Kaz Kylheku writes:
The main moderation problem on StackExchange sites is the abrupt closing
of questions. This is perpetrated by those contributors themselves.
>
But a constant stream of fresh question is the lifeblood of the site.
When visitors stop coming to ask quesitons, it dies.
>
Questions are often closed because they are duplicates. However,
they are often not exact duplicates.
Oh yes. The new question has a problem in its
int foo()
function. But the dupe question discussed the same error with
int foobar()
so it's not an exact duplicate.
I've heard this complaint repeatedly. Earlier this week I read a bunch of whiners on Slashdot, who were caterwauling about this. I can only speak from personal experience. Dupes can be reopened. I closed many, many question as dupes. I did not have many friends, over there, not exactly, but my dupes were reversed very, very rarely. There were quite a few other, similarly … strong… personalities, and they would've jumped at any opportunity to reverse my dupes. But I can count on my fingers how many times it actually happened.
Similarly I don't recall ever reopening someone else's dupe. Maybe I did, but I don't remember it.
Maaaaybe, just maaaaybe, you know what? They were, really, dupes.
I can say with absolute certainty that there wasn't any kind of a cabal over there, that banded together to close questions as dupes, left and right, with some kind of a gentleman's agreement not to reopen each other's dupes. So, if the complaint is that there were a lot of question that were getting closed as dupes, the explanation for that is that …they were dupes.
I recognize that this is cited frequently as a SO problem. I just don't think this is the case, based on direct personal experience. I conclude that, either:
1) All high karma contributors on SO were assholes who enjoyed closing questions as dupes, and who were doing it without any coordination between themselves, whatsoever. I never coordinated anything, of any sort, with anyone else. Or,
2) Maybe, just maybe, all the dupes – or at least most of them – were proper. Perhaps they're…
Moreover, people ask duplicate questions because the site's search
function is garbage: the answer is in there, but they were not able to
find it.
…reasonable dupes because of this factor. Perhaps. But the resulting dupe closure self-corrects because…
StackExchange pushes the narrative that questions and their answers
should be useful to future visitors. But then they rely on Google
for those visitors to actually find them.
… the dupe closures include links to the dupes. And I often spent extra times on the dupes, very often I added additional links links to two or three more extra questions, to the dupe closure. Once a question is closed as a dupe it was possible to edit the dupe question list, and add (or even remove) the dupe questions. I can only speak for myself, but I'd like to think that I'm not special in any way, so if I was doing that I want to think that others were doing the same too, to help point the teeming masses at all that prior art.
I think the problem went beyond dupes. It was three-fold:
1) The low-information teeming masses with self-esteem problems often interpreted dupe closures as a grave, personal insult of some sort. At the very least, if a dupe was not a 100% match, it was a 99% match and the remaining 1% was some secondary factor that anyone with a bare minimum of subject matter knowledge should be able to discern on their own. But the low- information teeming masses demand everything to be served to them on a silver platter and refuse to do even the bare minimum of neuron shuffling, to accomplish that.
2) I observed, from direct experience, a direct colleration between someone's newbie factor, and their snowflake factor. That is, the more someone is a newbie, the more likely is that this someone is also a fragile snowflake, who considers their question getting closed as a personal insult. See 1.
3) SO's attempt to remedy this growing conflict only exacerbated it, and only made it worse. Their imposition of, essentially, a Snowflake Bill Of Rights had the inevitable result. Telling a bunch of snowflakes that they're special, that they will be named George, and Stackoverflow will hug them, and pet them, and give them security, and will keep them warm like a mother, all of that only encourages the snowflakes to simply become bigger snowflakes. That's what always happens in these situations.
Appeasement never reduces social conflict, it only encourages more of it. The experienced contributors are far more likely to have better emotional control, but even they will eventually reach, individually, a point where contributing on SO is no longer worth it, for whatever it was worth for them. Getting shit on, in response, eventually gets old. It's only a question of how deep the shitpile has to be, before it's time to say: so long and thanks for all the fish. And without the experienced contributors, SO becomes just a dusty, flaming, dumpster fire with everyone else accusing everyone else of not being welcoming enough.
As I wrote previously: social media site who reach sufficient size will have sufficient momentum to carry themselves forward, while continuing to be a flaming dumpster fire, like Facebook. The problem with SO is that they did not reach the sufficient size to become that self-sufficient dumpster fire.
When you do that, you are handing (even more) control over your traffic
to Google.
>
Google served up site summaries without routing visitors to the actual
sites, even before the rise of LLM AI.
Dunno. I read those summaries too. And I still follow through the links. Just earlier today this happened: I was brushing up on all the crap that went into C++ in the last two revisions. I remembered the specific term I wanted to review. Google came back with an entire page of an AI-generated answer.
It was actually quite good, but I still clicked through to cppreference.com. Besides, before I forget: I wanted to mention another contributing factor to SO's fall from grace: everyone who ended up on SO, while looking for
www.pleasewritemycodeforme.com or
www.pleasedebugmycodeforme.net. I don't recall Google ever serving up complete code in their AI summary, so that entire crowd will still follow the links in the search results.