Syntax
Level: Smitemaster
Rank Points: 1218
Registered: 05-12-2005
IP: Logged

Having watched BT:L1:1N2E I would see that his demo is way too clean for a first pass-through of that level (ie no #1 overtaking). Very suspicious especially seeing the margin of over 25% between him and michthro. I've rarely seen such a relative gap.

NiroZ
Level: Smitemaster
Rank Points: 1302
Registered: 02-12-2006
IP: Logged

Having just checked the demo in question, I have to say that it is defiantly a bot.
For those of you who haven't done the room/can't be bothered checking, here are the stats

Player    Time taken in demo  Moves  Date
Halyavin  7:28                354    5th of January, 07
Michthro  3:53                467    June 9, 06

However, the most damming piece of evidence is (as I had a hunch it would be in the case of bots) the way Beethro moves. I have never seen a demo which had such an odd pace, spent the longest amount of time in thinking about moving the sword, or taking its time in what should be a straightforward drive down a line of roaches.

Now, with this overwhelming evidence, I invite Halyavin to play Counterstrike 1.6. I wish I would be there when he says 'How the **** did he see me? NOBODY should be able to move that fast! How in the heck did he get that many headshots!'

EDIT: just checked and found that he also got 8th in BT:L1:3E. The fact that it took 7 minutes is very suspicious, and the fact that it was made on the 28th of December is also interesting. However, an inspection of the demo seems that this is a different bot, perhaps it gave him instructions that he had to input.

---

[Last edited by NiroZ at 01-07-2007 02:12 PM]

michthro
Level: Smitemaestro
Rank Points: 679
Registered: 05-01-2005
IP: Logged

Yup. The pace of that demo is awfully unnatural. A pause between every two consecutive moves? Look at the first stretch. Why is it so jittery? So maybe for some reason that's how halyavin plays? Nope. Look at his demo for BT:L3:2N. That's the real halyavin playing - typical demo of an actual human in action. Keys are held down, everything.

My guess is that after obtaining a move sequence with his program, or recording it from playing with UU in the editor, or whatever, he feeds it to DROD with this program his "friend" wrote. He thought it would be better to throw in some random delays between moves. I'm not sure - the delays at the start are shorter. But I don't think that demo was recorded with a human playing.

There's too much wrong here. This thread plus halyavin's demos are leaving little doubt in my mind that he's cheating.

Mattcrampy
Level: Smitemaster

Rank Points: 2388
Registered: 05-29-2003
IP: Logged

While it's interesting that there could well be a DROD solver I'm heartened that if there is one, it hasn't had much of an effect on the highscore table.

____________________________
What do you call an elephant at the North Pole?

bandit1200
Level: Master Delver
Rank Points: 290
Registered: 10-29-2003
IP: Logged

Message content removed because I replied to one of Tim's remarks without reading further.

---

[Last edited by bandit1200 at 01-07-2007 04:03 PM]

Banjooie
Level: Smitemaster

Rank Points: 1645
Registered: 12-12-2004
IP: Logged

I like where this thread is going.

Maurog
Level: Smitemaster

Rank Points: 1501
Registered: 09-16-2004
IP: Logged

I think this sensational presentation of a rather primitive DROD bot was completely ruined by lack of people running in circles and screaming "Help! Help! Hide your sisters and nephews! The bots are coming! The bots are comiiiing!"

____________________________
Slay the living! Raise the dead!
Paint the sky in crimson red!

Elfstone
Level: Smitemaster

Rank Points: 1285
Registered: 03-01-2006
IP: Logged

On a side note and just because it has piqued my curiosity, how does a member have 32 rank points when he/she has totals of +5 and -9 in his/her posts? Should that not leave 6 rank points?

____________________________
Winner of: Novice Architect Excellence 2006.
FAPCA - Technical Design Excellence in Layout and Aesthetics

tokyokid
Level: Smiter

Rank Points: 453
Registered: 05-06-2006
IP: Logged

Elfstone wrote:
On a side note and just because it has piqued my curiosity, how does a member have 32 rank points when he/she has totals of +5 and -9 in his/her posts? Should that not leave 6 rank points?

Members start out with 10, and she/he might've gotten points from clicking the bar.

