Inside The Decline of Stack Exchange
thediff.co
thediff.co
In my programming job I consult SO multiple times, and I have never once felt the kind of hostile discussion environment that people try to potray. SO is very valuable to me especially for questions that have less to do with documentation but more about the essence of programming itself. It disgusts me that in future I will have to communicate with a brainless LLM instead where I have no recourse if its answer doesn't satisfy me, nor I have the confidence in the correctness of its word vomit.
I think most people dislike how hostile the main stack overflow site is to beginners, who despite being the ones who need the most help, are the most likely to get their question closed/ downvoted. It's unfortunately necessary to maintain the high quality of the site.
It's not just hostile to beginners. It's also quite hostile to experienced programmers (at least those subject to less than ideal design constraints), who are incorrectly treated like newbies and force-fed cookbook answers that fail to answer the actual question asked.
For example. Here's a paraphrase of some experiences I've had on Stack Overflow (obfuscated to preserve my anonymity):
> Me: I need to know about the implementation of this weird function in this weird proprietary language. I need to re-implement its output, which was (unwisely) directly exposed in our system's output and is depended on by our integrations. I know this sucks, but it's the problem I have to solve.
> StackOverflow: That's a bad design. You should use the XYZ function in Python instead.
> Me: You're not answering my question. I know it's a bad design, but I can't change it. The output is fixed, and I have no authority to force a change. If I can't re-implement the function, I'm suck using the weird language forever.
> StackOverflow: You should use XYZ function in Python. If you can't, talk to your architect and get budget allocated to rewrite all the downstream systems.
> Me: You're no help at all. Why did you even bother answering?
> Me (thinking to myself): Wait, I know: chasing after internet points and trying to feel high and mighty.
> StockOverflow: What if a newbie found this question? If I don't tell you to use XYZ function in Python, they might do something else instead! Like design and implement and entire less-than-ideal system!
> Me (thinking to myself): Good grief.
That being said, I realize you do have to be careful about subtle bugs.
My canonical example is “write a Python script that returns all of the AWS IAM roles that contain a given list of policies”.
The code it usually generates works correctly as long as you don’t have more than 50 roles in your account. It won’t add pagination support unless you spot the bug and tell it.
The second question I have is what type of questions do you find are better suited for Stack Overflow than ChatGPT? Are they questions based on post 2021 knowledge?
This is not always as easy for all types of questions. It's hard to come up with an example on the spot, but I tried a few queries I searched on SO in the last week, and found one to demonstrate. I was searching for "Where to store JWT in browser?". This is the SO answer [1] for reference. Now to prove my point, I ask the question to ChatGPT (3.5). Here's the chat [2].
On a first glance it looks like ChatGPT may have nailed it, though the information dump is HUGE. Among the 6 options it suggests, the last option "secure cookies" looks "correct". Indeed it is correct in what it says that this can prevent XSS attack. But it is not complete. Because it still does not prevent XSRF attack entirely. So I had to explicitly prompt it to think about XSRF and its response is weird. At first it incorrectly claims that XSRF attack is mitigated, but then in the response body it elaborates that we also need anti-XSRF tokens for complete protection. So I don't know what to make of it. Contrast this with the SO answer which is way more direct.
Honestly, ChatGPT's answer looks like that of a student who is trying to impress some examiner with their knowledge dump by beating around the bush rather than trying to precisely answer the actual question.
[1]: https://stackoverflow.com/questions/27067251/where-to-store-... [2]: https://chat.openai.com/share/c26fee93-5d3d-48e2-a820-297974...
Which is it you are referring to here?
Everybody is trying to do their best but the new askers don't have the experience to not ask their question and the experienced SO users have seen that question asked a dozen times already.
I have had questions closed as duplicate, where the "original" asked something else. Questions about choice of framework apparently are off topic on SE stack exchange -> closed. Weirdo moderators / keen users butting in on active, sane questions with "I'm voting to close because [random nonsensical rule]".
I feel bad writing it as I suspect most moderators on SO are great, but this is nonetheless my overwhelming experience of past 12 months.
