Bots Work
DeusExMachina
REGISTERED
I just recorded a 5 second score on today's Freecell game of the day (#1674554720), and it is utterly bogus.
On my iPad, I found a 16-tap solution that I can execute totally legitimately in 7 seconds.
I then buddied up with Google to find a Windows app that records and plays back mouse clicks. The first one I tried was no good, but the second one worked a treat, as my British friends would say. ReMouse. It's free. The only setup, if you can call it that, was to try with various speed settings for playback. 5x and faster didn't work. I'd wager clicks came in before super moves had executed. But 4x played smooth as silk, turning my 20-ish second game by mouse into a 5 second record.
I've read several threads speculating that bots might be behind the impossibly fast times, but I didn't see any posts where someone actually gave it a try -- and admitted to it. Anyway, I'm here to testify that it's not only possible but dirt simple.
It's worth emphasizing that all the app did was record and playback my moves at accelerated speed. The solution was entirely my own--not like some around have speculated about that a bot could be actually working out a superfast solution. That may be entirely possible, but if you've gotten good enough at speed play that heavily leverages super moves, then it's far simpler to use an app like Re-Mouse to juice your times.
The lesson being: you can't trust times. At least not without corroborating evidence of some kind. Like this video I posted a couple weeks back: 
Of course, a video like that could also be faked fairly easily. Trust me, I am an engineer on a video editing program called Premiere Pro. It would've been trivial to make that video look like I'd recorded a 6-second time just like a certain player who books the fastest time pretty much every day.
Having quenched my curiosity about how easy it is to log an amazing score by blatant and undeniable cheating, I hereby vow that I will never again use ReMouse or anything like here on Green Felt. And if anybody ever questions whether my time is legit or just wonders how I played a game so fast, I'll happily post a video.
[edited to add] My secret hopes are that any players who are cheating in this manner will be shamed into stopping it, and that others who are posting impressively fast scores legitimately will put up a video to show us all how they do it.
Cheers,
Mark
Comments
Nice idea, @DeusExMachina! I like that you were able to implement your solution without coding. We have always expected that one day someone would be able to automate turning in fast times, but I didn't expect this sort of solution.
With respect to your point about using a standard solver, one of the hard things about using one would be that they wouldn't support our super-moves, so their output wouldn't be efficient by our "move" metric without some post processing.
PS We used to have a way to easily export the game state into a solver, but it was just too tempting to use, so we dropped it.
I have to say I was surprised that this post didn't generate any discussion given how many times "bots" have been accused of domination of the leader board. I had satisfied myself that crazy fast times without bots were possible (at least in the few games I play) and had even tried it a couple of times for yuks though it still took me too long to be on top B^)
I would think this technique would only inspire others to "cheat" - I'm sure no one so inclined would be "shamed" into not doing it. I just wish there was a leader board that was based on moves so that I could better ignore the speedsters (if this sounds familiar, I have made the request before B^) I find the leader board useful to keep track of which games I have played since I only intend to play one Seahaven game per day. The other game I play is not on the leader board, so if I miss a day, it's gone for my purposes...
"show the players first score not their fastest" I would think that using ReMouse would still show BOTH scores??... I saw a hack that was really simple - the players ID's were "user1, user2 user3 ad nauseum" no bot just replay till you get it right. Another user simply kept replaying, I saw 4 in a row. My solution, click on "show the players first score..."
and play to enjoy YOUR game. If you must compete remember DeusExMachina's remouse and all of the ego's out there or you'll always be pissed off. Self compete, compare your moves per second with your previous games, compare your position with the total played -- You made 97 moves, game total 3650 - 365 is 10% - you're in the 90th
percentile! Can you live with that? It's an easy compute to find where you are in the total -- find your own way to "compete" just remember - If it isn't fun any more, it's your problem.
The "first score" box isn't very useful since if you do multiple games, you will just save your official name for when you figure it out. I've found that there are few changes with the box checked.
If your advice was for me, I always play for enjoyment and relaxation. I can't imagine getting that by obsessing with being first on this leader board. If it were based on moves, it might be interesting to see if I hit on an efficient enough solution to make that leader board on the first try, but once I've solved the layout it doesn't really matter any more.
don't you hate typos B^) " I meant I can't imagine getting obsessed" ...
No, the advice was for anyone who wandered in. Your comment - "I have to say I was surprised that this post didn't generate any discussion given how many times "bots" have been accused of domination of the leader board." -
Is why I made my comments - for the "accusers" if they show up.
Quick FreeCell question for DeusExMachina: Since once you've found the minimum-tap super move solution you're obliged to synch up the ReMouse speed option to the Auto-Finish algorithm in order to achieve the fewest-second score, would it be simpler and faster to use a 1-card-at-a-time solution (foregoing super moves and Auto-Finish altogether); capture all those taps in ReMouse; and then execute ReMouse at hyper speed? E.g., 70 moves at 100X?
I wondered about the same thing, @GammaRho496. That would almost certainly allow much faster playback speed. In fact, the tool you’re using may well be the limiting factor with that approach. If memory serves, Re-Mouse tops out 8 or 10x—it doesn’t “go to eleven” like Nigel Tufnel’s guitar amp.
Still, since I’m able to perform 15-18-tap solutions in about 4-7 seconds* (on iPad), I’d imagine that one who’s skilled at playing “straight” (without leveraging super moves) could execute 40-step solutions in 40 seconds, then play back at 10x for a 5-6 second time.
I strongly suspect that’s exactly what one player around here does. You know who I’m talking about: the only who routinely not only books the fastest time but also rates among the lowest move-counts. In my experience, reducing the number of recorded moves necessarily comes at the cost of increasing the number of taps/clicks. So I won’t believe that a 71-move/4-second time is legitimate until I see the video evidence.
Sorry. Contrary to my pledge above, I did it again.
After playing anonymously for a while--to avoid logging a bunch of 4 second scores--I logged in after booking a legitimate 3 second score. But my efforts at being a good GreenFelt citizen went awry because without thinking I hit 3 seconds a few times before signing in.
At that point, I knew I had no prayer of cleaning up the mess by knocking another second off--at least not fairly. I mean, we all know that nobody could complete a game of freecell in two seconds without the software equivalent of performance enhancing drugs. So I reluctantly resorted to ReMouse again.
So in case you questioned that 1 second score for game https://greenfelt.net/freecell?game=686960134, you were right to. (But I swear on my DVD of The Life of Brian that my 3-second scores are totally genuine!!!)
This is for the folks who are trying to get under the 5 minute mark.