Elfstone
Level: Smitemaster

Rank Points: 1285
Registered: 03-01-2006
IP: Logged

tokyokid wrote:
Members start out with 10,

emmm yes : 10+5-9=6 or did when I was at school about 98 years ago

and she/he might've gotten points from clicking the bar.

- of course; I had forgotten about the bar!

____________________________
Winner of: Novice Architect Excellence 2006.
FAPCA - Technical Design Excellence in Layout and Aesthetics

Tahnan
Level: Smitemaster

Rank Points: 2467
Registered: 11-14-2005
IP: Logged

OK, I'd like to change my answer on the poll to say that there's a definitely possibility that someone among all the players--namely, our poll-starting friend--uses a solver. (Then again, I'm not sure there are any useful DROD-solvers, depending on whether "useful" is taken to mean "can be used on more than a few brute-force-ish rooms".)

schep
Level: Smitemaster

Rank Points: 865
Registered: 03-01-2005
IP: Logged

Cheating is ill-defined.

I don't remember if CaravelNet came with a 'Terms of Usage' type thing, and I can't find one now. So it seems like whether certain methods of coming up with solutions are to be accepted is up to common sense and consensus.

Certain things I think we'd all agree are absolutely unacceptable:
Hacking the server and/or exploiting spider bugs to submit invalid demos or mess with other people's demos.
Submitting demos that would not be possible in-game (entering by an entrance you couldn't possibly get to before conquering the room).
And by precedent, copying existing demo move sequences exactly.

And now we're discussing, hypothetically at least, using a computer's aid to help optimize rooms. A lot of people have assumed or implied certain versions of doing so count as cheating. (Including the one user I notice specifically suspected of actually doing so.) I don't really see why this is "wrong". Whether going for conquering or solving, DROD boils down to coming up with a good move sequence. There are plenty of ways to do that. I've used computer power in small ways to help me conquer rooms before; was that okay? If I'm capable of writing a pretty cool program that significantly helps optimize and start using it, that's not okay? Does the answer depend on whether I tell other people about this script or whether I make the script publicly available?

Of course, it's obvious to me I don't really speak for the users who really enjoy and spend time at optimizing. I've never seen the point of making DROD more challenging by means of keeping checkpoints scarce or anything like that, so I thought Unlimited Undo would have been really nice (but too bad it was only going to be available for optimizing, not conquering). I was absolutely befuddled that a significant number of optimizers didn't want UU in DROD at all.

Operating on discussion and trust, and dealing with individual problems like the demo-copier, has worked okay, and we don't want to lose those. But whether the suspicions and possibilities about all this stuff are reality or not, maybe it's time for an official policy about rules for high-score demos.

zex20913
Level: Smitemaster

Rank Points: 1723
Registered: 02-04-2003
IP: Logged

I just want to comment, schep, that intentionally copying copious amounts of optimal demos is a bad thing--inadvertently coming up with the same move sequence is sometimes mandatory. Also, sometimes, in looking at a demo, the idea given will yield the same move sequence among optimizers after the lynchpin.

Personally, I feel that if you've written a program to assist you with DROD, use it all you want. 3-wide tar cutting has been hardcoded into my brain. (88168816...rotate if necessary.) Other sequences as well (767676767676 for mud, ctrl+q for some hordes in a 2-wide corridor, etc.) So long as you made the program yourself, and so long as the solutions remain local, I don't see much of any conflict, it's just a different skill set being emphasized, and I wouldn't count it as cheating.

My $.02.

____________________________

Jatopian
Level: Smitemaster
Rank Points: 1842
Registered: 07-31-2005
IP: Logged

schep wrote:
And now we're discussing, hypothetically at least, using a computer's aid to help optimize rooms. A lot of people have assumed or implied certain versions of doing so count as cheating. (Including the one user I notice specifically suspected of actually doing so.) I don't really see why this is "wrong". Whether going for conquering or solving, DROD boils down to coming up with a good move sequence. There are plenty of ways to do that. I've used computer power in small ways to help me conquer rooms before; was that okay? If I'm capable of writing a pretty cool program that significantly helps optimize and start using it, that's not okay? Does the answer depend on whether I tell other people about this script or whether I make the script publicly available?