EDIT I forget about the patronizing users who didn't read the question, but don't let that stop them from lecturing you. On SO rules, life mistakes...
There's also the issue of people trying to improve their SO score because that can impact things like job interviews, at least on the programming side, and as with any such system gaming it for high reputation (points for questions, answers, upvotes) has some side effects on people's behavior that aren't always pleasant. There's also a saturation effect, e.g. many common questions have already been asked, and the 'archive of truth' mentality means that you're supposed to search past answers before asking anything, which again tends to take a lot longer than ChatGPT does.
Same for that "answers over time" graph: that's literally what you'd expect for a knowledge base. There has to be an initial trend upwards as the site starts to accumulate information (i.e. questions with answers, not just questions) and initially gains popularity as a place to get help by asking questions. But, as the number of questions and answers increases, the utility changes from a place to "get help by asking questions" to "a place to get help by searching for answers". At that point, the curve levels off, and as the number of typical beginner and intermediate questions die off because they've already been asked and answered many times, the number of new questions has to trend downward. It'll be slow, because we're constantly inventing new programming languages, libraries, frameworks, etc, but the majority of questions for those are simply duplicates of things asked about previous languages, libraries, and frameworks, with their answers being perfectly usable by folks looking for answers.
And this is exactly the graph we see: rather than a sign of decline, it's a graph that shows the change from a predominantly Q&A site to predominantly being a knowledge repository.
ChatGPT will, of course, have some impact on the graph, but the number of posts on Stackoverflow today that are "How do I X? I asked ChatGPT but that didn't help me" is staggering. ChatGPT may have initially driven down searches across the rest of the web, from Google to Stackoverflow to your mom's recipe website, but the novelty's already worn off, and overall it's done precious little to cut into the utility of Stackoverflow, as far as this data suggests.
SO has accessibility problems for beginners and a meta game for more advanced users. But there is no evidence of a death spiral here.
If anything… GPT could help SO by relieving the massive number of basic questions that GPT actually can answer.
Yes, absolutely, and the article addresses precisely that point. But the questions graph shows a drop of about 40% at exactly the point in time when ChatGPT launches, which is much too extreme to explain away like that.
It's Q3 of 2023, if you're going to argue a decline, why are you not looking at the most important three quarters of data?
That said, some of these are going back up. Java and C#, for instance, are clearly trending up again, as is Swift. Python's down, but by 10%ish, not 40%. And JS is down a lot but that's the one language you'd expect ChatGPT to actually be the best at "helping" with, given that it's the web's own programming language, so it being trained disproportionately more on JS than anything else would be hardly surprising.
And then there's languages for which ChatGPT might as well not exist: Lisp, Go, Rust, etc. Heck, if we were to look at Rust, for instance, ChatGPT did literally nothing. Its rise has been, and continues to be, meteoric. And if we look at arduino questions, you might even conclude that ChatGPT was a factor in driving people to Stackoverflow, not away from it.
Data science is hard, and certainly much harder than most folks (especially those with a programming background) think it is.
That sneering and the general idea that some questions are worthy and others not is one of the most offputting things about SE, followed by duplicative and unhelpful me-too answers that are very obvious reputation farming. When I took up programming again after a long hiatus I found SE very helpful at first but got sick of it within a matter of months because the meta game/* is horrible.
* the social dynamics in a community driven website that are wholly orthogonal to and often end up subverting the site's stated purpose by leveraging the stated ethos and decision infrastructure in pursuit of selfish ends. Other examples include Wikipedia edit wars or abusive forms of legalism and political brinksmanship.
Their knowledge repository is slowly rotting under the weight of having to ask every time "okay, is this correct-sounding, highly upvoted answer actually (still) correct, or is it 10 years out of date?"
The fact that there's very no off topic also makes it nearly impossible to "build" a community out of the site as well, it's all too focused on just questions and answers and I feel no need to go back to talk to anyone because I've never met anyone at SO while I made many longtime friends in other forums.
It also saddens me that SO has killed all these smaller and more or less insular communities, i see very little movement on the old mailing lists and forums that are still up, all questions moved to SO in a way or another (even for non-English speaking communities) so I'm not sure if the site has been positive for the tech community in general.
