Richard Gebauer

Should these scores be allowed on the ranking?
Post Reply
thefinerminer
Site Admin
Posts: 113
Joined: Tue Jan 08, 2008 3:33 pm
Location: UK, Scotland
Contact:

Richard Gebauer

Post by thefinerminer » Fri Dec 18, 2009 10:57 am

Richard sent Recorder videos of 4-17-54 on 10th November. The mouse movement was very jumpy, but this happens sometimes with the Recorder. I also found the solving very strange, especially on the Expert video and the Expert 3BV was pretty high (176). All scores were made the same day.

I asked him to make MSX videos and on 06 Dec he sent 3-17-56 (all made same day). The 3BV are high (11-63-187) and the mouse movement is still very jumpy and some of the solving is very strange.

Should we ask for more videos of different versions? Do you think these are fake or real? And if he has hacked MSX, how did he do it?

This last question is very important. Gongor Erdenebilguun also submitted very good scores on MSX which suggests MSX is hackable, but no one knows for sure.
Attachments
RichardGebauer55,71-187-061209.mvr
(31.31 KiB) Downloaded 464 times
RichardGebauer16,58-63-061209.mvr
(11.22 KiB) Downloaded 441 times
RichardGebauer2,84-11-061209.mvr
(2.07 KiB) Downloaded 458 times

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Richard Gebauer

Post by EWQMinesweeper » Fri Dec 18, 2009 1:23 pm

after watching the beg video i thought it was someone with poor solving skills that has fast forwarded the vid.
after watching the int vid i am not so sure: those hesitations just scream "FAKE!" but i find that the cursor movements are even more robotic than those in the james shannon exp vid and it even seems to be a not so clever bot (flags all mines).

now after also watching the exp vid my theory is that it might really be just a noob who managed to make vis replays play at much higher speed. the way the original games were played is quite clever, all clicks have been done after roughly the same time differences, which makes the 56s game look convincing - if there weren't among other things the hesitations, when duing the original game he probably had to serach for easy patterns...


my suggestion is that you tell him we know that those videos are fake. if he claims they are not he should send arbiter scores (remember that gongor cheated on both msx and clone?). tell him that if his next videos are also fake he won't be accepted to minesweeper rankings in the future.
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

User avatar
RonnyDeWinter
Posts: 77
Joined: Mon Dec 01, 2008 9:11 am
Location: Netherlands
Contact:

Re: Richard Gebauer

Post by RonnyDeWinter » Sat Dec 19, 2009 6:27 pm

99,99% sure it's done using a solving bot.

- Inhuman mouse press/release time (watch the smiley face, it hardly does 'ooh' on a click
- Constant mouse speed
- Constant click speed
- Sudden pause at the end of each solved island (not sure why)
- Flagging all mines
- Always moves too the nearest solvable flag/square
- Unusually high 3BV's for record boards
- All records made on the same day
- No wasted clicks
- No wasted double clicks
- Double clicks aren't visible on the board (due to non-existing press time)

Need I go on? :D
NF 1 (0.96) + NF 15 (14.20) + NF 61 (60.18)

All my minesweeper records

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Richard Gebauer

Post by EWQMinesweeper » Sat Dec 19, 2009 9:18 pm

so sure it's a bot?

i mean gongor found a way to speed up his clone and msx videos and that would explain all your points except maybe #2 and #3 (how constant is he really - also to me it doesn't look like all clicks are done on the same pixel in each square)
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

User avatar
RonnyDeWinter
Posts: 77
Joined: Mon Dec 01, 2008 9:11 am
Location: Netherlands
Contact:

Re: Richard Gebauer

Post by RonnyDeWinter » Sun Dec 20, 2009 2:10 pm

EWQMinesweeper wrote:so sure it's a bot?

i mean gongor found a way to speed up his clone and msx videos and that would explain all your points except maybe #2 and #3 (how constant is he really - also to me it doesn't look like all clicks are done on the same pixel in each square)
It's easy to randomize the press coordinates by a couple of pixels, although I agree that I also expected that he would have forgotten that detail.

