134

For the new genAI site, SE is planning to dramatically lower certain privileges to 1 reputation. This includes allowing every user with 1 rep to up- and downvote as well as comment everywhere. Here (screenshot) is the GenAI Meta post announcing it. I do think this kind of dramatic change requires discussion on the main meta here. I am making the assumption here that this is a change that would be potentially expanded to the network, and even if it isn't expanded it has the potential to affect the entire network due to the association bonus.

And while I think that making it easier for new users to participate fully is a nice goal, the existing barriers mostly have a purpose for existing and you need to be careful about not breaking fundamental parts here when entirely removing them.

Lowering reputation thresholds to 1 is very different from just adjusting them a bit, it removes a barrier entirely. This gives an enormous boost to sock puppets and any person that intends to use socks to perform some shenanigans. It breaks the existing defenses in multiple ways, making it much easier to create socks with actual powers and also creating new ways to misuse tools while leaving fewer traces. The full discussion here would require disclosing aspects of how mods handle socks and vote fraud that are not entirely public, so this can't happen here.

Allowing users to comment everywhere breaks the spam and troll defenses we have right now in a fundamental way. All the tools are simply not set up for monitoring comments by new users, those can appear on very old posts and are almost invisible as they don't bump the posts. Dropping this restriction would require an entirely different set of tools and review mechanisms to prevent abuse.

And if vote fraud gets out of control on one site, this can easily affect other sites via the association bonus. So this affects the network even if it stays confined to this one site. Trolls could use the site as a boost to create privileged socks to wreak havoc on other sites.

If SE wants to fundamentally change the reputation privilege system in this way the discussion should happen here on the meta sites. Not decided by the company alone and announced on a tiny private beta site. I don't think it's impossible to adjust some privileges, but it's not a trivial thing and can have dramatic effects that have to be compensated by different tools and mechanisms.

8
  • 11
    ‘For the new genAI site SE is planning to dramatically lower certain privileges to 1 reputation. This includes allowing every user with 1 rep to up- and downvote as well as comment everywhere.’ — isn’t this the case for all private betas? Commented Jul 23, 2023 at 13:06
  • 10
    @user3840170 sorry, that might have been unintentionally vague from me. This is a change intended for the public beta (the phases are weird here because the site isn't created in the usual process). Commented Jul 23, 2023 at 13:08
  • 2
    I honestly can't see how this will affect the whole SE network. There are already extreme per-site policies, for example reputation and badges not showing at all on some sites. I don't see any problem with that, as long as it was discussed and agreed in the site meta, and even if not - users can simply leave that site. Commented Jul 23, 2023 at 13:08
  • 8
    @ShadowWizardStrikesBack Hiding reputation/badges is not a per-site policy as much as it is SE fulfilling their end of their legally binding contract with the original MathOverflow people.
    – Laurel
    Commented Jul 23, 2023 at 13:13
  • 28
    Maybe we should scrap the association bonus for this new site. Then at least the impact remains localized. Commented Jul 23, 2023 at 18:31
  • 3
    Why in the world would seeing vote counts be a privilege that requires 50 rep but voting on posts to indicate they are or aren’t worthy of other user’s attention not require any experience at all with the site and how it works? That seems very wrong-headed. You can’t build a community of experts or a high quality library of knowledge by turning votes into “likes” and “dislikes’. You’re just going to reward the influence grifters, of which there are currently many in the genAI space.
    – ColleenV
    Commented Jul 24, 2023 at 10:39
  • 1
    @ColleenV I'm sure the 50 rep requirement for vote counts is purely a server/network resource-based decision. (Note, I'm not saying it's a good decision... just that that's the reason).
    – TylerH
    Commented Jul 24, 2023 at 13:38
  • 5
    I understand the concerns people voiced in the comments here but I've removed a lot of it since much of the discussion was based on confusion caused that Spencer has addressed in the accepted answer, as such, I felt a cleanup was in order.
    – Catija
    Commented Jul 24, 2023 at 15:10

7 Answers 7

40

TL;DR: We are not going through with these changes unless the site's community specifically asks for it.


