Slate's stash

starball

00:08

@Mithical this has crossed my mind at times. what if instead of the current state of closing things as "needs more focus", in some cases, we had directory trees that break down questions? say someone wants to ask "how do I create a website"... SUPER broad. but maybe it could be broken down into smaller, but still broad categories, and then those broken down in multiple stages by questions about considerations.

our library canonically houses leaves, but what if it could house branches too?

deciding how to break big questions down wouldn't be easy... but I think it could hold significant value to newbies.

I vaguely recall that collectives have something sort of like that, but a flat structure.

@Mithical yes, this is in the direction of what I've been occasionally wishing for

wizzwizz4

We can currently do this. Post a question, VTC it as "too broad", then edit in links to the subquestions as necessary.

starball

@Spevacus new waterguy?

@wizzwizz4 but I have never heard of this being a normal practice. I'd expect a rain of downvotes if I tried to do this without consulting meta, and I'd fear a rain of downvotes on meta if I tried to consult meta :P

@Slate no problem. if//when you do, I'm happy to talk about it in chat too if it looks like it'll get into long-ish discussion

@Slate I'm very happy to hear that. I appreciate you :)

Journeyman Geek

02:12

(shared with permission)

@Slate ^ that's one container, and the cards

Slate

05:10

@JourneymanGeek man that's a GIANT pitcher of dancong

Journeyman Geek

I am told those crazy kids drank a LOT of tea :D

Journeyman Geek

05:25

That's another one.

Kevin B

06:58

The SO Jobs situation to me was more just a disappointment. It's not that i don't think people need all the help they can get landing a good job, it's that i don't see how this in any way helps in that regard. It's a low quality search that searches a limited resultset of jobs, leaving out a lot of great listings that one for one reason or another don't meet SO's standards, which leads to there being literally no jobs in a 200 mile radius around me according to SO Jobs. :shrug:

Careers had a similar problem, except it was caused by smaller businesses being priced out of participating.

Journeyman Geek

@KevinB one of the things I find 'amusing' is - I was mildly antagonistic to way Careers 1.0 was run cause it was taking attention away from what we'd call pubplat now, and to me that was the start of problems.

Kevin B

i loved the idea of it, because it provided members of the community a way to show off their resume alongside their participation trophies here

Their best answers, questions, etc

it drove a lot of people to participate

It felt more akin to companies looking for developers, rather than developers looking for jobs

Journeyman Geek

Oh, the product? Sure

The way it was managed with respect to the rest of the things, less so

Kevin B

If i search my state on the new one, i get "No jobs right now, check back soon!" there's definitely jobs here, lol

most don't tend to actually list their pay rates because they can't compete with more populous areas

Why should SO be the arbiter of what jobs are worthy of me finding them?

Journeyman Geek

They arn't? :D

Its literally a white label copy of indeed ... for some reason

Kevin B

07:10

then why don't they show!

there's literally a ML engineer job right up the road

shows up clearly on indeed

Tinkeringbell

07:25

Got another shower-thought.... Bit related to the whole 'what's the long-term plan' thing, but it goes something like this:

- In order to have 'a community', you need to have some minimum amount of recognition, I think. So, at the very least, some kind of 'hey, I've seen that username before'.
- If the 'long-term' goal is somehow focused/related to AI/LLM, human-reviewed or even 'answer assistants' going off on their own -> that's all "bot"

So, what happens to "community" when you only have askers, that get their answers from "previous content" (the old 'community') regurgitated by softwa…

Slate

15:50

@JourneymanGeek i'd be terrified of knocking this thing over, looks like it has a very high center of gravity lol

Journeyman Geek

I mean, I would be too

Slate

16:41

900 Miles to Go

►

starball

17:08

@Tinkeringbell impersonability?

Slate

17:30

@Tinkeringbell It's an interesting question in part because it suggests there's a difference between "someone goes to an LLM provider for their answer" and "someone goes elsewhere and gets an LLM-provided answer," at least emotionally and for the general public

like those things are definitely different for some people, just don't know how many. Could speculate I guess

Tinkeringbell

19:03

@Slate It'd be interesting to find out. It seems a lot of users/comments/posts I see from MSE or mods (so the more "stuck around here" users, I guess?) are in some way or another hinting at there being a significant difference ("If I want LLM content, I can ask the LLM, why would I bother crafting a post here" along with those remarks about interacting with other humans and not a system/duplicate)

But I guess no one ever asked the "drive-by" askers why they're asking here and not asking Chat GPT

To be fair, I'm of the "why be here for LLM if I can already get that elsewhere" kind too :)

Slate

@Tinkeringbell we have, I think

in internal research studies. I'll have to dig around a bit and see what I find. I don't remember summary results

Kevin B

eh, well

an LLM can't necessarily determine what someone actually wants either

when you are struggling to solve a problem and cant find the answer, it's often because you haven't properly defined your problem... when you're in that situation it's highly likely even an LLM assisted search would be unable to make heads or tails of what you're trying to accomplish

however... an expert dev has likely either been in that position or has already helped someone else who was in that situation and thus can recognize it and point someone in the right direction

at least, that's the only way i can justify people asking questions on a stack site in general

Imagine not trying to find the answer yourself first

'that's madness right?

