Suspicious Scores

[Mr_Loon]

Thought it might be a good idea to have a thread where people could draw attention to any suspicious scores on the C4A scores website.

I was going to draw attention to the rather large hiscore submitted recently for Donkey Kong (3X current world record!) but that seems to have been fixed.

There does appear to be something a little odd going on with Super Hexagon in terms of duplicate scores, click the show all button to see this, my guess is that your previous best score is submitted as a new score when a new session is started.

No big deal but thought it was worth a mention.

 

[pmprog]

Can we get Milkshake's SF2 removed? I believe it was scored by multiple "coins" (SF2 doesn't reset the score on a continue)

 

[Binky]

Pinned

 

[Franko]

Can we delete all scores in Frogger that are higher than 14,000?  Either that, or can you all spot me about 3,000 points?  Thank you very much for your consideration...

 

[notaz]

There does appear to be something a little odd going on with Super Hexagon in terms of duplicate scores, click the show all button to see this, my guess is that your previous best score is submitted as a new score when a new session is started.

The game is always only reporting the highest score it knows over it's Java API's. I don't really understand how skeezix's system handles that, sometimes it takes the dupes, sometimes it doesn't..

 

[Kumaki]

Can we delete all scores in Frogger that are higher than 14,000?  Either that, or can you all spot me about 3,000 points?  Thank you very much for your consideration...

LOL, nice try!

 

[skeezix]

There does appear to be something a little odd going on with Super Hexagon in terms of duplicate scores, click the show all button to see this, my guess is that your previous best score is submitted as a new score when a new session is started.

The game is always only reporting the highest score it knows over it's Java API's. I don't really understand how skeezix's system handles that, sometimes it takes the dupes, sometimes it doesn't..

The server has very limited 'dupe filtering', to the point it should be ignored For the earlier versions it was handy for the way arcade games were reported, but its not really useful the way things are done now; for indie titles it doesn't really kick it at all, anyway, hardly.

That all said .. if you submit the same score 50 times, they're all valid scores, so it accepts them.. why not? Due to how indie titles are done (they have to purposefully invoke the score submission tool), it is assumed they know what they are doing and aren't 'wasteful'; that said, most of the indie titles aren't being too careful.

example: If a game starts with a score of 100 every time, and then just publishes the score on exit, and the users all just exit/kill it, then it'd post 100 for everyone, all the time... which wouldn't strictly be right; but that said, anyone actually playing the game will show up higher in the scores, and eventualyl push off those 100's, and it'll be a self correcting problem (ie: its only a problem because C4A keeps the top 500 scores.)

To get out of that situation, titles should only push valid scores (scores that are earned). If the scoring mechanism tends towards certain levels everyone gets, theres not much we can do about it.

I could add some filtration to suppress duplicate scores (or duplicate consecutive scores?), but never struck me as something we need.

Consider.. if you score 10,000 in pacman every time (or whatever), thats a valid score.. on the arcade machine, you'd slowly take over the high score table, which is how c4a was modeled; in milkshakes c4a website, we filter out the redundant names just to clean it up (but you can toggle this as well, right?).

Naturally, I could be seeing this all wrong, I do not claim to be perfect or represent even 5% of the audience

jeff

 

[notaz]

Well reversing SuperHexagon to get the current score out would be a significant effort, so it's too "expensive" solution. It's not the same as almost automatically generating JNI interfaces from .dex file and implementing them for apkenv that has been done to get it going.

 

[b_o_b]

ASM Asmo 6797973 Ms Pacman seems a bit high. Please remove this score... uhmm.. and it is probably best to delete all scores from Asmo. Just to be sure

 

[Jumpman]

ASM Asmo 6797973 Ms Pacman seems a bit high. Please remove this score... uhmm.. and it is probably best to delete all scores from Asmo. Just to be sure

+1

Yeah, we can't take any chances

Chris

 

[Asmo]

WTF? That's not (even in jest) a score I submitted...

Seriously, unless there's a weird bug in there that can result in an impossible score submission (without even completing a game), somebody's having a laugh. Any logs that can identify any oddities with locale or anything?

All my scores are legit - but that one's not mine!

I did play a game of msPacman yesterday, but reset and quit it when I had to go do something else, so no score knowingly submitted at all - and there's never been a score in it's table THAT high to even get accidentally submitted!

Any explanations?

EDIT: Remember when something similar happened with DK:

http://boards.openpandora.org/index.php/topic/12127-release-compo4all-mame-ladderscoreboard-competition-for-classic-arcade-games/page-26#entry246085

What was the OTT score recorded on DK, anyone recall? Is there some correlation? I wonder if this was an instance of the same thing happening?

 

[Ziz]

Who believes this face?

 

[Asmo]

I should have worn a tie when that photo was taken.

 

[Jumpman]

I should have worn a tie when that photo was taken.