The question of the GenAI reputation thresholds for privileges has caused some confusion and needs some clarity.

As stated in the question:

Of course, this was all under the premise that the site would be launching straight to a public beta, which we did deviate from and started in a short private beta instead.

Would you like to keep these thresholds? Alter them or stick to a typical public beta threshold limit for the launch.

To summarize, the question was whether any of these abnormal reputation thresholds should be kept for its upcoming public beta launch. Currently, the plan is to launch them with normal public beta reputation thresholds, unless the small community on GenAI identifies something that would be beneficial to deviate from, and even then, just for maybe a month or two, with evaluation to revert it pretty regularly.

Admittedly, that may not have been clear, for which I apologize. My mistakes here were two-fold:

  • When the original plan was to launch the site directly into a Public Beta, we discussed adopting the referenced rep thresholds in June, which then was not revisited until Friday, the 21st.
  • The text of the referenced GenAI meta post could have been more concise about framing that this was an opportunity to deviate from typical site norms due to the unique situation of this site, instead of the ambiguous nature of which it was written.

To be clear, this, like the other experiments we mentioned in the announcement for the GenAI site is in no way any kind of preview for a network-wide rollout. This was an attempt to make standard site privileges more accessible had the site launched directly into a Public Beta instead of a private one.

As many of you have demonstrated, there are potential and actual risks to other pieces of the network if it were to be rolled out as framed in the GenAI meta post. Please consider your criticisms heard and acknowledged.

2
  • 2
    Exactly what is the difference between the "public beta launch" and the site right now? Based on other questions on Meta, apparently the difference between public and private is that during the private step only users that "joined/committed/followed" the Area51 proposal can participate. But there is no Area51 proposal here and I was able to join by clicking a link. Forgive me, but I don't know how the thing works Commented Jul 24, 2023 at 14:19
  • 11
    @SPArcheon Anyone can join a SE private beta site these days. You don't need to follow or commit to anything. One of the differences between the private and public beta is that private betas are just not displayed publicly. Examples, you won't find a private beta site listed in the hamburger menu, the list of sites at stackexchange.com/sites or in the HNQ. Commented Jul 24, 2023 at 14:45
51

Something worth considering is that if these 'new' users get to 200 rep without understanding the system, through 'lower standards' - we have a whole bunch of users without good mentorship, or an idea of how to use the site getting more privileges elsewhere without understanding what's expected of them.

It also feels like, well GenAI's getting every shortcut possible - we've had a abridged process to start up the site, to meet a specific deadline I suspect.

I'm assuming that the goal here is to allow new users to have an easier start - and well part of the process of new site creation is about helping new users familiar with the topic and not the network find their feet. Occasionally when we broke new ground it needed gentle, or not so gentle changes in direction.

Teach a seagull to fish, and it'll eat for life. Give a seagull a taco chip, and they'll be harassing beachgoers for life. (or commit shoplifting)

We want new users to be equipped with skills to survive the network - which also includes organically learning to use the network. (Experienced users are probably more understanding of why we have reputation thresholds).

For GenAI - both the 'technology' and the site to be seen as a net positive - it has to, in a sense earn its way, rather than being given advantages. Considering that genAI features have been prioritised over long standing feature requests and technical debt (including A51), and now a site's being given a 'leg up' cause of a accelerated lifecycle (and yet where its convenient, its a weird hybrid of public and private beta rules) - it has not shown that its viable on its own, but rather that the company has a thumb, or maybe an entire fist on the scales.

When that potentially affects the network in quite a few potential ways, starting with 'easier' network association bonuses this is worrying.

33

I don't think this is a bad idea for a new site - that's why we have lower thresholds for "private beta" sites after all. But lacking that particular bit of friction, other mitigation will be needed.

