Timeline for GPT on the platform: Data, actions, and outcomes
Current License: CC BY-SA 4.0
43 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 4, 2023 at 6:11 | comment | added | Karl Knechtel | "Why aren't we instead talking about the problem of loss of askers and questions? There are no answers without questions. Losing questions is logically a more root problem. Why not give attention to the more root problem?" - it's at the root, but it's not a problem. For Stack Overflow to improve, it requires an extended period of net negative question influx rate as the garbage is cleaned up. | |
| Jun 15, 2023 at 18:54 | comment | added | Braiam | "available questions" here is an aggregate of all visible questions that can be answered. Basically, open questions by a given date. What you are graphing in that section is new questions asked. Different definitions. | |
| Jun 15, 2023 at 5:29 | comment | added | JonathanReez | Fantastic answer. It’s mind boggling that C-suite execs at SE have less understanding of the platform than the users. | |
| Jun 14, 2023 at 5:06 | comment | added | Kevin B | effectively, his data shows that, despite the question rate going down by x%, the number of "active answerers", which are answerers who post 3 or more answers a month, dropped not also by x%, but by x%+n, aka more than he expected it to if the drop was solely due to the question rate dropping. There's certainly better arguments for why these rates are different than because mods are suspending them, but the data doesn't prove or disprove that it's simply due to less questions being asked either. that's just another hypothesis | |
| Jun 14, 2023 at 4:59 | comment | added | starball | @KevinB the perspective from which that argument is made and its data gathered and presented is (in my view) warped (and because of that, I agree that it's a weak argument). My second graph is intended to show that the falling rate of questions likely does explain the falling rate of answers. | |
| Jun 14, 2023 at 4:52 | comment | added | Kevin B | @starball yes, the data he provides next that shows that there are more questions being provided than there are answerers, on average on a day to day basis. He's claiming the falling rate of questions isn't as high as the falling rate of active answerers, and therefore the falling rate of questions doesn't explain the falling rate of active answerers. I agree that this is a weak argument, but it doesn't change that your data in the first chart doesn't in any way disprove it. He accepts that data as fact right out of the gate. | |
| Jun 14, 2023 at 4:50 | comment | added | starball | @KevinB but then why does that full sentence read "The alternative hypothesis for the above chart is that the number of questions available for users to answer has simply fallen, on account of question rates falling. This claim is hard to swallow given current data." ? My assumption is that "hard to swallow" here means "we don't believe this is the case given the data" and not "we believe it's happening and just wish it weren't the case". | |
| Jun 14, 2023 at 4:49 | comment | added | Kevin B | put another way, he's literally stating the same data you're claiming disproves his statement in the first sentence... how does that make any sense if he's trying to argue that question rates aren't falling? he literally asserts that they are. | |
| Jun 14, 2023 at 4:43 | comment | added | Kevin B | @starball Yes, that's exactly the quote i think you're misrepresenting. The first sentence in that section is "The alternative hypothesis for the above chart is that the number of questions available for users to answer has simply fallen, on account of question rates falling." which more aligns with my interpretation, as does the data he presented to support it. He clearly agrees here that question rates are falling in general. He's trying to disprove that that is why we're seeing a reduction in activity from frequent answerers. | |
| Jun 14, 2023 at 4:40 | comment | added | starball | @KevinB I disagree that that isn't a theory that was presented. Quoting the question post: "The total volume of questions available to frequent answerers continues to rise [...] What follows is the # of available questions posted per week [...]" | |
| Jun 14, 2023 at 4:35 | comment | added | NoDataDumpNoContribution | @Zhaph-BenDuguid """The total volume of questions available to frequent answerers continues to rise"" This observation from Philippe is not so relevant. Answerers predominantly focus on questions not older than 7 days. Older questions just hang around and don't seem to be that interesting fruit. Looking at the number of new questions as starball does is the more interesting quantity. The company unfortunately is looking at the wrong thing. | |
| Jun 14, 2023 at 0:51 | history | edited | starball | CC BY-SA 4.0 |
remove part of the answer where my wording / point is wonky and weak.