Of course, it's obvious to me I don't really speak for the users who really enjoy and spend time at optimizing. I've never seen the point of making DROD more challenging by means of keeping checkpoints scarce or anything like that, so I thought Unlimited Undo would have been really nice (but too bad it was only going to be available for optimizing, not conquering). I was absolutely befuddled that a significant number of optimizers didn't want UU in DROD at all.

Indeed. I have to wonder whether the optimizers use the orb solvers, or if they consider those cheating. And how about those checkpoints?

____________________________
DROD has some really great music.
Make your pressure plates 3.0 style!
DROD architecture idea generator

zex20913
Level: Smitemaster

Rank Points: 1723
Registered: 02-04-2003
IP: Logged

I'm not sure if I'm an "optimizer" or not. I'm close to one.

Either way, if an orb solution is not straightforward or interesting without being overly challenging, I have been known to use the orbsolver. And I use checkpoints...if they're on an optimizable path. I don't go out of my way for them.

____________________________

Rabscuttle
Level: Smitemaster

Rank Points: 2539
Registered: 09-10-2004
IP: Logged

Jatopian wrote:
Indeed. I have to wonder whether the optimizers use the orb solvers, or if they consider those cheating. And how about those checkpoints?

Orb solvers are not necessarily going to give you the optimal solution, especially if there is a choice in which orb to hit. (example, sort of. Actually, a better example would be King Dugan's Dungeon : Third Level : 1 South (going the correct way through the room) If memory serves, there are two five orb solutions, one of which is quicker.)

There will also be cases where the quicker solution involves hitting more orbs. Orb solvers can be used, but they need to give all possible hitting orders. Personally, orb hitting puzzles aren't that fun to optimise, so I don't.

I find that checkpoints are often off the optimal path. You suck, architects!

---

[Last edited by Rabscuttle at 01-07-2007 11:49 PM]

Citrus
Level: Smitemaster

Rank Points: 1249
Registered: 01-08-2006
IP: Logged

Rabscuttle wrote:
I find that checkpoints are often off the optimal path. You suck, architects!

Heh heh, I purposely do that to give optimizers a greater challenge. Too bad my puzzles aren't challenging to begin with.

____________________________
Call me Citrus. Small system. Sometimes we'll sign off our messages with who wrote it if it's relevant. Colored text indicates who.

halyavin
Level: Delver
Rank Points: 52
Registered: 02-20-2006
IP: Logged

That is why I have to sometimes write down my demos on paper and then retype them. I have to switch between paper and screen that why I have so strange pauses in some demos.
My friend reduces amount of work a lot recently - currently I use his get_demo.exe program to dump my demos and so I switch between two windows instead of window and paper. It is still inconveniently because DROD occupies too large part of the screen.

halyavin
Level: Delver
Rank Points: 52
Registered: 02-20-2006
IP: Logged

michthro wrote:
Here's where this thread gets *really* weird: I just had a look at halyavin's scores, and guess what. He himself has some pretty amazing scores. It makes no sense whatsoever for someone who obtained those scores to accuse others of cheating, on no more grounds than that their demos are so good.

Let's get this straight:
-halyavin accused players of using programs that brute force short sequences. I couldn't get an example out of him, but he made it clear that that's what it's about.

I am not accusing, I am suspecting that can be done now and would be able to done in future.

-He himself has some excellent scores for precisely the kind of room he's on about. Rooms that a computer could very well handle. BT:L3:1N2E would be an easy one for a computer. In some cases it would require setting various goals one at a time, just as halyavin described in Development. BT:L1:1N2E is a prime example.

halyavin's demo for BT:L1:1N2E is awesome. It makes no sense for someone who came up with that to be so amazed at other demos that he accuses their authors of cheating.

Ok, I admit that "too good demos" was only a ground to start a thread. It is silly to start a thread if drod solvers generate worse demos than human, isn't it?

So what's this really about?

Is halyavin so incredibly conceited that he thinks he's the only one who can come up with good demos without cheating?

Is it some twisted form of bragging?

I'm pretty sure it's a lot of both, but mostly I think halyavin has written such a program, and instead of simply discussing it, used it to cheat. Let's consider the facts. There's no hard evidence, but everything points that way:
-We know he's a competent programmer, and very interested in the topic of room solvers.
-We know he does things like unlocking editing rights, and as BeefRow points out, clearly sees it as a shady business.
-Throughout this thread he was just too sure of his case. As if he's talking from first-hand knowledge, rather than on general grounds.

I thought much about this topic or I willn't start in otherwise.

-He obviously is under the mistaken impression that writing such a program would be bad in itself. He even attributes a program that extracts moves from demos, and unlocking hold editing rights, to a "friend". That's simple to do, and from other posts he's clearly capable of doing it himself.

In order to write such program I need to know the structure of Drod database. And I don't want to study metakit library for this. My friend knows much more about metakit and DROD database. Also, in order to interact with metakit you should use C/C++ (I am not sure - I haven't deal with metakit yet). My friend likes it but I am prefering Pascal.

(There's nothing wrong with room solvers, as long as you don't use it for scores. Why shouldn't someone write one and discuss the results on the forum? If it can be done it can be done, and we have to face it. If you want to prove that computers can help with optimising, this is not the way to do it. Simply discuss it openly, and leave it to us to deal with the consequences.)
-He is so convinced that some demos must have been obtained by computer, yet he has similar demos himself. If he didn't obtain his own good demos by cheating, he would know that there's nothing so amazing about what can be done with practice, experience, and effort. I think he's the one who cheated, and, being new to optimising, can't believe that others may be capable of finding similar demos themselves.
-He accuses people of cheating, right. Why complicated room solvers? There are much easier ways to cheat. UU is a much more natural suspicion. Why are room solvers the first thing that comes to halyavin's mind? The answer is obvious.

I never thought that UU is a cheating. I have seen a topic about it (I didn't read it closely though) and I thought this is very probably coming feature of TCB. I was wrong ?

I have been optimising for a long time. I'm heavily into it. I know what I'm talking about. Anyone who fairly came up with those two BT demos I mentioned (and there are more), would NEVER have been so amazed by anything else that they accuse others of cheating. That's the work of a very good player, who knows what optimising is about, who understands where good demos come from, who knows that if they can do that, they can equal similar demos. Or it's the work of a computer.

In brief, if halyavin isn't cheating, he's a very good player, in need of some serious help for starting this thread.

If I start the thread that forum is storing password in cookies and send it in each transaction, you will think that I am a hacker?

---

[Last edited by halyavin at 01-08-2007 07:35 AM]

Maurog
Level: Smitemaster

Rank Points: 1501
Registered: 09-16-2004
IP: Logged

If you want my opinion halyavin, the one thing that strikes me wrong is your attitude. I mean, you probably don't hang in this forum much so you just don't know, but if instead of this poll you'd post something on the lines of

halyavin didn't write:
"Hey, look at that, my friend wrote this really cool program, just look at what it can do! It's not that useful yet, but pretty neat, huh? Huh?"

the reaction would be different. In fact I bet you'd get much constructive positive feedback instead of the negative vibes you're getting here. This poll gives the impression of accusing top people of using programs that gain unfair advantage which is bad to begin with. Then people discover your demos in which you don't exactly hide that you use such a program yourself. Seriously, it's all in the presentation, dude.

____________________________
Slay the living! Raise the dead!
Paint the sky in crimson red!

halyavin
Level: Delver
Rank Points: 52
Registered: 02-20-2006
IP: Logged

Sorry, that I haven't said from the beginning, that I just want to know what you think about probability of existance of usefull DROD solver, how you attitude to using it, and what you think should be done if we want to prevent this threat.

halyavin
Level: Delver
Rank Points: 52
Registered: 02-20-2006
IP: Logged

And I am not sure that such programs should be shared with public if they exists. I see nothing good in nuclear parity.

michthro
Level: Smitemaestro
Rank Points: 679
Registered: 05-01-2005
IP: Logged

halyavin wrote:
I am not accusing, I am suspecting that can be done now and would be able to done in future.

Look, please, don't try and tell us you weren't making accusations. You even called the thread "Fair play".

In order to write such program I need to know the structure of Drod database. And I don't want to study metakit library for this. My friend knows much more about metakit and DROD database. Also, in order to interact with metakit you should use C/C++ (I am not sure - I haven't deal with metakit yet). My friend likes it but I am prefering Pascal.

