Cheating/A Business Solution

Cheating/A Business Solution



1. Reason & Motivation for This Thread: Reason is that I empathize with the business challenges facing Russ and Chris. There are four...

RHP Site Integrity; RHP Member Morale; RHP Potential Litigation Risk; and RHP Reputation and Goodwill Within the Internet Community.

Motivation is personal. Recently withdrew from participation in the RHP General Forum (Even at High Tide, Clarence, Friendly Advisory threads

on pages 1-2) and, I guess, the energy had to find some alternative outlet. So I've been reading interesting threads in other public forums.


2. Qualifications to Comment: None, from a technical standpoint. Just a crusty senior dinosaur playing with a laptop gift, trying to figure out

cut & paste and on the internet for only the past twelve months. On the other hand I do bring decades of progressive career success with

several Fortune 100 and 500 Companies to the RHP table. Durable organizations have leadership at the helm which insists on seeing things

objectively as they are and are likely to become, in context of present goals and market trends. They are also creative, nimble and decisive.


3. Present RHP Situation: Private suspicions unfounded accusations, public ill will, voluminous posts on the subject, negative internet press.


4. Simple Suggestion: Eliminate the cloak and dagger of moderation team identity and the bandaid reporting of suspect individual games.

How? By taking an open book approach. No idea how data base tools or detection equipment work but if it is available and if it does work...

then apply it in a routine manner to all active chess players. How? By publishing the findings next to the rating on each member's profile.


5. Implementation Details: Three index numbers come into play: (1) Player's rating expressed as a percent of maximum achievable (let's

say a base of either 2500-2600). Top ten players average rating is now 2366, which yields 94.6 on 2500 and 91.0% on 2600. (2) Data

base or engine equipment match percent. (3) The signed difference between those two percents. Rating - match percent = + would seem

favorable and a minus difference would appear questionable. Just publish the raw index information (for the most recent 50-100 games)

for all players... STARTING AT THE TOP, then work down to the 1500-1600 level. Update it monthly. There's probably a logarithmic rather

than a straight line correlation between rating and match. That's okay. Refinements of interpretation will follow. Meanwhile there's an open

book policy approach and cheating deterrent in place, which treats all RHP Members equally and self documents policy decisions to ban.



-gb

Originally posted by Grampy Bobby
Cheating/A Business Solution



1. Reason & Motivation for This Thread: Reason is that I empathize with the business challenges facing Russ and Chris. There are four...

RHP Site Integrity; RHP Member Morale; RHP Potential Litigation Risk; and RHP Reputation and Goodwill Within the Internet Community.

Motivation is personal. Recently withdre ...[text shortened]... ch treats all RHP Members equally and self documents policy decisions to ban.



-gb

I already asked Russ for this open view of 'matches' and it's not the direction he wanted to got.

What do you do with the info?

Who decides if a matching % is too high?

Cheats will watch their percentages and keep them to an acceptable number.

Everyone would have a different idea what percentages are acceptable.

So, it's not a new idea... not exactly a bad idea... but not the direction Russ decided to go even after 2 game mod explosions.

P-

Originally posted by Phlabibit
I already asked Russ for this open view of 'matches' and it's not the direction he wanted to got.

What do you do with the info?

Who decides if a matching % is too high?

Cheats will watch their percentages and keep them to an acceptable number.

Everyone would have a different idea what percentages are acceptable.

So, it's not a new idea ...[text shortened]... ad idea... but not the direction Russ decided to go even after 2 game mod explosions.

P-

Please read the suggestion again. Nobody would be looking at one %, rather at the comparative index of rating versus match percentages.

Initially, little if any action would be justified or taken, since we'd be exploring virgin data territory. Within a few months, however, a profile

will emerge... most probably a normal distribution Bell Shaped Curve (bi-modal would be within the realm of possibility). At minus three

sigma (the small clustering of match percents significantly exceeding rating percents) eyebrows will be raised. Russ and Chris will decide.



-gb

Originally posted by Grampy Bobby
Please read the suggestion again. Nobody would be looking at one %, rather at the comparative index of rating versus match percentages.

Initially, little if any action would be justified or taken, since we'd be exploring virgin data territory. Within a few months, however, a profile

will emerge... most probably a normal distribution Bell Shaped C ...[text shortened]... tly exceeding rating percents) eyebrows will be raised. Russ and Chris will decide.



-gb

The idea alone seems extremely logical. This would probably lead to tightening the noose around cheaters. However, to Russ is it cost effective? The amount of hardware/programming/upkeep on a system which automatically checks moves would be very large. While your indexing solution is favorable, the issue is WHAT to use to get your match %.

It would take multiple machines running endlessly to gain the data needed to supply such a system. I don't see that as lucrative in this case... Our subscription $ would triple.

That said, I'm subscribing before the end of the day in case Russ buys into this. 🙂

Originally posted by Nowakowski
The idea alone seems extremely logical. This would probably lead to tightening the noose around cheaters. However, to Russ is it cost effective? The amount of hardware/programming/upkeep on a system which automatically checks moves would be very large. While your indexing solution is favorable, the issue is WHAT to use to get your match %.

