Clojurians Log v2

I contemplated posting one of my libraries there to showcase a "REPL was more useful than the LLM" type of project, but then I remembered... I had partially generated my presentation notes using LLM and committed that to the repo. 😆

> The last usage of that reaction was more than 3 weeks ago. Just a single reaction. I stopped usage after a mod called it "inapprorpriate" and made it clear they did not approve, it reaffirmed my impression that this Slack cares a whole lot about being a welcoming space for vibe coders, and not a whole lot about the people that will quietly leave because of that.

If it was me, I said that 🔫 was inappropriate, not 🤖 .

Yup, I mostly check this slack now to say morn in #C061XGG1W and practice reading Norwegian and because I'm logged into slack for work. Most updates I care about are covered by deref. That's the thing. People just silently checkout. Most communities skew 90:10 lurkers/chatters, so you have no idea how many lurkers you are losing. Even this channel is non stop about AI.

This channel is about AI. And yes, nobody knows how many lurkers there are at any given moment - that cannot even be a data point. We have had maybe a couple of dozens of announcements that were notably AI-assisted. Among an insane amount of other content that's not related to AI at all. If that's what it takes for some people to check out - well, I can only sit and watch.

@p-himik I've been finding you quite aggressive and dismissive through this whole process (here and on #C0BLJS63J2Y). I don't know if that is your intention, but it is certainly the effect you are having on me. If you feel that means I just need to toughen up, then let me know and I shall proceed according to that response.

Apart from that, the amount of effort to engage in such conversations is extremely disproportionate to any other kind of conversation here. This channel has 400 members. Less than 2% of #C03S1KBA2. Any "we must do X" thread gathers fewer than 10 reactions (correction - a thread on the 24th of June gathered 18 reactions, the only exception I found) - and surely a lurker, if they care about this server and they agree with something, would at least leave a reaction? So from the perspective of admins, it's an extremely small but extremely loud minority that demands "X", without even enunciating what that "X" is properly, thoughtfully, in an engaged manner, without incessant quips, question dodging, cartoon gun pointing.

And of course I'm not saying that a tiny minority's feelings are invalid. But admins don't manage feelings here, we can only manage tangible (well, digitally) things. Creating channels, posting announcements, removing messages, blocking accounts, that sort of stuff. None of that can deal with "I want to feel better, like in the days of yore."

On a separate topic, @otfrom, you said: > I've been finding you quite aggressive and dismissive [...] on #C0BLJS63J2Y. Genuinely asking - could you please help me understand what exactly in my messages in #C0BLJS63J2Y was aggressive of dismissive? And how could it be phrased differently, while retaining the same info?

I for one think the admins and moderators in this workspace do a fine job of keeping things neutral and fair

While true, I wouldn't police too much based on that word. Some people still do need an outlet that wouldn't fit a skeptics channel that's true to its name, and there's little point in creating something like #ai-complaints or #ai-concerns or #ai-remonstrance.

Sure, I agree, I have no desire to police anyone. I have my own concerns about AI that, maybe luckily for me, have nothing to do with the viability of my career. Be that as it may, what I observe happening here is that people are lumping the mods into this box that makes them seem like the "this is fine dog in the burning room" while they are the principled defenders of human autonomy, when clearly the issue is significantly more complex

If I were to insist upon a definitionally sound channel, which I would not, I would push for #ai-snobbery, describing a person who looks down on something or someone because of, among other motivations which are far too messy to debate here, a feeling of emotional insecurity, a motivation for which I have a great deal of sympathy, having experienced much of it myself. (I am now taking off my advocatus diaboli hat.)

re https://clojurians.slack.com/archives/C0B7Q4XRM1U/p1785320482770149?thread_ts=1785244731.366779&cid=C0B7Q4XRM1U and on down

thank you very much for your work as a mod. Things can be very difficult especially in contentious topics like this one. My desire was to move away from skepticism and back towards building things and finding other people who wanted to build things with people who have similar principles to me. We seem to be finding each other. I hope we can keep to "lets build a thing. here are things you can use. ooh -- you built a cool thing!"

I can see that this channel must be very difficult to deal with. I'm not a supporter of AI and I don't like it. Now that there is something constructive I think I'll leave the debates on AI here. I hope you are able to find the reserves of patience you must need to deal with channels like this.

I don't "like AI" (in its present configuration) either but the thing is that people keep bandying about these phrases like "building things" (I know this comes from the CoC) without giving proper acknowledgement to the apparent subtext that that's a very vague guideline and in no way does it set any parameters on the tools people use. As other people have pointed out before here and elsewhere, many or even most of the ethical problems of LLM usage are upstream of LLMs (data control, environmental impact, the subscription model of everything, to name a few off the top of my head). I hope I am not erring too far from the realm of civil discourse by making the point that I think too many of these arguments are terribly affect-laden and emerge from these sort of romantic conceptions of what programming is as a discipline. I've seen at least one person on this slack invoke the metaphor of being like a medieval guild artisan, but take it as a literal interpretation of their job.

It's also probably worth pointing out that some of the loudest "anti-AI" voices that caused this channel to get created in the first place, quit the channel shortly after joining it, and one of those quit the whole server (only to return when their company had a job to post here). The latter protested that their "anti-AI" voice was being oppressed because we had to remind them, several times, of what the Clojure Etiquette document says. I share a lot of Eugene's frustrations as a moderator, to be honest, because we've tried really hard to get clear answers about what actions the admin team should take, to satisfy the (very vocal) minority that are complaining about AI "overwhelming" this workspace... and we just can't get those answers. The upside of this channel is that it has had the desired effect of stopping the drive-by negativity in other channels, that was in violation of the Clojure Etiquette doc—even tho' it has made for a fairly contentious channel here 🙂 I would request, again, that folks read https://www.clojure.org/community/etiquette which has been very long-standing guidance for the Clojure community (it was originally posted by Rich to the mailing list fifteen years ago!).

The etiquette guide was part of made me want a channel where I could talk about building things and have the lack of LLM involvement be part of the criteria (same as things like accessibility, bitemporality, written in clojure). I don't feel this channel makes room for any of that, and it feels like to big of an ask for other channels (and leads to people starting to debate Gen AI instead). So a channel where the lack of Gen AI was part of the filter was something I would hope would cut some of that out. After a bit of a bumpy start last night, that seems to be doing what I want. The signal to other people seems to be clear. The answers to some of my questions seem to be there too.

I hope the anti-AI voices find their way there and find some place they feel they can participate. If not, and it is a ghost town then that is fine. If it becomes a place where things are just contentious arguments all the time, then shut it down and say why. Those who want something else can follow the rule of two feet and move on.

I apologise for putting my suggestion in here without reading more of the history of this channel. I can see from reading a bit further back now and in more depth that people were really primed for a fight about it.

@otfrom I think the good thing about this community is that you can start this conversation and can have a positive end to it. I find the discussion complex on many levels, but what I find disturbing is that those that do not want or like AI, sometimes seem to project that onto people that do use it (in whatever capacity). That is something that troubles me. I am a vegetarian, for many different reasons. Am I now supposed to dislike people that aren't vegetarians? I hope not. I hope the new channel works out for you.

At least for me, I think one issue is trying to find a black and white logic, which I understand if mods have to enforce it, but in doing so it made it appear like many people have a hard-line, when in reality I suspect what most people want is common sense nuance. The constant challenge of what the "hard-line" needs to be, ironically made all opinions feel more extreme and those discussions can escalate quickly. Most of us are now forced to consume AI slop at work, pretend like we're all 10xing velocity and 100xing value, and consume AI slop at home 😝. Sometimes you need to expense more energy to consume it then the person has too to generate it. The desire for filters, or standards comes from that place. And it doesn't need to be some mod enforced process, we don't have to worry about abuse until it shows itself a problem. It can begin as norms that we hope people follow, such as disclosure (or hand written announcements). As for "human-clojure", as I see it: Did you write something yourself (code or other) that you think is interesting to share and discuss, then go ahead, if it's not true of everything in a project, point out what is and showcase those parts. Do you want to discuss abstractions, tooling, design, code style, targeted at hand-written code (not agents), then that's the channel to discuss it.

@alexander.vanelsas I mourn my loss of community. I thought I understood it, but it is smaller than I thought it was. People often react badly out of grief. It is a difficult thing to pick apart. I know I struggle with it. I see some of my drive by posts above in this channel and I feel bad that I was posting out of hurt. I'm sorry I've made life more difficult for people here.

I haven't fully developed the cross-over in my mind between the popular use of AI and what I've come to think of "profit vs value" or "extraction economy", but the Smarter Every Day pursuit of a Made in the USA SmarterScrubber occupies very similar areas of my mind. These videos I think are pretty well done. https://youtu.be/3ZTGwcHQfLY https://youtu.be/vS6HEes8daw

Nothing to feel bad about imo @otfrom I see no issue with your posts at all, and that certainly wasn't my intention to suggest that!

@didibus I think one problem with the "hard line" is that people are citing examples from OSS projects about committing generated code artifacts and trying to shoehorn that into a modification of chat etiquette in a social media platform

@bhlieberman93 Can you explain what you mean, I didn't get it.

Well, what I've seen here is that people are posting links to these manifestos about such and such a language or software project banning AI contributions, which is one thing that has many legal and moral and commercial parameters. But then the discussion drifts into the topic of how "slop" projects are discussed and announced and treated in this slack workspace, which is a pretty different issue imo, both in degree and in kind

They are not at all commensurate and I think it has led to there being all this confusion and tension amongst people in re to what the mods are supposed to be doing, etc.

People want "Clojure the language" to take a stance in the same way but afaik this is a community channel with no strict affiliation at all with anyone on the core team, despite the fact that some of them participate here socially and professionally

Clojure "core" has taken a pretty no-AI stance as far as I can see, at least for now (not that they were open much to human contributions either). And Rich has one of the most https://gist.github.com/richhickey/ea94e3741ff0a4e3af55b9fe6287887f haha.

I think people are now asking what the Clojure community etiquettes around AI should be? Some may be in the ban everything AI camp, and mods might find that both not what everyone wants and also impossible to enforce. But I feel most are more interested in finding ways to keep the community lively and engaging. That's why the discussion often turns to discuss the problem of AI slop. Slop can make it feel like the community is now just you being talked too by AI, seeing AI made stuff, and asking questions to AI about it. I think it's fair to try and address this as a real problem. There's probably a way to even welcome AI assisted work, but how do you make it feel not-sloppy ? And something genuine that you'd be interested in engaging with the author about? Or that you'd trust to use yourself?

If you're referring to https://www.clojure.org/community/etiquette I don't see anything in there that relates to or is affected by AI. It's very specifically about "making stuff" and "technical arguments" rather than "opinion pieces or diatribes". It specifically says "Occasionally, there may be disagreements about how something has been, or will be, made. These disagreements should take the form of technical arguments." (my emphasis).

I really appreciate the moderation on this slack. I agree with just about everything the mods have said here and elsewhere. I do think it's missing the point a bit to say that non-technical arguments about AI are against the spirit of the community etiquette. To illustrate, do you think that posting the https://gist.github.com/richhickey/ea94e3741ff0a4e3af55b9fe6287887f link is against the clojure etiquette? The effects of LLM usage on the community, the software industry, and the world more broadly is important. Some of the effects aren't strictly technical. Like I said, I think the mods have done a good job creating spaces for discussion, but I think AI presents new challenges (both technical and non-technical) that we shouldn't ignore.

I don't think that linked etiquette speaks of AI, it's a new problem that wasn't anticipated. Specifically on if it's okay to post messages that you didn't write yourself but generated using AI. Or if you didn't build what you're posting, have no ownership on it, and used AI to generate. Specifically I'm not saying that as an anti-AI stance, I use AI myself (in case that's surprising). But how to navigate that world and keep a healthy, vibrant and inclusive community seems a new challenge.

I think especially the closing justification: > Ignoring these guidelines fails to respect the time and effort of people who make things, which you should care about if you intend to be one. > Now seems like in a way, might apply in reverse. How do you respect the time and effort of those you're asking to check-out what you made?

fwiw, I also agree with the general sentiment that there haven't really been any concrete suggestions that would be practical from a moderation perspective.

I haven't seen it posted here, but I found the https://youtube.com/playlist?list=PLEMXAbCVnmY6U_pA-7GKuP9xiv9utLaP4&si=jsohMiKwQn1mtCVc series to be very interesting. It's a systematic discussion of many AI related topics with Casey Muratori and an AI researcher, Demetri Spanos.