The Dustbin

[stockwellpete]

Remaining unbeaten the entire season and with six victories to his name, Challenge1 and his Arab Conquest army have won automatic promotion from Early Middle Ages Division C. Well played Challenge1.

[MikeMarchant]

Great stuff. A well deserved title, Iaian.


Best Wishes

Mike

[Nosy_Rat]

Thanks, guys.

[devoncop]

Really well played. Totally dominant all season.

Good luck next season !

Ian

[snooky51]

Division E

snooky51 (Ghaznavid 962-1187 AD) defeats stormcrow (Viking 900-1049 AD) -- 60-45

Vikings came out with an all infantry army, mostly dismounted huscarls on their left, mostly offensive shield wall on their right and their regular huscarls in the middle. Ghaznavids threw the weight of their army at the Viking right and center, a holding action by what was left over in the Ghaznavid army kept the dismounted huscarls from getting engaged until it was too late. Ghaznavids had 58% routed going into the final turn and was able to rout an archer unit to grab the win from the jaws of draw. Good game

[Ludendorf]

Well done on the victory, and good luck in B division next season.

[DanZanzibar]

Division F

DanZanzibar (Carthaginian 235-146 BC with Spanish 300-10 BC allies) deafeats Blagrot (Samnite 355-272 BC with Italian Hill Tribes 490-275 BC allies) 42-14

Thanks for the match - I'm looking forward to our next one!

[Cunningcairn]

Well played!

[kraff]

I think the rule of zero tolerance for cheating is the right way. The problem is with determining of what constitutes cheating.

In my opinion it is a strange idea to suggest that even one reload should be considered cheating. For the sake of argument, my opponent and I can agree, for whatever reason known to us, a reason that nobody should have any right to object to, to reload our turns as many times as we please when playing a tournament match. This is a purely theoretical scenario, edge case scenario if you will, but as long as this reloading of turns is consensual it is hardly cheating. This scenario may be more probable if my friend and I decide to have a private match and, again, reload turns as we please, in order to learn more about the game's mechanics and to play better.

As for determining Internet connection problems - if everything has been set up correctly on the server side, this should be logged in server's logs. It should be evident from the logs whether there were any lost connections between the server and a player's computer during turn upload/download. If there was no problem with connection, then maybe a player is up to something and we proceed with verification. If there were problems, nobody is cheating. To put it simply.

[lydianed]

Division B

lydianed (Moors + Byzantines) defeated paulmcneil (Parthians + Seleucids) 60:36

[SpeedyCM]

Congrats mate well played.

[SpeedyCM]

Division C

SpeedyCM (Median 626-550 BC) defeats Barrold713 (Syracusan 550-461 BC with Libyan 549-301 BC allies) 53-24.

[stockwellpete]

Division B (Missing Result)

sunnyboy (Syracusan) beat edb1815 (Egyptian) 40-2

[phoyle3290]

Division C

phoyle3290 – Carthaginian, Hannibal in Italy, 216-203 BC with Bruttian or Lucanian allies defeats devoncop - Etruscan 330 – 280 BC with Samnite 355-272 BC allies - 64-50

My opponent made an excellent effort to rightfully earn that final point. It was an interesting matchup from the get-go, but some good luck just pushed the match in my favor.

Great playing and thanks for the game!

(3-1)

[MikeC_81]

We're currently collecting data on all games to find out what "normal" behaviour is to be able to set the thresholds that will prevent cheating . . .

Another disappointing response. Let me reiterate that "normal" behaviour is 0 reloads. ZERO. The threshold you are setting is not for "normal" behaviour. It is the tripwire for the amount of "abnormal" behaviour at which an investigation must be launched . . .

This is where I disagree, Mike. At the moment, apparently the system flags up the 10 worst cases of excess downloads whenever it is checked (how often it is checked I don't know). In a previous post I asked whether this was likely to catch all those abusing the system. What about those players number 11, 12 and 13 on the list? Were they shown as having zero excess downloads? Or what? I don't suppose we even know. So when Iain says Slitherine is collecting data on all games to find out what "normal behaviour" is, I read it as him wanting to establish how many cases should be flagged up each time in future. Should it still be 10, or maybe 15 or 20? It is no good warning the worst 10 players if the next 10 players are also save scumming as well, albeit at a more judicious level. So maybe Slitherine's "net" will have to be a bit bigger in future.

Without meaning to beat the dead horse, the issue isn't the size of the net it is that the holes in the net are too big so to speak in your analogy. They are so big that you could drive a truck through it before someone noticed. Anyone with a reasonable understanding of how computers, files, and servers work can easily understand how to cheat the system. The number of people they round up per report is irrelevant. They could dredge up the top 100 names if they want and it wouldn't help because as I keep saying, they don't have context. We have no idea what algorithm for this is although from Ian's comments one can reasonably infer that a player has an overall upload vs download ratio that must be maintained or it flags the system. This would explain Ian's comment that the system is better at catching chronic cheaters than sudden spikes since a spike style usage of save-scumming could retain a reasonable upload vs download ratio while savescumming in one game (a critical game where they are behind or even) while doing no save scumming in several other games (ie the ones where he is ahead or unimportant and thus has no need to savescum).

The problem right now is that since the details of the extent of dkalenda's cheating are now public, any reasonable person with said knowledge of how a computer can now target the automated checker's weaknesses with surgical precision. Want to cheat in the DL? Enter as many sections as possible, target the section you want to win, save scum that division within reason (ie not dkalenda's ridiculous levels), while playing all other sections without cheating to pump the system full of clean games to keep that upload vs download ratio below the level of scrutiny.

Ian doesn't want false positives, so the threshold must be set high so there is no way for his system to defeat a targetted attack on it like I just described. They simply don't have the manpower to see where the cheating is going on. On the other hand, a system of transparency where players are given reload notifications effectively outsources the manpower dilemma to the player base. We know from his results in past DLs that he doesn't win every division every time. So it stands to reason that he too understood that save-scumming needed to be used in measured doses. Yet it likely is the fact that as he kept strategically save scumming more and more, he realized that the tolerances for the automated checker were so high that he decided to go haywire and that's when he finally got caught. Even then it wasn't until he was dumb and sent an email admitting to it that it went public.

So let us say that another dkalenda shows up, understands how save scumming works, and understands the likely safeguards that are in place like I do but notifications are now in place. If he was to do a targetted attack on the system like I described, a player in Classical Antiquity might all of a sudden ask why dkalenda is reloading so often in Classical Antiquity along with abnormally lucky results in the saves that got reloaded. He might fire off a quick PM to you or other players in that division to see if he is reloading in those games as well. A bunch of quick affirmatives from 3 or 4 other players with confirmation with reloads and abnormal luck in those reloaded turns and bingo, we have reasonable cause to summon that player to provide a good reason.

Without player notification of reloads, we are in the dark. With player notifications, it shines a light and when enough light starts coalescing on a player, we ourselves can act as the policeman when Slitherine clearly doesn't have the manpower to do that. By all means, they should still tighten up their automated checker but without transparency, cheaters will continue to prosper especially now that we have so much more information on where the weakness of Slitherine's defences are.

Another disappointing response. Let me reiterate that "normal" behaviour is 0 reloads. ZERO. The threshold you are setting is not for "normal" behaviour. It is the tripwire for the amount of "abnormal" behaviour at which an investigation must be launched. And no one should be afraid of investigations. I have had to reload over the course of 2+ years of competitive play due to technical problems. Yet I am more than happy to have my upload vs download ratio public because it will prove that the the wins I have gotten are legitimate by the simple fact that you can count the number of reloads due to all problems on both hands.

But you make it sound like reloads should be a regular occurance, indeed "normal" behaviour. For a reload due to internet failure, the following has to happen. Their internet must be working to download the file. Then sometime in the next 10 minutes while they play their turn, their internet has to die preventing an upload and it must stay permanently down long enough for them to have to quit the game since FoG2 allows you to retry uploads. The other possiblility is that the game constantly crashes while being played. We don't see masses of people posting on tech support forums or emails saying the game is crashing do we? I mean if there are, you better tell Pip and RBS to fix it because they aren't doing their jobs.

Mike, how do you draw the conclusion that zero reloads is normal behaviour?

You would need to analyse the date in order to conclude that. If we lived in a perfect world where internet connections were 100% reliable, where computers never crashed, where storms never interfered with electronic equipment, where players never made stupid mistakes (it's not impossible to hit Alt-F4 by accident), and nothing else could possibly go wrong while playing FoG 2, then zero reloads might be normal behaviour. But we don't live in that world.

In fact, even after Slitherine have performed this analysis and established some baseline norm, deviation from that norm only flags up abnormal behaviour. This abnormal behaviour does not, by necessity, identify cheating. It means that Slitehrine have the very difficult job of making a decision about whether to level that accusation or not, in a case where, however abnormal the data might be, the player could still be innocent. After many years of interacting with the good people at Slitherine, I do not see them wanting to be hang 'em/flog 'em judges.

This is an extremely difficult problem to solve. It is not simply a technical challenge, it is also a judicial and social challenge. Navigating through all these complexities and coming up with a solution that works, that is fair and that is acceptable to the community is not a task I would want to take on. For my part, I am happy to accept whatever solution Slitherine proposes and move on. If I have to live with the possibility that one of my opponens is cheating (and I can't see how it is possible to ever rule that possibility out) then I have the choice of either accepting that gracefully or no longer playing multiplayer against opponents I don't know and trust. Persoanlly, I will take the former course of action.


Best Wishes

Mike

Zero reloads is normal behaviour because that is how the game works. What would you say to your internet provider if they told you that loss of server was normal? Are you in a place where lighting storms occur everyday knocking out electronics? Don't even get me started on Alt F4. That is a meme. No one hits the specific combination of Alt F4s for no reason. How many times does your kid run into your room to turn off the power to computer? These are "normal" events to you?

Of course, it is a difficult job. I said it from day one that it is an impossible job to police all cheating. I knew exactly how it was possible to cheat the system the very first time I saw the MP interface and finished the first upload/download cycle. You can go back two years to the bbogensic affair where I broke it all down. What I am saying is that Slitherine is doing such a bad job that they can't be trusted again so that transparency must now be filtered down to the players so we can police ourselves. I don't care if they never ban a single person again off their servers since as a player base, we can now have the information available to police ourselves..

[SnuggleBunnies]

I think the argument that having reloads displayed ingame would cause suspicion is a bit silly. Doing so gives the player a chance to explain what happened. It also gives the other player a chance to analyze whether or not there is a pattern, and if they want to choose not to play a certain player in the future. Sure, it could lead to false accusations etc. But guess what? Right now, players have no way to tell who is reloading, so ANYONE could be doing it! As things stand, you can be falsely accused without even having reloaded at all. How is that any better?

[Ludendorf]

So, question. Because right now, what we really need is as much information as possible. Just what is the highest frequency of downloads to uploads right now? A list of 40 were mentioned. What is the size of the ratio for individual 1, and what about for individual 40?

Furthermore, do we actually have any idea what the mean frequency is? A baseline is always good to have in these matters.

[Nijis]

How about make it so if there is a reload in a tournament, the reloaded-against player automatically wins that game? (Or maybe can claim a partial win, or a score adjustment, or is given the choice of restarting the match?) You don't need to worry about proving cheating beyond a reasonable doubt because this removes the incentive to cheat.

It's bad luck for the reloading player if their kid really did unplug the computer, or there's an internet outage, but very bad luck did happen to ancient commanders. Think of it as an eagle relieving itself on your general's head just as the sky darkens thanks to an eclipse, or something of that nature. But you just lose the one game, not a whole gaming community that you enjoyed being a part of.

MikeC_81:

Zero reloads is normal behaviour because that is how the game works. What would you say to your internet provider if they told you that loss of server was normal?

I've lived in places with very unreliable internet, and you couldn't say anything to your provider because they were the only game in town. Also, I assume that there are scenarios where Slitherine server outages could be to blame. RBS explained why my situation in the last game would not count as multiple downloads, but if a player tried once, failed, then quit and reloaded at a later date, would that not count as two downloads?

[pantherboy]

I'm in the camp of MikeC. Just allow instances of reloading to be included in the pop-up dialog at the start of your turn. Shift the onus onto the player to explain the reasoning. By having the info on whether a redownload occurred or not a player will be able to analyze the replay for the frequency of unusual event before beginning their turn. For example if it happens in the first turn or two then obviously it will have no impact on the game and lends weight to any excuse. But if it is at a critical juncture of the game and a number of outcomes go the way of the re-loader than it would not be unreasonable to be suspicious. In such a case I would support the position that a player whose opponent reloads at a point in the game that they feel may give an advantage will automatically be awarded a win if they request one. Yes it could of been due to one's daughter switching off the PC (happened once to me while watching youtube) but that is still no excuse for affording your opponent an advantage whether sort or not.

[Challenge1]

Thanks for the kind words and to all my opponents for the interesting and challenging matches! We all know Arab Conquest to be a strong army list which definitely helped a lot!

Ehm, Div B.... Norwich or Sheffield Utd!?

Iain