It wou iple.

That said, I'm subscribing before the end of the day in case Russ buys into this. 🙂

Agree. That's why I qualifed with 'IF such equipment is available and IF it works'. Totally a black box approach and I have no idea what's

in the box or what the box might cost. One other STRONG benefit of the Open Book Approach is that each RHP Member would be provided

with an ongoing protective shield. Today, most typically look at invite parameters or a prospective opponent's rating before accepting

a game. Suppose each player's profile also displayed rating percent, matching percent and the +/- difference. Wouldn't that disclosure

inhibit many equally rated or higher rated players from accepting games from players with a match percent dispproportionately higher

than their rating percent? So the business question you need to ask is "Who really cares if some criminally minded players are using chess

steroids prohibited by tos, with intent to cheat, if they're unable to hurt anyone because no thinking player will give them the time of day".



-gb



Edit: Probably no real need to run the data game by game live. After the fact EOM processing of last 50-60 games completed would be fine.

Originally posted by Nowakowski
The idea alone seems extremely logical. This would probably lead to tightening the noose around cheaters. However, to Russ is it cost effective? The amount of hardware/programming/upkeep on a system which automatically checks moves would be very large. While your indexing solution is favorable, the issue is WHAT to use to get your match %.

It wou iple.

That said, I'm subscribing before the end of the day in case Russ buys into this. 🙂

One last clarification. Always nice to conclude on a bright note. Player Tables indicate an RHP total of 19,280 active/inactive players.

Applying Pareto's Curve, we know with some reasonableness that, in terms of rating categories, 70-80% of all alleged cheating can

be attributed to a relatively small 10-15% of all members. Only 191 are rated above 2000; 629 above 1800; 1230 above 1700. Start

there in a rich target zone, then swoop down to 1600, then 1500. Below that level who really cares. Why? Any chess steroid users with

lower ratings will sooner or later become emboldened. Their ratings will climb. Their graphs will spike. They, too, will qualify for scrutiny.



-gb

😉

Originally posted by Grampy Bobby
One last clarification. Always nice to conclude on a bright note. Player Tables indicate an RHP total of 19,280 active/inactive players.

Applying Pareto's Curve, we know with some reasonableness that, in terms of rating categories, 70-80% of all alleged cheating can

be attributed to a relatively small 10-15% of all members. Only 191 are rated abov ...[text shortened]... gs will climb. Their graphs will spike. They, too, will qualify for scrutiny.



-gb

😉

Indeed. This is why I believe in the point approach. I think you have the right idea, however I think you may have made it slightly more complicated than need be.

The fact is, you can achieve the same result by utilizing feedback instead. If you were to set up a feedback system for cheating and multiply it across rating for a point value, you would know whether to challenge a player as well. Say a player rated 2000 has played 600 games, and of those 600, 200 submissions of cheating have been submitted (thats frighteningly high, but a good round example). This would be 200 points, across a multiplier (say .05xrating). So [ 200 x .05(2000) ]/ 100 in this case, you would get exactly 200 points. If you saw a player with 200 points, you'd know that he's extremely suspicious. The only thing I think needs to be added, and I'm not sure HOW is the addition of the amount of games played per point submitted.

If someone can create a function which will reflect the amount of games played and feedback submitted in a feasable way, then I think it needs submission.

My only idea for this has been:

{ [T/(F1)]x[ F1 x .05(R) ] }/ 100

Where T is total games played, F1 is feedback points and R is rating. The division of 100 is simply to put the points in a lower more understandable perspective. This expression however I don't think is very representative of the danger of a player. If anyone knows something better, please submit it.

Originally posted by Nowakowski
Indeed. This is why I believe in the point approach. I think you have the right idea, however I think you may have made it slightly more complicated than need be.

The fact is, you can achieve the same result by utilizing feedback instead. If you were to set up a feedback system for cheating and multiply it across rating for a point value, you would presentative of the danger of a player. If anyone knows something better, please submit it.

Your math is well beyond my reach. Appreciate your desire to quantify, while entertaining concern about the subjectivity of your data source.

I'm not qualified statistically for this discussion, but would like to make a comment about subjectivity.

Player A tags Player B as a cheater. If Players C, D, E, F, G ... Z do the same, the element of subjectivity should be mostly dealt with by the sample size. If each player placing a tag is required to complete at least 30 moves in a game against the suspect, the possibility of an orchestrated gang-attack should be significantly reduced.

Originally posted by MissOleum
I'm not qualified statistically for this discussion, but would like to make a comment about subjectivity.

Player A tags Player B as a cheater. If Players C, D, E, F, G ... Z do the same, the element of subjectivity should be mostly dealt with by the sample size. If each player placing a tag is required to complete at least 30 moves in a game against the suspect, the possibility of an orchestrated gang-attack should be significantly reduced.

