1
$\begingroup$

AI StackExchange is still a beta site, still has growing to do, and appears to still be relatively small/niche. I have gotten the impression that it is more vulnerable than huge sites (like Stackoverflow) to people trying to game the system (for example by running multiple accounts themselves). This increased vulnerability appears to be due to:

  1. More difficult to detect. Due to the small nature of the site, it's quite natural that you see the same accounts interacting in multiple questions if they just happen to be very active users. On a huge site like Stackoverflow, it would be a much larger coincidence if the same two accounts kept repeatedly interacting with each other.

  2. More negative impact. If two accounts (operated by the same person or by close friends for example) are consistently upvoting each other and double-downvoting "competing" answers, that has a huge impact on a small site like AI StackExchange where there generally are only a handful of votes to begin with. On a larger site like StackOverflow, such votes would more easily get drowned out by "real" votes.


My question: do others agree that AI StackExchange may indeed be extra vulnerable, and if so, should additional actions be taken?


There is indeed a concrete case of two rather active accounts on the AI StackExchange from which I've seen so much suspicious stuff happening that I can't believe it's a coincidence anymore as described in my first point above (and googling for their names actually lead to additional off-site evidence that they're both the same person as well). Because there is really only a rather small number of new questions on the site every day, this has lead to a tangible negative impact on the site in my opinion.

I have already reported this case to StackExchange using the Contact Us link, as should be done in general on StackExchange. Therefore, I'm not reproducing the evidence here / calling the accounts out by name, but I can of course still do that in an edit if that would be deemed useful. Nevertheless, I have no idea how long it's going to take for action to be taken in response to my report, and am curious if anything else should be done to speed things up on a small site like AI StackExchange since this behaviour appears to have a disproportionally large negative impact on the site in my experience?

$\endgroup$
  • 1
    $\begingroup$ The system will clean up the mess. Just hold on as they accumulate. $\endgroup$ – quintumnia Jul 21 '18 at 16:38
  • $\begingroup$ I notified Jeff Atwood,about the issues our community is experiencing. $\endgroup$ – quintumnia Jul 22 '18 at 18:09
2
$\begingroup$

I do agree that the impact of shenanigans on a small site is greater because of the smaller size of our rep economy and the lower beta privilege thresholds. This is partially balanced out by there being fewer users to keep track of, but as you said, that can also make it difficult to tell organic interaction apart from problematic behavior.

The "contact us" link goes to Stack Exchange the company; moderators never see those requests. I imagine the Stack Exchange community management team is pretty busy (they oversee all 174 SE sites), so it might take a while for them to get to your request. In many cases, though, site moderators can do the job. In a sense, this site can respond with more agility than larger ones because the mod workload is far lighter.

To reach a moderator about suspicious patterns, cast a custom flag on any post. If you need more space than the flag box affords, I can create a private chatroom for you to share your findings with the mods. Thank you for your vigilance.

$\endgroup$
  • $\begingroup$ I notified Jeff Atwood,about the issues our community is experiencing...i think by next month, everything will be sorted out. $\endgroup$ – quintumnia Jul 22 '18 at 18:10
2
$\begingroup$

Vulnerability

The primary vulnerability of a publication systems with a large global readership runs in another direction than the one mentioned.

Members can post information in a way that appears to be well grounded in theoretical proofs or empirical study when the information may in fact be invented to sound good. The information may be authentically reproduced from the results of a web search and that information may too sound good but lack rigor or peer review.

The classic example of this later case is Wikipedia. The oversight of that site is well aware of the fact that many Wikipedia pages and sections lack peer review or reflect popular trends that have not been properly tested or very deeply considered. Most universities have departmental policies to not allow sources that are essentially blogs in theses.

Another example that I've seen in this AI beta and the SE sub-sites is the referencing of a sentence from the abstract or conclusion of a paper accompanied by an interpretation that conflicts with the facts in the body of the paper. I'm aware that people want to support their views and do searches, and I've done that on occasion too. However, we all must be aware that the quality of the site is highest when we all work to investigate the reliability of what we are quoting or referencing.

People are vulnerable to misinformation, even if accidental, and this is an information disseminating community. That's the concern of greatest importance, but that vulnerability is not a vulnerability of the SE system. It is a vulnerability of the readership.

Example of Negative Global Impact

The SO site and associated SE sub-sites are powerful. For instance, some semantic impressions have formed in SO that design questions are too broad. That may have been done for reasons that seemed wise, but the result is that SO has a corresponding impact on the global development community: Dismissal of design. Again, a vulnerability of the readership.

Size and the Influence of an Individual or Small Group

Regarding the correlation between size and the impact of activity of an individual or those who work together in a team in a workplace or lab, yes that is true. The larger the group, the lesser the relative impact of one or a few people. For that very reason, some prefer smaller groups because they feel more relevant, which is why the sub-sites are growing.

However, the correlation flips when considering detection.

It is actually more difficult to detect especially large impact from one or a small group in the larger group. A few people who just happen to have similar backgrounds or have read a similar set of material for work related reasons will stand out more in smaller communities than in larger ones.

The question has that aspect backward.

Defining Positive Impact

Whether that impact is positive or negative is reflected (and should be reflected) in the up votes or down votes, not the opinions of those that may notice what appears to them like excessive corroboration. When scrutinizing game-play, the developers of the system will naturally examine questions about trends. Regarding individual accounts or groups that work or live together, the question should be ...

  • Are there any votes that appeared that are without merit?
  • Were there any published member rules broken?
  • Are members collaborating to trounce on others?
  • Are members trolling to crush other individual members or ganging up on them to remove what they imagine to be a threat?

Those are the questions I have too.

Game Play

The question mentions gaming, where in fact all SO and SE sub-site members are in game play. The system is deliberately set up as a game. It is a zero sum game in that the point scores are relative as in some sports, therefore the increased success of one member may threaten the success of another. The game system is not the design of the member.

What is the choice of the member is intent. Some are here to compete; others to contribute.

That privileges must be earned is the more fixed point, and less relative, so there is some communal value to those who have decades of experience and tend to apply mathematical rigor and interest in social responsibility to questions to game out sufficient point score to positively impact the larger parameters of game-play. That too is how the system is set up.

The best leaders in this community (as with the larger community of humanity) are the ones that are willing to engage in the silliness of game play without particular interest in winning but to game well enough to get more factual and productive information out to people so they can make wiser decisions.

Actions to be Taken

The actions to be taken are the rules of the game. Of course, as in tennis, appealing to the umpire is permitted, but not always productive, especially if the umpire tips the scales against a contributor as a result of the appeal in such a way as to suppress or censor useful perspectives that contain needed qualities.

Censorship

Mathematical rigor is not always appreciated by those who don't value it, yet it is objectively useful, especially in science and technology. Those with decades of experience are not always appreciated by those that cannot speak from experience as easily. Because this is natural, those with more experience should respect the need of novices to gain experience gradually as they once did. So the experienced and the novice should yield to one another.

Minority views are by their very nature unappreciated. When they are repeatedly expressed with vigor and they have a disruptive quality or challenge the status quo, the reaction can be a mob scene. The mob is usually made up those that have used the status quo or trendy ideas to gain reputation.

Of course, SE AI members are above that.

Multiple Accounts in One Household, Workplace, or Laboratory

Although it is true that in communities, local ones, FaceBook groups, and SO ones, there are people who know each other and work together so that they have similar patterns of speech when talking on the topics in which they collaborate for family life, activism, or work. It does not mean they are cheating in some way. To those in the group working together, what they do is collaborate, which is no threat to an SO or SE site because they have thrived with couples, work teams, and many other collaborating groups interacting.

Truth be told, those who appear in public to always be in agreement may not always be in agreement privately. Families, laboratory teams, and those who work closely in the industry have just learned to withhold argumentation with their close friends, family, and work and laboratory associates.

Summary

In summary, others may have legitimacy or malice in their intent, but score based systems are technically games, and we all play together. The power and importance of AI is substantial and some will enter game-play reluctantly solely for higher purposes than points.

  • Infusing mathematical rigor where lacking
  • Encouraging backing facts with proper academic references and attribution
  • Encouraging proper analysis of papers and articles
  • Exposing the misuse of terminology
  • Scrutinizing practice that, while common, may be detrimental
  • Questioning blind futurism when it may realistically lead to mayhem
  • Asking any of the larger questions that should be asked
$\endgroup$
  • 1
    $\begingroup$ On the question "were there any published member rules broken?", it does look that way to me. Having multiple accounts is allowed, as long as you don't do anything you couldn't also do on a single account. That means rules are broken as soon as someone with multiple accounts upvotes their own posts, or double-downvotes other posts, or posts comments from accounts with different names on a single Q+As page. $\endgroup$ – Dennis Soemers Jul 29 '18 at 18:18
  • $\begingroup$ Such actions are harmful because they can affect how others (especially non-experts, who are the ones that will look to this site to learn) perceive the quality of certain answers. "This one got instantly upvoted? It must be good then, let's also upvote it!" obviously that's not what should happen, but it does. This effect is much more pronounced on a small, niche site like this one. $\endgroup$ – Dennis Soemers Jul 29 '18 at 18:19
0
$\begingroup$

The subtext of the question was, that any kind of content moderation in a forum distorts the normal user interaction. According to this assumption the ideal forum doesn't contain any kind of moderators under fake accounts. Let us describe what the purpose of content moderation is. The aim is to delegate tasks into the workgroup. It is the same principle used in companies for improving group working. The main purpose of content moderation is a service for the group. That means, management doesn't hurt the social community it makes it stronger.

AI StackExchange doesn't have to much people who are trying to gaming the system and influence the group but too little. The result of missing moderation is, that the group is not attractive for the environment and that the group members are not communicating efficient.

$\endgroup$

You must log in to answer this question.

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