Timeline for answer to Do you agree with Gergely that "Stack Overflow is almost dead"? by Steve Bennett
Current License: CC BY-SA 4.0
Post Revisions
24 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 19 at 20:28 | comment | added | Michael Kay | In my area of expertise, new questions continue to attract good answers, but there are very few new questions. And I am in no doubt that the decline in questions was caused by moderators closing them before anyone was allowed to provide an answer. | |
| Mar 26 at 12:59 | comment | added | Steve Bennett | Wikipedia's editor population is pretty stable. commons.wikimedia.org/wiki/… | |
| Mar 26 at 12:58 | comment | added | 463035818_is_not_an_ai | it would be interesting to look at similar figures of eg Wikipedia and compare that to the situation of SO. I don't know any numbers, but I'd be surprised if Wikipedia did not experience a decline of influx, and its certainly not dead. | |
| Mar 26 at 12:53 | comment | added | 463035818_is_not_an_ai | I think at this point it is certain that soon it won't be the site that is has been. I still don't buy that this necessarily means "dead" | |
| Mar 26 at 12:51 | comment | added | Steve Bennett | @463035818_is_not_an_ai Correct - hence in my answer I referred to "StackOverflow as a collaborative enterprise". If there is no question/answer activity than StackOverflow is just some other kind of website, it's not the active collaboratively-edited database of knowledge that it has been up until now. | |
| Mar 26 at 12:42 | comment | added | 463035818_is_not_an_ai | your premise is that no influx of questions and answer is synonym for "dead" and you present good figures to show that the influx is actually decreasing dramatically. Though, suppose I am not convinced of the premise, ie I dont have an answer to "Is number of new questions the right metric for "life" or "death" of SO?" then your figures do nothing to convince me | |
| Mar 24 at 17:58 | comment | added | Christoph Rackwitz | if we could get daily/weekly/monthly Active Users (logins at least) broken down by each user's A/(Q+A) ratio, and then aggregated to taste (number of DAU/WAU/MAU per decile of answerer-ness), that would be interesting to see as a heatmap-type plot. then we could look for shifts in that composition. I suspect all the answerer types still visit regularly. such data would require cooperation from higher powers, if they even have historical data of logins. from the data dump, I could extract evident activity only, not logins. that might be good enough for MAU stats. | |
| Mar 21 at 19:35 | comment | added | NoDataDumpNoContribution | @Roland Sure, but it's probably not that simple as A/Q>X means growth and below X shrinkage. The graph shows that answers and questions declined with roughly the same rate for the last years. The ratio of both was probably roughly constant over the last years. Still activity shrunk in an accelerated fashion even. | |
| Mar 20 at 7:00 | comment | added | Roland | @NoDataDumpNoContribution Since you need questions to post answers, there is an expected strong correlation. What might be slightly more interesting is the number of answers per question and the time it takes to get an answer or even an accepted answer. It looks like A/Q was significantly larger during the growth phase than during the decline phase. | |
| Mar 19 at 11:51 | comment | added | NoDataDumpNoContribution | Also the absolute values from 2025 on are very small. Statistics from 2025 on might be dominated by other effects. Like the network simply not being well known as a place to ask anymore, even if one would get help here or LLMs aren't getting better. | |
| Mar 19 at 11:48 | comment | added | NoDataDumpNoContribution | "I've updated the graphs to include answer counts as well." Thanks a lot. Both react to each other. But it's too close to say which one drives the other. Could be that experts would have answered more, if only more questions would have been asked. Could be that askers would have asked more, if only experts would have answered them. Instead one or both groups decided to call it a day. | |
| Mar 19 at 7:46 | comment | added | Steve Bennett | You can see in the "decline" graph that the rate of decline was fairly steady, around 10-15% until November 2022, when it began climbing, and it hasn't stopped since. What happened in November 2022? ChatGPT was released. It's clear that LLMs are the dominant factor here. | |
| Mar 19 at 6:44 | comment | added | talex | @SteveBennett you graph doesn't show quality of answers. It may be one of factors in this situation. | |
| Mar 18 at 22:12 | comment | added | Steve Bennett | @NoDataDumpNoContribution I've updated the graphs to include answer counts as well. Answer counts and question counts behave exactly the same, although the decline in answer count is very slightly less. | |
| Mar 18 at 22:11 | history | edited | Steve Bennett | CC BY-SA 4.0 |
updated graphs to include answers
|
| Mar 18 at 19:11 | comment | added | Christoph Rackwitz | I am on the side of the counter that says ex-parrot. I also say NI! and demand a shrubbery :) | |
| Mar 18 at 10:39 | comment | added | NoDataDumpNoContribution | @MisterMiyagi You're right. It could also be both. But it also could be primarily only one. Maybe askers stopped asking earlier and more than answerers stopped answering or the other way around (after all answer to question ratio was dropping all the time). The numbers are so clear, even a blind man can see clearly that SO is dead now. But what was the cause of it, and what was only an effect? That is notoriously hard to know. | |
| Mar 18 at 10:20 | comment | added | Gimby | @ChristophRackwitz Monthy Python eh. I think the dead parrot sketch is more appropriate here :) "It's sleeping!" | |
| Mar 18 at 10:01 | comment | added | Christoph Rackwitz | to reference a Monty Python joke: it's an ex-change | |
| Mar 18 at 7:04 | history | edited | Steve Bennett | CC BY-SA 4.0 |
added 86 characters in body
|
| Mar 18 at 6:35 | comment | added | MisterMiyagi | @NoDataDumpNoContribution Why does it have to be one before the other? Both groups seem to have gradually lost people. (And it’s not like the groups are completely distinct to begin with.) | |
| Mar 18 at 6:21 | comment | added | NoDataDumpNoContribution | The numbers are very clear. What is unknown is who is ultimately responsible: did askers leave before answerers or the other day around. | |
| Mar 18 at 0:42 | history | edited | Steve Bennett | CC BY-SA 4.0 |
edited body
|
| Mar 18 at 0:33 | history | answered | Steve Bennett | CC BY-SA 4.0 |