Agreed, this is an oversight in my previous post. Moves made would also have to be dealt with. Perhaps the "Feedback" or "Ticket" option would not be available untill after an agreeable amount of moves in a game. I would think that after 12-16 moves this would probably be appropriate.

Originally posted by Nowakowski
Agreed, this is an oversight in my previous post. Moves made would also have to be dealt with. Perhaps the "Feedback" or "Ticket" option would not be available untill after an agreeable amount of moves in a game. I would think that after 12-16 moves this would probably be appropriate.

Representative game volume sampling (last 30-50 games, rather than one specific game in progress), an objective standard (player's

rating strength relative to black box matching), targeted application of scrutiny to 1500 or 1600 and above, simplicity of site administration

and cost effectiveness relative to benefit still seem primary factors. You do see math solutions. Do you recognize non-formula variables?


-gb

RHP site tone and membership morale will remain crux considerations.

post by grumpy bobby

"3. Present RHP Situation: Private suspicions unfounded accusations,
public ill will, voluminous posts on the subject, negative internet press. "

Oh how right you are.

I came here to post another idea and read this thread.

Excellent idea to have this out in the open.
The other poster is pointing out all the difficulties but 10 programs
permanetly running through the system picking games at random
checks two players at the same time.

If clean the program moves on and green flag (another idea from another thread)
is placed against the name.

If it finds a 'suspect' the name is passed to MOD Team.

As for cost. Russ can approach makers of say Rybka and if they
give him 10 programs then he will advertsie Rybka on the site.

There are few things to iron out and remember I'm here on another matter
so I have not thought all this out. But it's on the correct path.

AVOIDING THE POTENTIAL RISK OF COSTLY DEFAMATION LAWSUITS/A COMMONSENSE APPROACH



1. Premise: All lost chess games are lost for the same reason. All fragmented and messed up human lives are fragmented and messed up for the same reason. All floundering companies suffering self induced headaches flounder and take expensive aspirins for the same reason. Conclusion: That same reason has always been, still is and always will be... sloppy thinking.

2. Premise: All men are created equal. Conclusion: Despite the fact that this gentle rose colored myth has been around for a long time, nothing could be further from the truth. Genetic, cultural, social class, wealth and other privileges of birth (and time and place) always cause great disparity. All men enjoy equality only in the sight of the law.

3. Premise: Presumption of innocence is one of the pillars of western hemisphere law. Conclusion: Presumption of innocence is not only the starting point but, also, the mid-point and the concluding point within the astute mindset of human nature savvy, risk averse, well run companies (who prefer to stay out of court and protect their bottom lines).

4. Premise: Language matters. Conclusion: The RHP Site is leading with its chin to the extent that the C-word is permitted to be used.

5. Premise: The inflammatory words 'cheat', 'cheating' and 'cheater' convey a grossly presumptive, often inaccurate, rush to judgement bias and, therefore, represent an ill advised point of view in the public domain. Conclusion: These words should be stricken from all RHP contractual statements, including terms of service, and should be prohibited from casual use by any and all members in the public forums as well as in permanently archived in-game messages.

6. Premise: By definition, these words focus on the member, i.e., the person. Conclusion: The correct focus is the member's game history.

7. Premise: We may have reason for concern, speculation or even suspicion whenever play in a given series of games appears contradictory to a player's previously demonstrated overall strength profile. Conclusion: That's all we have. We do not have conclusive evidence, hard proof or any other legally sound basis for accusation.

8. Premise: Use of the words 'ban' and 'banned' pose similar business risk. Conclusion: If and whenever a player's machine matching percent substantially exceeds his or her rating percent (beyond the impartial and established tolerance limits as stipulated in written RHP Site Policy of record) the player is simply notified of his or her 'out of tolerance rating versus machine matching profile' with specific and objective reference to site policy. Membership may be allowed to continue to permit further fact finding and confirmation (some will voluntarily quit during this interim) and/or the membership is summarily and amicably concluded... without accusation or labelling of any kind. RHP assumes no burden of proof of member misconduct. A documented out of tolerance profile is the only issue. Prorated or full subscriber fee refund should probably also be promptly made.

9. Premise: You seldom go too very far wrong in this life or in a business venture by giving the benefit of doubt and treating others better than they deserve, Conclusion: No membership should ever be concluded on the basis of only a handful of games. In most instances, several dozen or more would likely insure a fair sample size and allow for random brilliance and the flukes of exceptional play.

10. Premise: The C-word is neither accurate, justified nor useful. Conclusion: Regret having used it as part of the title of this thread. Another example of 'sloppy thinking'.



-gb

😉

I do agree that the "c-word" is part of the problem. And I think Russ has that covered. People are now simply banned, with no mention as to why. Maybe Russ should switch from "banned" to "refunded" (which is the policy, I believe) to completely remove any implied stigma and keep the lawyers at bay.

Of course, that means no post-ban discussions of engine use could be allowed either. Right now, 3b'ers are pretty much fair game after they've been ejected.