I had so much coaching and help from people in the local and regional communities I was a part of when I was in college and later starting my career and now there's nothing like it around anymore because SO sucked all the oxygen around them. Not sure where i'd be if I was starting right now.
I do not believe the google trend graphs,like "the scariest graph", in this post accurately reflect a decline of SO.
There is a place for AI tools and StackOverflow is not the friendliest of places but the quality disparity between the answers you get when comparing the two is very clear to me.
> This year, overall, we’re seeing an average of ~5% less traffic compared to 2022. Stack Overflow remains a trusted resource for millions of developers and technologists. [...] Conversely, in April of this year, we saw an above average traffic decrease (~14%), which we can likely attribute to developers trying GPT-4 after it was released in March.
https://stackoverflow.blog/2023/08/08/insights-into-stack-ov...
I wonder if there's a systems lesson to be learned here around communities that do a good job welcoming new people and creating a magical onboarding experience staying vibrant for longer.
Here's how I've seen it played out before:
---
johnnydoe2002 posts a question titled "How do I get my code to work?" with two sentences explaining their problem and an attached screenshot of their code. Within two minutes it gets downvoted thrice, flagged seven times as vague, and modman♦ closes the question, but not before posting the comment reply "Welcome to SO! Thank you for your interest, blah blah, your question has been closed because it needs clarification, blah blah, [link to new user portal]"
OK maybe that was a bit hyperbolic and most new users aren't that bad, so let's say johnnydoe2002 is undeterred by this and actually formats and titles their question in a reasonable way. Within four minutes it has one short answer from repseeker and a long-form answer that is more of a code review from overeagerverbosity. The question gets no upvotes despite having two answers. Within seven minutes poweruser98 closes the question as duplicate because it received enough flags.
If I were johnnydoe2002, I would have a mix of feelings. On the one hand, someone (two people, in fact!) answered my question; on the other hand, it was summarily closed and nobody found it interesting. I can clearly tell that I am not a part of the "in-group."
But I can also commiserate with modman♦ and poweruser98. johnny's question only adds noise and they clearly didn't read the contribution guidelines. The mods and power users are willing to be nice, but they don't have the time to hold anyone's hand: in the past ten minutes they've had to deal with all sorts of questions from larryloe2001, minnymoe2003, and randyroe2000. I wouldn't be surprised if for every useful question on SO there are two that have to be closed.
---
I honestly think that changes in phrasing and formatting could help the new user experience a lot.
Like if instead of closing questions as duplicates, the mod/user in question would post an answer that says "I think this other thread answers your question" and the page is invisibly redirected everywhere else to the other thread. Essentially "shadow closing" so that it appears to the new user as if they just received immediate feedback in the form of a (helpful) answer, while not being told that they are bad because they dared post a duplicate.
Or instead of closing a question as "needs clarification," sending it back to the edit modal with some adjoining commentary or links that explains the clarification it needs.
I think that for a community with standards these exacting, most users will not be able to just pop in - which is difficult because we usually refer to SO in passing. But I also think that those exacting standards can be upheld without stomping over new users with downvotes and close votes.
34.223.203.0/28 what is the end ip of this range?
Write python code to strip out characters that are not alphanumeric or spaces
How do make FastAPI not return nulls in the response model when values don't exist?
Rewrite this javascript to python <snippet of javascript code>
ChatGPT doesn't offer that kind of user feedback, so you have no measure if the answer is valid if you don't know the topic well enough
ChatGPT does offer user feedback of a kind though -- e.g. I can tell it "no, not like that, more like x" and it will take that into consideration.
It’s a bit like reading HN comment threads vs. asking an LLM for its take on a submission.
With ChatGPT, when it provides incorrect answers I've had good luck explaining the error. ChatGPT usually revises answers into useful output.
I am frequently saying things like "no, that's not right, ..." or "no, that's not what I meant, ..." or "could you keep this part, which seems good, but replace the rest with a different approach like ...".
Think how rude it would be to micro-manage a volunteer answer on SO!
The site is a big douchey middle finger. A dismissive know it all that actually doesn’t know all that much.
