I don't have any formal education in statistics, so I can't provide what you're asking for (that I know of).

That's cool, I guess I can't get my info then. I would suggest that you look at a well planned z-test as being your goal in this, otherwise you're just pumping numbers at people, and they are easily shot down.

Thanks. I tend to learn things as I need them, so knowing I'll need this definitely helps if I come to do something like this.

Would a z-test work?

AFAIK , you need a normally distributed mean value and compare it with a given constant. In this case you are looking at when a user(s) are online and when down rep happens, so there will be no normally distributed mean value (either a person is or is not online, kind of like a Bernoulli trial). But that will not work, for if the person is not online then he can not be the downrepper... The mean you will get is either 0 or 1 (is on, is not on) and the constant you will compare it with is 1 (is on). The only thing I can see would work with a z-test is the number of total users online when a massive downrep happens, and compare this with the number of total users when it is not, but I am not sure what that would prove (or is it this you are looking for?)

Is it not better to do an associative / correlational test? Or am I mixing up things here.

That would work fine, in fact that's my example point. There are users online, and there are intersections in their being online. Xyem appears to be arguing that these intersections are heavily associated with the user doing the rep stuff. A z-test seems obvious given the huge amount of sample data. The test on the otherhand seems hard, and this is where you've cut it to binary. I'm suggesting that the criteria for evaluating that test is far more complicated, i.e. were they involved in the thread, have they posted in the past, how often do they speak about downrepping - I don't know.

So yes, a z-test, but with more than just is "this person blue".

If you are referring to the timestamp request I made, I was looking at how closely that user's activity in that specific thread related to the rep losses that happened in that thread. It wasn't just that the user was online, it was that they were definitely looking at that specific thread at the time the rep was lost (given the 15 minute precision from RepLog anyway).

Given all the data that GOG has access to, the test would include things like "did they only interact with the site to downrate a post".

To be honest, this would likely be a multi-stage thing. The first thing would be to detect the abusers "tool" accounts (the means). This is easy given things like the test above. The second stage would be identifying who the "parent" account is (where you may need to go into useragent matching and the like). The last stage could be the "human" check e.g. did they argue with the target? (the motive).

Rep me down but i have the tendency, strong tendency if not a habit to down rep folks complaining too much about DRM Free or not stuff, saying screw this it is Steam, Origin, etc.. Their loss, deal with it stop voicing it every fucking bundle time like you have something to prove or here i may and will rep you down hehe.. I just got fed up with this attitude on GOG, so here goes my ugly repping! Useless but Fair retribution to those showing up feeling superior not using Clients, there you have it! :O

Following this topic, I went ahead and asked them if they could provide a reason for not providing the timestamps. So, at least on the face of it (and I have no substantiated reasons to think otherwise), it wasn't a refusal to provide it but a matter of access (at least by the support team), which makes sense.

I need to have a think if it is worth pursuing this by asking for the request be passed on to someone who can (i.e. the web dev team).

I guess you could always ask Little Bobby Tables and see if that helps ;)

Hm.. it's actually quite possible that there are injection vulnerabilities. We've seen lack of escaping before..

.. if I found one, it'd be so tempting to add the removed games I don't have to my account, buy the other games I'm missing and truly complete my GOG collection! Though I could just add them, update my GOG Wiki page to show I owned them, then remove them again..

Possibilities :)

There are indicators that the person I believed was abusing the rating system before has begun taking the same action in this thread, so I went ahead and asked the GOG support staff member I have been talking to, to pass on my request to whoever maintains the site (probably the web developer team).

Hopefully they will provide the data and put my suspicions at ease.

Le posted, I shall keep you.

Sorry dude, people disagreeing with you is not abuse.

Apparently, outright telling you I wasn't going to talk to you isn't enough to get you to leave me alone. Now you're misrepresenting me.

I never said people disagreeing with me is abuse and such a belief is clearly one I do not hold.

I have been in many disagreements, some even very heated, over my 3 years here and not only have I never claimed abuse, I've never even lost the rep to even give me cause to... until I disagreed with you.

In fact, me losing rep is so recent, Google literally has no record of me having any low-rated posts. Seriously.. guess what I was looking for when I found out about the attack on Ubivis.

Indeed, of the 6 results for "xyem" and "low rated", they are all where I have posted in a topic where someone else has been low-rated (and not a single one of them is my opponent in a disagreement, by the way).

So, pray tell.

How does 1 person (who says they have no friends) enter an argument against 3 or 4 other people (at least one of which is "popular" enough for anyone even remotely helping them or on their side to be immediately be accused of only doing so to [url=http://www.gog.com/forum/general/replog/post385]favour) and yet.. it is the 3 or 4 people who end up with low-rated posts but "Billy No Mates" does not.

Even if that one person was 100% absolutely, irrefutably correct, it does not necessarily follow that they would get fewer low-rated posts than their 3-4 opponents. If all the participants downrated all their opponents posts, it would only take 1 or 2 to send the one persons posts low-rated while it would take another 4 for their opponents.

So you are basically saying that my position was so wrong that even though you had zero vocal supporters, you had more than 4 other people agree with you, whereas even though I had at least 3 vocal supporters, not a single other person disagreed with you.

I think this to be pretty unlikely (mainly because I was getting supportive PMs from a few people who didn't want to become targets..). Something was going on.

Sure, it may not be you, but I am clearly ruffling someone's feathers and you are currently my prime suspect.

During that entire argument with you in RepLog, the phrase "Thou doth protest too much" was a prominent thought. Especially so when you continued to attack me after I disengaged you in debate (like you have just done here).

You also seemed adamant that it would take hours to create 100 accounts, when it only takes about 30 minutes. It's almost as though you have made multiple accounts and it took you a long time.. otherwise, why not believe me when I said it took me ~5 minutes for 9 accounts (for the rating testing)?

It all comes across like whoever is abusing the rep system is trying to dissuade me from investigating, so perhaps instead of obstructing me and making yourself look it is you, you should try to help?

Ok now I think we have entered Insanestan, your giant ego can not comprehend that it is possible for random readers here to downrated your logically false or paranoid posts? A bit narcissistic, are we? Instead of just admitting to yourself (not even to me or the forum) that you might be wrong, you just blame the other person of multiple accounts abuse?

As with any conspiracy theory, this is easily debunked: I am way too lazy to spend even 10 minutes to create accounts just to affect some pointless rep.

But yes, now I sincerely hope that gog will reply to you, just so you realise no one is out to get you and people just disagreed, even though I am a lowly nobody here and you are the public darling.

Although I guess it is possible someone else is really out to get you, that would be a bit of an obummer for me, I guess, because my ego grew a fair bit during that other thread. :D


Edit: 30 minutes for 100 accounts??? Maybe for a pro like you, it took me like 5 minutes to create even my own account.

This is really starting to feel like one of those movies.
You know, one of those movies that seems to start out with a clear antagonist and protagonist but the well-meaning character eventually devolves into a less sympathetic character? Then the two less-than-sympathetic characters hash it out until they've both destroyed themselves?

I like those movies.