In particular, I want to highlight the cost of cleaning up abuse much later on: abusive comments poison everything they touch and discourage constructive participation, while removal of fraudulent voting after recipients have become accustomed to a certain reputation is demoralizing. You already mentioned the trouble that a determined attacker can cause with elevated privileges...

So to do this successfully, this site will likely need very different rate limits (40 votes and unlimited comments per day is just asking for it)... And also a dedicated team of human abuse-preventers with good tooling and experience in using it.

Fortunately, staff at SO have been handling flags during the ongoing moderator strike. With 15000+ flags under their belts I expect they have the experience and skill needed to make this a non-issue for this new site.

1
  • 1
    I think a more interesting experiment than not requiring voters to be correlated at all would be changing up/down votes to a 1-5 rating. That way you don’t have posts that look ok score-wise, but are actually quite controversial and zero can be displayed as “no votes/not enough votes”. One of the most useful bits of information when I’m looking at product reviews is the distribution of the ratings.
    – ColleenV
    Commented Jul 24, 2023 at 11:06
27

I think there are a few privileges that must be fixed

  1. Close votes: 15 rep

    I have the 100 rep association bonus, which means that I, someone who has zero posts on GenAI and limited experience with AI in general, can help decide what is on or off-topic. Usually, close votes are granted after edit privileges and review queues, but for some reason their rep requirements have been massively reduced.

  2. Comment everywhere: 1 rep

    This is going to be abused by spammers; especially since there are less restrictions on commenting than on answers and questions (I think).

  3. Vote up/down: 1 rep

    As others have said, this could make sock-puppeting easier and disrupt the reputation system with undeserved association bonuses. With downvotes, new users can downvote answers as much as they want without losing any rep, since rep cannot go below 1.

  4. Remove new user restrictions (more links, answer protected questions): 1 rep

    Then what's the point of protecting questions?

  5. Chat: 20 rep

    This allows trolls to have access to chat after just two upvotes (which are easier to get on GenAI). See Laurel's comment.


In general, I believe rep thresholds on beta sites should be lowered to allow for early-stage moderation, but the system in GenAI needs some reworking. Maybe disregarding the association bonus for some privileges could help a bit (such as close votes).

1
  • 8
    To be fair, chat is a different beast because even on graduated sites like SO and MSE, they only need 20 reps. Commented Jul 24, 2023 at 4:32
10

I'm not sure why they are (wanting to) do(ing) this. Perhaps they got complaints from new users unable to comment and entering a rant.

I would think they are trying to include everybody in the site instantly after they create the account, allowing them to perform most actions (not even "privileges" anymore) before they even earn their first reputation points. By doing this, I think they're wanting to increase the number of new and active users even after the few first days.


While I agree with reducing the commenting reputation threshold, it shouldn't be set to 1 (maybe 35-40?). As explained in the question, this would encourage spammers.

Also, the "new user restrictions" (if they can be called "restrictions") are removed at 1 rep, so instantly after the creation of the account, as explained in other answers, it enables new users to perform more actions. One of those actions is called "ask or answer questions too rapidly". This means that the new user, possibly spammer, could post anything as quickly as they want. That, to me, encourages spamming.

Voting "privileges" shouldn't be adjusted if their reputation requirement isn't increased, serial voting can have serious impacts on the site and on the users.

Anyways, that's to me one of the worst ideas SE had (at least during my time here).


For info, here are the privileges acquired at 1 reputation on the site (see genAI privileges):

privileges acquired at 1 reputation on genAI

2
  • 1
    How can someone participate in the new site?
    – Ghost
    Commented Jul 23, 2023 at 18:03
  • 4
    Open GenAI and click the button. Commented Jul 23, 2023 at 18:23
5

I would like to give individual sites a bit more autonomy, let them decide for themselves more how "the system" shall work for them.

If they would decide that certain thresholds should be lower and subsequently they are overrun by spammers instead of active users, it's their problem. If they however have a capable mod team and feel up to the challenge, why not. The idea behind is that maybe not every community and every topic benefits from the same set of rules equally well and that the specific communities know best what is good for them.