I don't at all think it's making me dumber. I'm spending more of my time learning and doing the far more interesting and creative domain-specific parts of my work. That is, it's making me smarter by freeing up time to focus on things that matter a lot more. Just like any other tool I've learned to use to automate away drudgery in the past.
Take a pocket calculator for example. If you already know arithmetic, calculators make you more efficient, but they are a terrible way to teach arithmetic.
ML trained on all the things ever written about programming is going to get things wrong quite a lot of the time, and you'll only have the context of its reply to you, for you to figure out whether the ML answer is correct. And of course, ML always wants to please you with an answer.
SO, on the other hand has accepted answers, and upvoted answers that are easier to decide on whether they're correct (or what you're actually looking for). And there's more nuance to it as well.
Good luck figuring out if it incorporated the context in the content of the answer though.
The internet already had too many low quality, misinformed pages. With Chat GPT, it now has an infinite number of pages like that. I’m not sure that makes a qualitative difference.
Is that correct? I would think that its ‘interpolation’ and ‘extrapolation’ capabilities can easily make it produce incorrect statements.
As a simple example, train it on “π is a number”, “1 is a number”, “3/7 is a number”, “1.3457 is a number”, etc.
and
“1 is a rational number”, “3/7 is a rational number”, “1.3457 is a rational number”, etc.
and I give you a good chance that it will ‘think’ “π is a rational number”.
A remarkable thing about asking ChatGPT questions is that you don't have to be defensive, you can just ask your question no matter how basic, dumb, or simple it is without justification.
Seems a bad state of things that the inhuman AI is extending more grace than the human community.
If you ask a bad question to ChatGPT, you waste no one's time but your own.
If answering SE was a job where you got paid, I would expect a lot more hand-holding (and some amount of kindness, to keep the revenue stream coming back), but instead, the people who answer questions do it more-or-less for fun. Having someone waste your time isn't fun.
> hundreds upon hundreds of easily-researchable questions titled "plz halp i am n00b" asked by people who didn't bother to apply a slightest amout of elementary analytic skill
I'm sure ChatGPT gets not just "hundreds upon hundreds" of these each day, but thousands upon thousands of them each hour. And it remains cheery about it. Which is a much better experience for its users. It just turns out to be a better tool for this job than relying on the good will of human volunteers.
The content: A random guy looked at Google trend charts and fabricated some conclusions.
The opposite of "inside".
(They sold it years ago to a South African private equity venture and made the pile of money they deserved.)
But a serious question: is Joel still in charge of SO? Has he retired?
Obvs your use case may be wildly different, but I detest ms office for emails: 'received three days ago...Two days ago...'. Great, now i have to remember what today is and then do some mathematical gymnastics (i dont work m-f).
I've had some seasoned developers review my (ChatGPT's) Python and got positive feedback with some constructive suggestions.
so it checks out, it’s AI generated
I am able to give a lot of information about what I need, and I'm experienced enough to know precisely what to ask for. There's also a huge amount of training data about these tools. I'm also experienced enough to be able to evaluate the output for correctness. I would never ask anything like "write a program in Python for <blank>, at this point.
It is pretty good at writing little DTOs based upon a sample JSON too.
But harder questions for poorly documented libraries will exhibit a lot of the same challenges as Googling.
[0] Who is on HN; hello, nice app :) https://news.ycombinator.com/user?id=liuliu
... that it doesn't work exactly "as printed" when copy-pasted into an IDE? Or that it fundamentally misses the intent of the question? Or that it got some important key details right and messed-up other things?
Usually, in my experience, it's the last one. Often all you need is a clue to get you thinking in the right direction. Interactively asking chat-gpt questions and critically evaluating it's answers has been far more helpful than I could have ever imagined.
Does it ALWAYS help? No.
Is it an exciting and emergent alternative to what we've had before? Hell yes.
The first party docs are, of course, the gold standard. But they're often far more detailed and canonical than folks are ready to absorb.
SO can be helpful IF you can find your question already there and the answers are still relevant, or if you manage to ask your question in exactly the right way and the duplicate-police haven't shut it down (regardless of whether it's actually dupe). Or if a kind soul has provided a clue for you in a comment as your question was downvoted into oblivion.
Another option, IMHO, is github issues/discussion. The library developers are often there and generous about helping people out (without the perverse and infantile incentives of gamification that afflicts SO). But this option should really only be reserved for very carefully asked questions. In a way it has a higher bar than SO (but without the negative reinforcement of downvotes).
If it counts, I have a SO account with >15k "reputation."
1. Stack Exchange, a network of Q&A sites built around a single model
2. Stack Overflow, a specific Q&A site for questions about programming, on which that network was founded
Programming- and computing-related SE sites are the most popular, but they are not the only ones. ChatGPT cannot read a circuit schematic, should not be used for legal questions, probably won't recognize an obscure science fiction book based on one plot element you hazily remember, and in general is not a reliable expert in anything.
I say this is pedantic because if the programming sites "decline" they'll probably take the rest of the network with them, but as a long time denizen of EE.SE I would like if it people at least recognized that SE is used for more interesting things than "how do I do [basic programming task X]?".
https://trends.google.com/trends/explore?date=today%205-y&q=...
Maybe a little something going on with typeerror? Or just noise. But it definitely doesn't look to me like evidence supporting the article's claim.
It becomes tiresome to wade through the self-important moderation to get to actual content. That said, spam and unmoderated channels are clearly worse too. I'm not sure how to get to something in the middle of that Laffer curve.
There's also a supply and demand problem of double-ended Q&A sites: there's rarely direct incentives to give away subject matter expertise. There are indirect incentives such as reputation, but behind nicknames, this doesn't necessarily translate into industry reputation. Furthermore, there are potential human language and subject matter terminology difficulties of throwing geographically-, and industry-diverse groups of users together. The platform should attempt to ameliorate language differences and steer user-contributed content away from devolving into ego battles, bikeshedding, nitpicking, or handwaving.
I almost was a moderator for one of the SO sister sites, but I thought it was just too much thankless work, and I can understand why it yields petulant people who would rather see less content that conforms to their expectations than to create and make space for new ways of doing things.
The content itself is an attractant to people like that, once you have an established corpus to corral you perceive ownership and want to "protect" it.
I genuinely don't know if there's any way to fix what I see as inherent problems in online communities, that generate "open source" knowledge at least.
They have already banned machine generated content. When the question of training legality will be settled in courts - predictably, not to the advantage of the AI companies - Stack Exchange will be left with the human network generating the content, while AI companies will need to license any new answers it they don't want their models to remain stale and out of date.
Having said that with Stack Overflow the real reason I hit ChatGPT first in many cases is FRICTION. The UI, the users, the answers that are technically correct but "weird". It's a lot of filtering I have to do as a user that I don't do with ChatGPT or other AI tools.
I would have chosen another word for them...
I'd say that it is in OpenAI's interest to pay for SO and Redit so that they don't kill the hen with the golden eggs.
Anything involving genuine reasoning needs to be done by a human. It’s a great documentation parser. A step up in search engines. But a thinking machine it is not (yet).
I'm also not sure how "reasoning" at inference time is relevant here. You seem to be conflating unsupervised offline training with inference. The ironic thing is that the behavior you describe actually works, you can try it right now. Add code documentation to your context window and ChatGPT will happily write code using that documentation as reference.
In ChatGPT's instance they can be downright wrong and I have to verify if what it's saying is indeed correct on many occasions.
For me, 95% of the time it is enough to read the first couple lines of the question, then hit the "End" key and read the least-upvoted answer.
I’m in my 20’s and this is the first time I feel like my work routine and ways of interacting with tech is getting antiquated. God damn.
The ridiculous (IMHO) moderation under the guise of “we have only one purpose” often didn’t even align with that goal.
The writing was on the wall.
So many questions that could have served answers are locked, deleted, and unfortunately, never asked due to the unpleasantness of dealing with the moderation.
I speak for myself.
The meta games going on in SO are a whole lot more of a mess and ChatGPT is more than happy to hear my question a 3rd time and spit out the same answer as before but change a few things. SO, not so much.