|
| Jun 13, 2023 at 14:19 | comment | added | Kevin B | eh, zhaph is correct. "Your assertion that the number of available questions is rising is completely contrary to reality" Philippe never made that assertion. I'm all for pointing out flaws in the theory, but lets at least make sure we're pointing out flaws in the theory presented, not one that wasn't. | |
| Jun 13, 2023 at 5:59 | history | edited | starball | CC BY-SA 4.0 |
added 57 characters in body
|
| Jun 13, 2023 at 5:00 | history | edited | starball | CC BY-SA 4.0 |
whoops
|
| Jun 13, 2023 at 1:41 | history | edited | starball | CC BY-SA 4.0 |
added 657 characters in body
|
| Jun 10, 2023 at 22:07 | history | edited | starball | CC BY-SA 4.0 |
added 497 characters in body
|
| Jun 10, 2023 at 20:34 | comment | added | starball | @Zhaph-BenDuguid how so? I'm a frequent answerer. All else being constant, other frequent answerers answering less doesn't mean there are more questions for me to answer. It could mean there are more questions with no answers for me to answer, but that's not part of the argument Philippe presented. And all else is not constant here. As I've shown, questions influx is declining rapidly. Either I'm still missing something (and if so, I apologize), or I think you've falling for a pretty obvious logic trap. | |
| Jun 10, 2023 at 18:48 | comment | added | Mazura | relation between decreased rate of incoming unique questions, and decrease rate of answers. - I have an 8yo answer that I've linked to twenty seven times. It wasn't GPT, it was getting to 20k, now seeing all the deleted garbage making it look like the rest of the internet. And there's quite a few of my peers who joined around the same time as me who are now ~100k because they keep answering dupes, instead of doing it right. | |
| Jun 10, 2023 at 12:42 | comment | added | Zhaph - Ben Duguid | I agree with a lot of your comments but I'd also point out Philippe's statement (emphasis added) was "The total volume of questions available to frequent answerers continues to rise". If the number of frequent answerers is falling (which Philippe states is happening) faster than the number of asked questions, then this statement is still true. | |
| Jun 10, 2023 at 0:19 | history | edited | starball | CC BY-SA 4.0 |
added 404 characters in body
|
| Jun 9, 2023 at 22:31 | history | edited | starball | CC BY-SA 4.0 |
added 386 characters in body
|
| Jun 9, 2023 at 22:17 | comment | added | NoDataDumpNoContribution | Very instructive answer. In the graph with the ratio of average questions and average answers there seems to be a roughly yearly periodic effect. Are incoming students with lots of questions the reason? Also in spring and summer 2014 SO took a huge hit before stabilizing itself. I wonder why then the number of new questions went down dramatically. GPT want invented yet. Maybe we can learn something from then. The situation looks a bit similar. | |
| Jun 9, 2023 at 19:44 | history | edited | starball | CC BY-SA 4.0 |
deleted 2 characters in body
|
| Jun 9, 2023 at 11:22 | comment | added | Johannes Kuhn | The first 2 statements do not necessarily contradict each other - there can be more unanswered question than ever right now, some of them old. | |
| Jun 9, 2023 at 6:25 | comment | added | tripleee | I'd say the graph agrees with my expectations around Monicagate (late 2019) and COVID-19 (US lockdowns starting March 2020). There was a weak downward trend before COVID, then a huge uptick, then a stronger downward trend; the Monica incident might have contributed, but then in non-obvious ways which are hard to impossible to decouple from the COVID effects. Finally, starting in 2023 or late 2022, a steeper still downward trend, apparently due to ChatGPT. | |
| Jun 9, 2023 at 6:18 | history | edited | tripleee | CC BY-SA 4.0 |
Formatting + typos
|
| Jun 9, 2023 at 6:05 | comment | added | S.L. Barth is on codidact.com | @BryKKan I'd have expected that traffic to Stack Overflow would increase with COVID. Software development can largely be done from home, so the questions would keep coming. And answers too, because with no place to go after work, you might as well answer a few questions on SO. | |
| Jun 9, 2023 at 5:55 | history | edited | starball | CC BY-SA 4.0 |
refactoring
|
| Jun 9, 2023 at 5:37 | history | edited | starball | CC BY-SA 4.0 |
deleted 691 characters in body
|
| Jun 9, 2023 at 2:32 | history | edited | starball | CC BY-SA 4.0 |
added 405 characters in body
|
| Jun 9, 2023 at 1:34 | history | edited | starball | CC BY-SA 4.0 |
added 286 characters in body
|
| Jun 9, 2023 at 1:01 | history | edited | starball | CC BY-SA 4.0 |
added 285 characters in body
|
| Jun 9, 2023 at 0:15 | history | edited | starball | CC BY-SA 4.0 |
added 21 characters in body
|
| Jun 8, 2023 at 23:39 | history | edited | starball | CC BY-SA 4.0 |
added 1483 characters in body
|
| Jun 8, 2023 at 19:01 | history | edited | starball | CC BY-SA 4.0 |
okay that was an exaggeration. I went and got an accurate number
|
| Jun 8, 2023 at 13:48 | comment | added | BryKKan | I gotta say, it's quite interesting to me that the trend of new answer decline happens to start around the same time as the fallout from the Monica debacle. Of course, the pandemic started to ramp up around the same time as well, but I have a lot of uncertainty there. It's hard to suss from the scale of your chart, but it seems like the decline started just before COVID hit its stride. Any thoughts? | |
| Jun 8, 2023 at 8:45 | comment | added | ꓢPArcheon | no, sorry, I was quite unclear indeed. What I meant is that it would probably be better to consider December an atypical month either way because between Christmas and winter bash any data on that month is probably mostly worthless. Not a point against you, just a general consideration. I have removed the original comment, it was indeed misleading. | |
| Jun 8, 2023 at 8:42 | comment | added | starball | @SPArcheon what point are you making with that first comment? overall traffic in drops a lot in Decembers, and 2022 was no different. Is there a flaw you're pointing out in my queries or conclusions? | |
| Jun 8, 2023 at 8:42 | comment | added | ꓢPArcheon | Worth noticing that a line as "Supposing every suspension is accurate, the magnitude raises serious concerns about long-term sustainability for the site" seems to imply that since it is not sustainable to suspend every accurate detection then we should stop doing so. Which is quite in line with the picture that apparently the mods got in the private version of the policy | |
| Jun 8, 2023 at 8:32 | history | edited | starball | CC BY-SA 4.0 |
added 116 characters in body
|
| Jun 8, 2023 at 6:41 | history | edited | starball | CC BY-SA 4.0 |
added 87 characters in body
|
| Jun 8, 2023 at 5:47 | history | answered | starball | CC BY-SA 4.0 |