Slate

@KevinB surely no one would do that; it would simply be a waste of time, and we can categorically disregard this without further thought

there's a point I mentioned recently on the MSE thread actually that might be relevant

I think that's probably somewhat true, @hkotsubo. I'd be inclined to hypothesize that many people, in practice, don't actually care for what's true in the "thorough, vetted, attributed" sense, but rather what's true enough in the "I can make something with this" sense. If an LLM is wrong, some people probably believe it will either be so wrong as to be obviously useless or not wrong enough to worry about. As a knowledge service we are, of course, most interested in the "thorough, vetted, attributed" sense. Maybe there is a diverging need there - but then, I'm just speculating. — Slate ♦ 3 hours ago

again I'm sort of guessing here. But it's hard to imagine that someone who is really committed to the deep, factual answer over what's quick & dirty & gets the job done is going to an LLM

but that's not a categorization of type of person. some people might lean more onto one side than the other, but it's really a per-question assessment

some people are going to be asking "is the answer I need to this question one that requires a thorough, vetted, attributed answer? or is something to get me started, or an approximation of the truth, good enough?"

maybe the answer is the latter more often than we like to think it is

Kevin B

19:23

for myself, categorically, i'm usually not looking for a deep, factual answer... either i'm looking for documentation, a hint that will lead me to documentation, or a relevant example. search overviews do often contain enough information for me to refine my search and get to where i need to be, or to move on myself without further research

despite it potentially being useful in that regard... i'd still hate to see it be a feature on SO

Not because i don't want SO to succeed, but because the value of such things is in the fact that overviews on google or whatever else are drawing from the entire search index, where as SO would only be drawing from what is on SO; a limited set of often stagnant QA pairs

i would prefer my search results to instead include relevant stackoverflow results the way it used to

it still sometimes does... but not quite as often, and when it does it seems less relevant to what i'm looking for.

Spevacus

balpha wants us to break out the sledgehammer

thoughts?

Slate

I really like that answer

Bryan Krause

19:44

@Spevacus Not sure I'm sold that the things described as different now necessarily relate that closely to the role of Stack Overflow

And, balpha prefaced it with "if", so, I'm not sure he's convinced either

Kevin B

👍

i feel like it very well parallels my thoughts on SO, but in a more positive light, with the idea that stack can still evolve to something relevant again.

i just feel like i've been pushing that narrative, that change is needed and it will be unpopular, for a year or two now with nothing but pushback, :shrug:

meanwhile... the page i use to browse the site gets mangled

Slate

it's an uncomfortable narrative for a lot of people, internally and externally

Stack Overflow being what it is made it successful. I don't think it's exclusively due to nostalgia that folks are reluctant to approach the idea of deep change

starball

20:10

@Spevacus I know he said he doesn't know what it is, but I wonder what it is / what we (variously) think it could/should be

I'm sure the marriage of rep with privileges is on the list, which I saw Cesar's comment about

Slate

I almost think we'd need to look deeper to start

Like, lemme ask a broader question. What is the thing we are trying to generate by running the Stack Exchange machine?

"Questions and answers" is a mundane answer, almost blithe. The real questions are, what role does it serve? Does it serve it well (how useful is it)? How do we make as many questions as possible serve that useful role to as many people as possible?

Kevin B

early on, stack served two purposes, and one benefitted the other

> Ask questions, get answers, no distractions

to the new user, it was literally a place to ask questions

just ask, and you get answers or directed to a dupe

but... when the site has existed for 15 years and has 10's of millions of questions and double that answers....

more often than not the asker is deemed by default lazy because obviously the answer exists somewhere in this massive collection of answers

I don't believe there is a pure javascript question that can be asked today that isn't a duplicate if it's broken down into its fundamental parts

so... how does stack serve users who need help with javascript but can't find what they need?

i'm sure there's people out there that woudl be happy to give them a hand, if doing so wasn't seen as trying to farm rep from dupes

Slate

maybe that's really part of the problem

starball

@Slate my understanding is "searchable library of bite-sized, topical Q&A- practical questions". does it serve it well? not when for various reasons, it's not doing well on the external search interfaces it sort of depends upon or expects to serve people through. reasons like bad titles and questions, which I mentioned in my answer post, and maybe declining SEO juice, which may not be unrelated

Slate

there's nothing intrinsically wrong with providing a duplicate answer to a question, or providing an answer that points someone to where another answer might be. the problems are ex-post: a) that it creates noise for future searchers, and b) that someone theoretically could have spent their time providing new knowledge

starball

20:24

@KevinB well, the language is continuing to evolve, and so are the platforms built on it (Ex. the DOM). there will be new questions.