It's not like I'm not validating SO answers, there's not any extra steps from my SO experience. Also SO likes to give oddly, technically correct, but also hard to work with code sometimes to the point of being unusable and I take someone else's example anyway. ChatGPT is really good at making super generic answers and quick iterations / alterations.
ChatGPT's power is often in the "conversation" where I add or take things away or change things "I need to break out X,Y,Z before I render it because..." and blamo I get new code. SO has none of that.
This is the internet -- lies and false claims are the standard. It's up to individuals to corroborate that the info they're looking at is reliable. Google provides a list of relevant results to a keyword search -- every one of those results can be bogus and no one bats an eye. ChatGPT provides text output based on a text input and the text can be bogus. Why should ChatGPT be held to a higher standard than Google?
GPT will tell you it’s right… well, it’ll apologize, offer another solution that could also be wrong and now you are in a loop.
I 100% believe that ChatGPT can output absolute garbage but for the same reason you don't cite Wikipedia/Google you won't be citing ChatGPT. Just like Wikipedia or Google It's just an excellent starting point to rapidly get a working draft going.
I’ve been trying to explain to people that GPT is good for subjective things, not object of things. But that didn’t really jive with the programming part that works OK for certain tasks.
Then I realized it does and programming is just subjective a lot of the time. You need a right answer but how you get there is open.
Using it as a spring board is good advice.
I feel like the people writing those answers should also be well aware that often you can't just change the problem or just add in a library to solve one specific problem on a whim. It's kind of demeaning to push those kinds of answers anyway rather than maybe question the task but provide a solution anyway.
The problem to me always seemed like the goals of the company didn’t line up with the goals of many (most?) of its users.
Most question-askers - who actually post a question - want to have a response customized to their exact problem. ChatGPT is infinitely patient, if a million users give it the same prompt a hundred times a day it will provide each of them an answer. They want to interact with it like a Discord chatroom - post a question, get answer.
Most answer-posters want to avoid posting the same response over, and over, and over. They want other users to interact with it like an encyclopedia of FAQs - or rather, questions frequently pondered but only asked once.
I'd argue that most answer-seekers probably want a single, high-ranking canonical question and answer as well; they don't want low-quality copy-pasted single-line responses in a chatroom. Maybe they upvote the pre-existing question, maybe they're just anonymous viewers. Stack Overflow the organization, like this group, wants the best-written, researched, articulated response to any particular question to rank highly on Google.
What's for sure is that no one looking for an answer ever wants to see "Closed as duplicate". The question is how you keep that from happening, unfortunately it's probably untenable to do anything but have a small cadre of community moderators (with all the self-selection trouble that causes) closing questions subjectively marked as duplicates and intensely frustrating the people who posted those questions. Nor does anyone want to (nor are they able to) scroll up through 500GB of chat logs to find a prior Q&A by someone else.
On the one hand I get their argument that they want only the highest quality answers on there but in that heavy handed moderation they now have another problem which is no one outside of a handful of individuals actually wants to post or contribute.
I think if SO didn't have such heavy handed moderation practices they would have lower quality answers on there but they would also not be experiencing such a gradual decline.
My take is that SO was born in a time where ethics among the primarily male and tech-fanatical audience of the internet was less developed. At the time, what we now would call toxicity was expected: we thought it was shameful to not ask a question properly, to waste others' time, to not have tried significantly before asking. Now, we have a different view: first, the asker may not be a technical person and may not be someone you would feel comfortable berating; second, even among hardcore tech people there is a much stronger emphasis on appropriate communication and mental health. I remembered stories of my step father giving his subordinates "bollockings" (British term for heavily and aggressively shouting at and disciplining someone who has made a mistake). Well, I work for him now, and he does not do this anymore. Everyone is quite scared of upsetting everyone else, everyone is more aware of the suffering of others, etc.
Somehow, SO has partly retained these aspects of the old internet, when almost all other spaces have shed them.
EDI: You can see the user "rejectfinite" has replied to you with a flagged comment which demonstrates how people still try to act out this kind of toxicity, but note the swiftness of his downvotes and his almost immediate flagging. We just don't tolerate that anymore. But we used to
A point made well. I saw that post.