The only thing that needs to be adapted is the association bonus. My proposal is to keep the reputation earned on that new exchange localized, i.e. do not count the reputation there for the bonus. Only if after some time (say at least a year) the general sentiment is that the experiment is successful and the amount of spammers or sock puppets is under control, consider including that exchange in the association.

The start of new exchanges is typically always a good opportunity to gain lots of reputation, there are lots of low hanging fruit, less competition and the rules aren't yet that strict. The focus is first more on quantity and only later shifts to quality. In a way that's wanted in order to grow and lowering thresholds could help even more with that. Let's see this as an experiment. It might end well or it might result in chaos. As long as the company turns off the association bonus for that site (it really should do it!), we could safely lean back and observe what's going on.

6
  • 2
    Interesting. What I’ve observed only haphazardly on PLDI is that, even in Beta, the focus is on quality. Commented Jul 23, 2023 at 21:11
  • 9
    What about chat? You only need two ill-gotten votes from GenAI to be able to troll almost the entire network's chat system (I've seen it happen several times this year, even without the software making it trivial).
    – Laurel
    Commented Jul 23, 2023 at 21:12
  • @Laurel I don't use chat much, so I don't have much experience with trolling there. One could insulate the rest of the network chat system too from votes obtained on that site by not counting then toward that privilege. Commented Jul 24, 2023 at 6:04
  • 1
    @D.BenKnoble That was my impression with a few betas. But let me add that a focus on quantity doesn't mean that quality is really bad. Only that over time more things get closed or downvoted that were still accepted in the beginning. PLDI did it differently? How did they develop and grow? Commented Jul 24, 2023 at 6:07
  • @NoDataDumpNoContribution I'm not sure if/what PLDI did differently, and it's still early enough that maybe I've extrapolated too much from what's available. Your sandwiched idea ("a focus on quantity doesn't mean that quality is really bad") is well-received :) Perhaps I reacted a bit harshly in assuming you meant to draw a stronger contrast than you did. Commented Jul 24, 2023 at 13:28
  • PLDI went through a normal, fairly boring early beta didn't it- the process is kind of meant to handle growing pains like scope and moderation. While the infrastructure is creaky, it's a well tested system for standing up new sites Commented Jul 24, 2023 at 16:03
4

*Puts pitchfork down...*

According to SpencerG, in the Gen AI site meta post mentioned by OP, there is no plan to "dramatically lower certain privileges to 1 reputation."

To be clear, the plan is currently to launch with typical, public beta rep thresholds. This post is intended to discuss if anything should be changed given its unique creation process.

As per the public beta rep thresholds listed at What are the reputation requirements for privileges on sites, and how do they differ per site? — up/downvoting and commenting everywhere will require more than 1 rep.

7
  • 15
    When SE says they "plan" something like this it means it's done deal. The request for feedback is empty gesture. Commented Jul 24, 2023 at 10:30
  • 4
    I wish I could be optimistic enough to believe that the 1-rep thresholds will be restricted to the private beta. That already has potential for a lot of damage. Sadly, I think this 1-rep threshold on the genAI private beta is harbinger of lowering rep thresholds in more places. Commented Jul 24, 2023 at 11:22
  • 2
    Have you looked in genai.stackexchange.com/help/privileges? :)
    – Thinkr
    Commented Jul 24, 2023 at 11:35
  • 5
    @Thinkr Seems like the typical private beta rep thresholds, unless I’m missing something. The site is still pending its public beta launch. Commented Jul 24, 2023 at 12:52
  • @galacticninja You say that "up/downvoting and commenting everywhere will require more than 1 rep". But that's not what the link I gave says.
    – Thinkr
    Commented Jul 25, 2023 at 6:48
  • 1
    @Thinkr By "will", I mean in the future. The site is currently in private beta, hence the typical private beta rep thresholds. Commented Jul 25, 2023 at 7:03
  • Oh. I got your point.
    – Thinkr
    Commented Jul 25, 2023 at 7:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .