upvote
Yeah, I don't understand the HN title. The "downfall" seems to have began in 2018-2020 sometime, what AI was launched and popularized at that point that would have killed SO? LLMs were basically useless until GPT3 which appeared in middle-2020 sometime, after the downfall seemingly already had begun.
reply
I'd call it significant that the number of questions halved within one year following the release of ChatGPT, the biggest relative or absolute rate of decrease in the timeseries.
reply
Right around the time Google rolled out new search results removed many information based sites.
reply
"How AI precipitated SO's downfall" would be a correct title then
reply
Add it to the list:

- the downfall of junior devs

- bad hiring market

- layoffs in practically every sector

theres a ton of things where AI took credit for a trend that had already started before it started being even halfway capable.

reply
I think if you won't even admit that AI greatly accelerated these trends, you're in some kind of denial. There's no reason to believe that we would see a rapid coordinated decline in all of these things at the same time without AI, and strong reason to believe that we would see it with AI. So we have a model that makes testable predictions, and data strongly consistent with those testable predictions, in the form of an acceleration of existing downward trends. What more do you want?
reply
>I think if you won't even admit that AI greatly accelerated these trends, you're in some kind of denial

I think if you actually look at the data for these trends rather than asking AI what it thinks you might experience some cognitive dissonance.

>There's no reason to believe that we would see a rapid coordinated decline in all of these things at the same time without AI

It's called hiked interest rates. The economy is not doing so great for several reasons but the main one is wars.

reply
Hiked interest rates cause a decline in Stackoverflow activity?

My answer doesn't change. Against the background of other phenomena already causing various trends, we see acceleration of those trends consistent with a testable hypothesis about the effects of AI adoption in industry. In most fields of science that's good evidence in favor of the model.

And no I didn't ask AI about it, this is my own opinion and my own perspective.

reply
People love simplistic narratives, i usually don't mind but this is just ridiculous. AI hate is gently overtaking AI hype as the most stupid thing around
reply
AI hype is still a million miles ahead and a million times dumber, especially thanks to online astroturfing.
reply
By volume sure, a part of the anti AI crowd is pretty extreme though, death threads, bomb threads, etc.
reply
It's not really a bell curve. There was obviously a downwards trend from 2016 onwards, but 2023 definitely precipitated the fall to zero. Without AI they might have lasted at least a couple more years, or the activity might have stabilized to a new floor greater than zero.
reply
https://postimg.cc/n9nZGLmb

Goodness of Fit 0.911, Kurtosis -0.849, Skewness: 0.073

It's very much a bell curve

reply
Just because it's approximated by a bell curve doesn't make it a bell curve. There are quite obvious separate phenomena shaping the curve at different times.
reply
> just because it's approximated by a bell curve doesn't make it a bell curve

I'm going to assume this is bait...

reply
Bell curves are probability distributions. This is a time series, so it can’t be a bell curve. It just has the same shape.
reply
Are you discarding the utility of Gaussian functions in analysis simply because the independent variable is time? A Gaussian curve can be used as a descriptive model without claiming that the observations themselves are a probability distribution

The fit does not prove causation, but it does show that the decline was already well described by a trend that began years before generative AI. If the claim is that 2023 created a separate structural break, it's different claim then the title describes

reply
You can't call it a bell curve unless it's from the Charles Murray region. It's just sparkling statistics.
reply
Do you mean "just because it's a bell curve, doesn't make it a normal distribution"?
reply
No, I meant what I said.
reply
deleted
reply
Those graphs look nothing alike, except for "going up and then vaguely going down."
reply
I don't know what to say other than learn math
reply
You mean the fall to a thousand questions per month. Now that the volume is low enough someone has a chance of looking at every single one of them, maybe the StackOverflow community can finally collaborate in peace, safe from the onslaught of questions that could be answered by reading the documentation.
reply
I am not sure. I think SO died way before AI and that graph seems incorrect too.

> Without AI they might have lasted at least a couple more years

Nah, their decline was already readily apparent before AI. You only need to go through old discussions and other people noticing it. AI may have accelerated the decay, but the decline happened already largely prior to AI.

reply
Yes.

At the same time, this is graph is something that really should not look anything like a bell curve. So the format is probably just a coincidence.

Except if the "all the questions have been asked" hypothesis is correct. What I really doubt.

reply
this. Thanks for pointing it out, I fell for "oh it was just AI" at first.
reply
This isn't really a bell curve.
reply
check my other response, it's very much a bell curve, statistically speaking

https://postimg.cc/n9nZGLmb

reply
> statistically speaking

That's a very big word you're using there for what is basically making shapes out of clouds. A bell-curve is the amortised function of a random variable with a mean and standar deviation. What does that have to do with a timeseries dataset?

reply
A bell curve is not an "amortised function." Amortization applies to accounting and algorithmic time complexity, not probability distributions. You're likely thinking of a Probability Density Function (PDF). If you are going to police terminology, it helps to use the correct words. Second, fitting a curve with an R^2 of 0.911 is the exact opposite of "making shapes out of clouds.
reply
> A bell-curve is the amortised function of a random variable with a mean and standard deviation.

The general notion of a bell-shaped curve is broader than that. Wikipedia has a reasonable overview: https://en.wikipedia.org/wiki/Bell-shaped_function

> “typically continuous or smooth, asymptotically approach zero for large negative/positive x, and have a single, unimodal maximum at small x.”

reply
You don't just fit a Gaussian distribution to a timeseries dataset. That's not what a Gaussian curve is designed for at all. https://www.explainxkcd.com/wiki/index.php/1725:_Linear_Regr...
reply
You are confusing a Probability Density Function (PDF) with a phenomenological curve fit. No one is claiming that time is a random variable drawn from a normal distribution.
reply
> No one is claiming that time is a random variable drawn from a normal distribution.

You are doing that implicitly by fitting a Gaussian curve.

reply
Fitting a mathematical function to a dataset does not implicitly adopt the ontological baggage of probability theory. That is a fundamental misunderstanding of applied mathematics
reply