I myself have been generously downvoted, which is understandable given the audience here on HN.
This came up in a podcast episode I did and then turned up on twitter.
At one point in 2013/2014, an exec asked me if any of the developers had ever treated me badly because I was a woman. My honest reply at the time was “not because I’m a woman, but they definitely have treated me badly because I’m not an engineer”.
Ben: This was a meetup at, I think it was Denver 2014. We were standing outside an arcade. The evening outing was a lot of pinball fun and air hockey. And one of our community managers was standing outside, and we got to talking, and he said to me, “You know, I was actually kind of surprised. You’re a really nice guy.”
It's interesting to me as a warning.
I think treating developers really well is important. But SO seems like an interesting case where they allow entitled behaviour of some devs to get out of control and it led to problems.
Dev's who are labeled as high performers get greenfield projects that can be pumped out quickly / look good ... someone else does the maintenance, they get all the flack for an obvious bug that wasn't fixed even if they didn't make it.
The random executive who is thought to be highly successful who jumps from department to department spreading his fingerprint long enough to make an impact and leaves quickly enough to avoid any responsibility for the results.
I fear it is a human thing and people are really good at sensing when they can get away with things.
A well-designed system at launch is not sufficient to provide sustained value, but competent ongoing maintenance is. Phrased another way, you can sustain value by maintaining a badly designed system but not by not maintaining a well-designed system.
Ideally systems are well designed and well maintained but we don’t always have that luxury.
Circling back to your point, it’s ironic that maintenance is the lowest-value activity in terms of big tech company reward systems.
I worked on a periphery of a big mainframe system that was designed and implemented before I was born. I think release 1 was in 1975. It still runs really well and is hard to displace because operationally, even though the people who understood it fully are retired or dead, they left behind strong materials so a moderately intelligent person can run everything.
The “other” systems, old .Net, Java, etc that support the interfaces to the old app are hot garbage, and tend to steal the budget dollars that are needed to migrate off the 1970s stuff. My team had to fix one of these… argh.
God bless those other non maintenance devs, they were good, but man they could take an RFC and turn it on its head sometimes.
It's well worth the read for the second half, and I made a post to HN because I'd love to hear more discussion on the issue. (https://news.ycombinator.com/item?id=37116593).
This is hard to quantify, and to make objective.
E.g. in a previous company HR once sent an email asking for feedback from employees, and one of my engineers gave some - polite, thoughtful, and clear. This caused quite an incident as the HR staff in question were not used to getting anything but encouragement in feedback, and I advised him to filter feedback through me, his manager.
Being "made to feel unwelcome" is both very real when you feel it, but looking slightly deeper it's not being made to feel unwelcome; it's feeling unwelcome. A mismatch of what's welcoming doesn't mean the person feeling unwelcome is correct; there may be no "correct" in that scenario. But it will be reported as though it is correct.
Any improvement suggestions might need to either get thrown in the bin (even if they're valid and simple) or filtered through their manager, or have a guided conversation so they think they came up with it.
It just depends on the person and your relationship with them, really.
This just raises so many more questions, like why bother your employee to consider the question and formulate anything if the feedback doesn't reach HR? I can imagine being someone who actually considered the question, took time to write and re-write something, another pass for clarity, a check to make sure that nothing was going to sound accusatory. Writing the email, sending it off, and hoping that the time I took away from my actual job to do this would result in even the chance of a change. But it doesn't, it goes into the round file, apparently.
I don't know, I haven't had as many jobs as some people here, but I have been through all of those regime changes and heard "we want your feedback" over and over again, yet it never seems to go anywhere. I had months on a "strategic vision" committee where we were trying to formulate a mission statement by slurping up data, questions, and forms from employees, going through various iterations of it, but what came back had nothing to do with the content we had received.
Is this always just an exercise in making HR feel good about themselves, then, without hope of any improvement?
No, I said I wouldn't send it in an email. Somewhat ironically, I wrote it clearly, but you took it differently and built a large rant on that incorrect foundation.
Next time I'll filter my comments on HN through my skip manager (-:
I sometimes wonder what innovations would be required to make HR completely outsourceable.