Slate

@KevinB part of me wonders how often duplicate voters are closing questions because they've done the work of breaking down someone's question in their head. like, if someone asks a JS question, and an expert goes "that's really a subvariant of question Y," they close it as question Y. but does the asker see the connection? ....maybe

Kevin B

that absolutely happens, has been for as long as the site has, but how often? i'm not sure

@starball cool, so, maybe there will be 2-3 new novel questions a release

Slate

probably off by a couple orders of magnitude but it's still small in comparison to the # asked so no challenge on the point itself

Kevin B

the majority of questions that arguably aren't duplicates take on the form of this stackoverflow.com/questions/79276399/… effectively, questions that ask how to accomplish a very specific task that is unlikely to 1:1 align with anything anyone else is going to do. but... techinically not a dupe, and is an entry into the knowledgebase none the less.

I don't think it necessarily adds anything of value long-term, but it's a valuable service to askers and a fun exercise for answerers.

starball

@Slate it brings up a different topic of new answers to old questions and "answer versioning", but for me the more interesting part is how new better answers take time, or now struggle to reach the top

Slate

20:29

I suspect it's sort of a red herring for an proxy problem. there are really two kinds of questions: those that get reused repeatedly for knowledge, and those that don't. the merits of a question are probably weakly correlated but the only underlying measure is: "how often are people landing here?"

put another way, if we took the 1000 questions that were most viewed in the last month, cleaned them up, made them really good, would that effort have higher ROI than the general activity of the site as it stands?

Kevin B

logic questions are effectively an infinite source of new questions that are unlikely to help other users, but at teh same time are incredibly helpful to those who have them and fun for those who answer them

but to the people who want to maintain a useful knowledge-base, they're sand

starball

@Slate wizzwizz's answer post is nicely related to this idea (referring to the first sentence in this message)

Kevin B

btw, this whole idea also stems from my battle against debugging questions, ;)

the moment you turn a logic question into a debugging question, it's 100% a dupe

that's literally the purpose of doing so

if you can figure out how the OP messed up, you can point at a dupe of someone else messing up that way too

there is value in understanding the misconception the op had that lead to their problem, and it's fun to give someone that "aha" moment of understanding where they went wrong, but more important is providing a better solution, not just the existing solution fixed

starball

@Slate it would take time to tell. and generally the most widely useful questions have been asked before ~2014, so on average, I expect no.

Slate

curious to see an example of what you mean, @KevinB, if you've got one on hand (no pressure)

Kevin B

20:47

I don’t at the moment, generally they take on the form of “How do I do this thing, I have x and I need it to become y”. The first reaction is often “What have you tried? Please provide some code!”

Probably some examples in the SG currently

And… the ask wizard does this implicitly

Slate

true, that probably preempts a good number of them

though I'm not convinced that's a better experience. if it allows someone's q to be closed as a duplicate faster, that's a curator win, but if it means a human never talks to them about their problem, that's an asker loss

people really want to talk with people non-judgmentally about their problems, even if it's just to say "this might help you"

Kevin B

the example i linked above is actually one instance where code being provided didn't turn it into a debugging question

starball

@KevinB if "this thing" is small scoped enough, I'd want to encourage that asker to just ask for the thing and don't provide any attempt, and don't entertain requests for attempts.

I think I had a chat with Catija in the tavern before on how I think failed attempts are often annoying noise in how-to questions.

chat.meta.stackexchange.com/…

Kevin B

heh, i think i've had that same conversation with catija

starball

@KevinB we might be talking about the same conversation :P

you replied to chat.meta.stackexchange.com/transcript/89?m=9692065#9692065

Kevin B

20:59

i have an answer somewhere on a similar quesiton.. i wonder what my opinion was way back then

7

A: Should Stack Exchange in general be awarding "A"s for Effort?

Should Stack Overflow be awarding “A”s for Effort? No. Effort doesn't make a question a good question or a bad question. Someone who doesn't know how to debug their code won't be able to ask a good question. They may put forth a lot of effort in writing the question and providing details, h...

definitely different in a few ways

for some reason i opted to answer one of the linked questions in the comments, and it ended up deleted months later by roomba

starball

@KevinB can you clarify what you mean in that last senttence?