:lol:

Chris

 

[skeezix]

There could have been a glitch somewhere/somehow that incorrectly determined a score (despite havbing thousand of submissions okay, you never know..)

The one good thing, without heavy end to end encryption etc, is its hard to impersonate someone:

- don't share your unique-id c4a profile id; if you do, you're boned Its a big large number, so its very hard to guess (nigh impossibru!)

- it is possible to re-use someones initials, since I designed it originally like an arcade machine .. any number of people could be SKZ if they wanted.

---> if everyone desires it, I can force unique-ness on the initials as well .. that could be clumsy down the road with 50000 players across many platforms, so not sure its a good idea; or perhaps enforce unique-ness of email address +initials combo, and then show a hashed version of the submitter email address, so that you can see SKZ 1234 and SKZ 5678 to know they are different.. or something.

----> I don't log the IP address of a submitter, but the general log has it; I tend not to wipe that log very often, so could be traced back; should I start actually logging IP with the score in the DB, so that its easier to track back?

Anyway, short of really painful (to users and me alike) closed system, we're open to attack; but protecting your c4a profile is the key --

If we suspect bad scoring, we can wipe all one someones profile-id from the system pretty easily, or from a specific game or whatever.

The Big Problem in my book is.. smart cheaters; what if you submit a bogus bot reasonable score every few hours, slowly increasing over a week or two, until you get top but reasonable score; it'd look legit.. nothing anyone can do :/

So in the end.. its all about fun, but we have to accept it can be cheated. To avoid cheatingy you have to have a closed system where only I can sign and build the binaries (say), , which introduces too much load (especially once we open up Windows, gcw, r-pi, etc etc..)

If anyone has any ideas (and comments on the above), let me know

jeff

 

[Neelix]

Given that Asmo has disclaimed the score, I think we can rule out cheating on his part in this case.

Does the database currently record the profile Id the score was submitted with?  

If so it would be interesting to find out if it actually matches Asmo's ID or not.   If so that suggests to me that the issue may have come from Asmo's end.  (Corrupted hi-score file when he terminated his session perhaps?)

@Asmo:  does the timestamp on the score match your terminated MsPacman session?

- Neelix

 

[Ziz]

I just think, that Skeezix' MAME-Score-Grab-Code, which is based on well known memory parts, where the score should be located, grabbed the memory without a highscore ever added, which results in a random (?) number, which was committed as Asmo exited the game ahead of time.

 

[Kumaki]

Setting aside, for the moment, that Asmo himself disclaimed this score...

If someone is smart enough to cheat, they would also be smart enough to know that Ms, Pacman can score multiples of ten only, so this would be an obvuious fake.

No cheater submitted that score, it's a glitch of some kind, like Ziz suggests.

 

[Kumaki]

A thought, regarding profiles?

Is it possible to have more than one profile on the same Pandora?  And you could switch between them?

I'm thinking of this in terms of when/if this goes to other platforms...and, say, someone has a profile on the family computer, now no-one else in that family could have a profile...there should be a way to have multi profiles on the same unit...although this opens things up to scoreboard hogs.

 

[Asmo]

Given that Asmo has disclaimed the score, I think we can rule out cheating on his part in this case.

@Asmo:  does the timestamp on the score match your terminated MsPacman session?

- Neelix

Thanks! Thought a few months of furious button mashing were on the verge of ruin for a moment there..!

Re: Timestamp - there's a timestamp? I see only the date, but it would appear to tally with the aborted game I played that evening. Didn't see a stupid score when I started the game, reset (F3) and quit C4a quickly when I had to do something else. Logged in the next day and...

I just think, that Skeezix' MAME-Score-Grab-Code, which is based on well known memory parts, where the score should be located, grabbed the memory without a highscore ever added, which results in a random (?) number, which was committed as Asmo exited the game ahead of time.

My profile/id etc has never been seen or compromised, so I'm guessing this is the case. The score even looks like 'other data' somehow. I wonder if (IIRC) 'chompjil'* who had a crazy Donkey Kong score before was a victim of something similar and abandoned it as a result..

If someone is smart enough to cheat, they would also be smart enough to know that Ms, Pacman can score multiples of ten only, so this would be an obvuious fake.

No cheater submitted that score, it's a glitch of some kind, like Ziz suggests.

Thanks! I was hoping there were signs that this wasn't even plausible for Superman, let alone Billy Mitchell on his best day.

AFAIK the world record score on Ms Pacman doesn't even exceed 1,000,000 - the 'score' of 6,797,973 is literally impossible, from every angle.

*EDIT:

It was 'Chompjil' back in May, same thing, a DK score way beyond the world record (and it seems Chompjil never went back)-

http://boards.openpandora.org/index.php/topic/12127-release-compo4all-mame-ladderscoreboard-competition-for-classic-arcade-games/page-26#entry246085