Also I can't explain the stationary pause at the end of each island: It's not like a human with a mouse but it's also not typical for a bot. The only thing that seemed somewhat logical, would be a touch screen.

If it would be a touch screen implementation of minesweeper it would be quite interesting:
- How does he distinguish right, double and left clicks on a touch screen?
- All 'mouse movements' would be straight lines, are they?
- All 'mouse movements' would have a constant speed, do they?
- Wouldn't there still be visible 'mouse' presses in the video using a touch screen or did he program a touch as a 'mouse click' in stead of a separate press and release? (that would be cheating because it gives a definite unfair advantage)
- It does explain the pause at the end of an island, because the cursor wouldn't move till the next touch.
- If he presses the squares faster than the cursor is allowed to move to squares you wouldn't see a pause, which would explain the smooth movement around nearby squares.

What definitely sucks, is that to be sure about these things, you would have to be able to slow down the vid, see the mouse path, see the left, right and double clicks. MinesweeperX doesn't offer these features, so being able to watch a vid from him on Clone would be perfect.
NF 1 (0.96) + NF 15 (14.20) + NF 61 (60.18)

All my minesweeper records

dcalafiore
Posts: 9
Joined: Thu Aug 27, 2009 7:49 pm
Location: Italy, Sicily

Re: Richard Gebauer

Post by dcalafiore » Sun Dec 20, 2009 2:53 pm

According to me, it's a fake.
I agree with Ronny, for the reasons he explained and another thing: after each hesitation cursor quickly moves towards a square and hit immediately the target, without slowing down or wheeling around it. Who is able to do it in the world? :?

Though I think it's a good idea asking him to make Clone or Arbiter vids, because we could slow down or stop the video, look at mouse path and analyse his game better. And it's an occasion to test anti-hacking skill of these programs! :)
My times: 2.05+20.17+78.76=100.98
18th in Italy ranking

Mucleus
Posts: 7
Joined: Tue Dec 02, 2008 1:58 pm

Re: Richard Gebauer

Post by Mucleus » Sun Dec 20, 2009 6:41 pm

I don't think it's a bot. The solving is too random (esp. around 24s), and there's places where he does sensible chording, and then other bits where he does random clicks when he needn't, etc etc. There are programs available which can track and modify memory used by programs, so (say) you can freeze the timer, see where the mines are, work out where to click, start the timer, make the move, repeat, etc.
As for the large 'gaps' after the end of a solved area, I have no idea. Particularly on the int game it just makes the solve look weird, so you'd have to wonder why he put them in.

Regarding slowing down playback, this is where Cryslon's <someformat>2txt idea would come in handy :P Does anyone know the minesweeper x format (i.e. is it data+checkdigit)?

dcalafiore
Posts: 9
Joined: Thu Aug 27, 2009 7:49 pm
Location: Italy, Sicily

Re: Richard Gebauer

Post by dcalafiore » Sun Dec 20, 2009 11:24 pm

Assuming it's a bot would explain hesitations. Maybe the bot is slow in calculus and, if it looks for the nearest solvable square, it could take more time when the square is further ( I assume that it checks the squares from the nearest to the furthest).
My times: 2.05+20.17+78.76=100.98
18th in Italy ranking

Cryslon
Posts: 122
Joined: Sun Dec 28, 2008 7:41 pm

Re: Richard Gebauer

Post by Cryslon » Mon Dec 21, 2009 6:19 am

msx format is terrible obfuscated binary mess, it's not data+checksum. I can't even figure where and how it stores the board layout, because the same board generates entirely different file. I guess it's XOR'ed with some random bytes or sth like that, but i have no time to study it further right now... Where are Curtis&Rodrigo?!

[gone to write that clone and msx are crap in gb]
Go IRC! (try mibbit)

thefinerminer
Site Admin
Posts: 113
Joined: Tue Jan 08, 2008 3:33 pm
Location: UK, Scotland
Contact:

Re: Richard Gebauer

Post by thefinerminer » Thu Aug 11, 2011 10:43 am

Consensus was that his scores are fake - this topic is now closed.

Post Reply