Kevin B

of the answer? I'm not sure i can explain my train of thought from 2013 precisely, but maybe i meant i see no purpose in answers existing for questions like those because they can't be helpful to future users, they don't serve a purpose for the knowledgebase

i want the users to get the help they need, but i don't want the question sticking around

it's a conflict of interest, between wanting to be helpful toward users who need help, and wanting to maintain a useful knowledgebase

starball

@KevinB I feel the same

Kevin B

21:14

it creates a somewhat... toxic for lack of a better term, environment for answerers who want to serve both purposes. Produce a great answer and... this question you think should eventually be deleted never will be.

so you have to decide. Answer it and just give up on it being removed, or don't, or do and downvote it... which isn't great either and likely still won't result in deletion, :shrug:

the solutions i liked exploring to this problem revolved around making the roomba a bit more aggressive toward older posts that don't receive many views... but i'm not so sure anymore that it's necessary

the core of the problem always tends to go back to... site search sucks

and now with google becoming worse it's no longer a great alternative to it

starball

even if search didn't suck, garbage in, garbage out. titles and question content need to not suck searchability-wise

Kevin B

the only problem old questions sticking around causes is we can't find the good stuff

Slate

@Tinkeringbell I asked our researcher who's worked on questions tangential to this, and they had this to say. I'm going to quote here, but understand that a) these are qualitative findings and are the impressions of the researcher, and b) we haven't done the research that would serve to characterize prevalence:

> - From talking to newer users, early - mid career, I am hearing that people still ask questions on Stack Overflow because it offers reliability, depth, and human input that can still be super important.
> - And while we are hearing more cases of developers starting with LLMs like ChatGPT or Copilot for quick answers, we also hear they turn to Stack Overflow when they need trusted, peer-reviewed solutions, especially for complex or high-stakes problems (or when they dont have someone senior or knowledgeable to ask, e.g. small company, team of one etc)

They also note that not all developers use LLMs in their workflow to begin with - a considerable proportion don't at all.

This isn't a complete or direct answer to your question but it's quite relevant

Kevin B

i don't currently use them... but also the language i write is likely to be poorly covered by it

i did try copilot with it a year ago or so, and it tried to turn my code into python

Slate

well... have you tried taking the hint?

clearly python is the correct language for whatever you were doing

Kevin B

21:29

heh, well,

coldfusion is quite stagnant at this point, so moving to python is actually a good idea in the long run, but it isn't happening

who knows. maybe one day

Slate

coldfusion is sort of a funny name for a language

starball

@Slate fartlang.org

Slate

fear.jpg

"fartlang" is either one of the most used frameworks in modern cloud computing, or a SIGBOVIK entry, and there's no in-between

starball

@Slate it's just a joke. someone cloned the dartlang website and did some minor character changes. I have the sense of humour of a 5 year old. also, fartlang.org/guides/language/effective-dart/index.html

Slate

reminds me of the "cloud to butt" thing from like eight years ago

the only thing it's good for is pranking yourself when you install it, forget it's there, and spit-take months later

cocomac

21:41

@Slate I don't recall exactly when (and if disclosing this is over the line, feel free to delete it), but IIRC SE did interviews at some point involving questions somewhat related to what you asked on MSE (incl. why people participate). Not sure how involved you were, but you might find the results relevant or interesting.

Slate

Nah, if it's public it's public

It'd be hard for me to pin down since we've run a lot of interviews, but the person I asked is probably the person who did those interviews

starball

@cocomac is "on MSE" attached to "interviews", or "what you asked"? I can't tell and am afraid to guess.

Spevacus

@cocomac I recall hearing this too, pretty sure it was from Catija, and prior to my appointment. I'd say it's public... Somewhere.

cocomac

@starball I meant the answers might be relevant to Slate's What's on your mind? question. "on MSE" was meant to go with "what you asked". The interview wasn't conducted via MSE.

starball

@cocomac thanks

Slate

21:49

i, for one, think we should conduct all user interviews on MSE so the crowd has the opportunity to boo bad answers

4

(this is a joke)

cocomac

@Slate Found it. I believe it was September 9th, if that helps. I'll put their name in the next message

Slate

gotcha

i'll look for it later

Tinkeringbell

@Slate Interesting indeed! I'll chew on this for a while, while almost falling asleep :P (Or, as a famous character from a Dutch children's show would say: I'm going to look at this on the inside of my eyes :))

Slate

I'm somewhat concerned as to the process of getting this message inside your eyes.

Transcript for