend someone against a harm, or a moral position. I think it's honourable to credibly threaten to act in self-sacrificial ways. Yet, there are right and wrong ways to do this. This one strikes me as wrong.

I'll admit I don't have much need for optimizing physical bookmarks myself these days, but I expect you're underestimating how reusable and replaceable an adhesive tag can be. Not that I think it matters; at best it'd be an essentially similar item to what you already use.

These seem neat but not obviously meaningfully better for the referenced tasks than a sticky index tab, if running multiples. The darts do look suave though, and I'm sure they last better.

I also recently had need to homebrew some permanent book labels without them, and found that folded over labels from a label maker made a pretty nice and reasonably quick index.

Yes, your understanding matches what I was trying to convey. The feedback is appreciated also.

It's just Bayes, but I'll give it a shot.

You're having a conversation with someone. They believe certain things are more probable than other things. They mention a reference class: if you look at this grouping of claims, most of them are wrong. Then you consider the set of hypotheses: under each of them, how plausible is it given the noted tendency for this grouping of claims to be wrong? Some of them pass easily, eg. the hypothesis that this is just another such claim. Some of them less easily; they are either a modal part of this group and uncommon on base rate, or else nonmodal or not part of the group at all. You continue, with maybe a different reference class, or an observation about the scenario.

Hopefully this illustrates the point. Reference classes are just evidence about the world. There's no special operation needed for them.

Firstly, it's just not more reasonable. When you ask yourself "Is a machine learning run going to lead to human extinction?" you should not first say "How trustworthy are people who have historically claimed the world is ending?"

But you should absolutely ask "does it look like I'm making the same mistakes they did, and how would I notice if it were so?" Sometimes one is indeed in a cult with your methods of reason subverted, or having a psychotic break, or captured by a content filter that hides the counterevidence, or many of the more mundane and pervasive failures in kind.