You're only giving yourself away even more. You are well aware that you don't need to know anything about the database structure. All you need is something like Stefan's VerboseHold, in which you were very interested. In fact, you were disappointed when it didn't work on official holds (BT?), so you modified it. (Link.) Once you have the room data extracted you could use any language you want. And besides, extracting room data programatically would merely be convenient. Obviously you don't have to. You could feed room data to your program manually. What do you take me for?

If I start the thread that forum is storing password in cookies and send it in each transaction, you will think that I am a hacker?

This doesn't really make sense, but let me try : Knowing about cookies isn't the equivalent of some of your demos. My point is that it doesn't make sense for someone who obtained those demos without cheating to start a thread implying that similar demos must have been obtained by computer.

halyavin
Level: Delver
Rank Points: 52
Registered: 02-20-2006
IP: Logged

michthro wrote:

halyavin wrote:
I am not accusing, I am suspecting that can be done now and would be able to done in future.

Look, please, don't try and tell us you weren't making accusations. You even called the thread "Fair play".

How I should call it? It was hard for me to think out a meaningful name for this topic. Sorry, if I am hurt someone .

In order to write such program I need to know the structure of Drod database. And I don't want to study metakit library for this. My friend knows much more about metakit and DROD database. Also, in order to interact with metakit you should use C/C++ (I am not sure - I haven't deal with metakit yet). My friend likes it but I am prefering Pascal.

You're only giving yourself away even more. You are well aware that you don't need to know anything about the database structure. All you need is something like Stefan's VerboseHold, in which you were very interested. In fact, you were disappointed when it didn't work on official holds (BT?), so you modified it. (Link.) Once you have the room data extracted you could use any language you want. And besides, extracting room data programatically would merely be convenient. Obviously you don't have to. You could feed room data to your program manually. What do you take me for?

I was speaking about unlocking holds and dumping demos and not about dumping rooms. Demos are not located in .hold files as far as know. This is probably possible to remove edit protection from .hold files directly though, but I never tried that yet.

I agree that extracting room data with VerboseHold utility is convenient if you want to feed room to other program or to take into account monster order tricks. I was said that switching windows in order to type demos from dump is a bit unconvenient.

Beef Row
Level: Smiter
Rank Points: 471
Registered: 12-28-2005
IP: Logged

Ezlo wrote:

Rabscuttle wrote:
I find that checkpoints are often off the optimal path. You suck, architects!

Heh heh, I purposely do that to give optimizers a greater challenge. Too bad my puzzles aren't challenging to begin with.

I always figured everyone was doing that purposely, since I see it all the time.

____________________________
"Now I will repeatedly apply the happy-face rule"

michthro
Level: Smitemaestro
Rank Points: 679
Registered: 05-01-2005
IP: Logged

halyavin wrote:
That is why I have to sometimes write down my demos on paper and then retype them. I have to switch between paper and screen that why I have so strange pauses in some demos.
My friend reduces amount of work a lot recently - currently I use his get_demo.exe program to dump my demos and so I switch between two windows instead of window and paper. It is still inconveniently because DROD occupies too large part of the screen.

I see. So you're saying that for something like BT:L1:1N2E, that first stretch, until just before the orb, you painstakingly copy a sequence 8888888888888... one move at a time, instead of simply holding down 8 until you get near the orb. That explains it. Thanks for clearing that up.

halyavin
Level: Delver
Rank Points: 52
Registered: 02-20-2006
IP: Logged

After several occasions when I pressed the wrong key in the middle of long demo I became very carefull and sometimes check beethro position even in so obvious situations. It is so easy to type something wrong after several hours of DROD'ing .
BTW, I did used retyping in Beethro Teacher:L3:2N. Why pauses more natural here? I don't know, may be I have been less tired.

michthro
Level: Smitemaestro
Rank Points: 679
Registered: 05-01-2005
IP: Logged

schep wrote:
And now we're discussing, hypothetically at least, using a computer's aid to help optimize rooms. A lot of people have assumed or implied certain versions of doing so count as cheating. (Including the one user I notice specifically suspected of actually doing so.) I don't really see why this is "wrong". Whether going for conquering or solving, DROD boils down to coming up with a good move sequence. There are plenty of ways to do that. I've used computer power in small ways to help me conquer rooms before; was that okay? If I'm capable of writing a pretty cool program that significantly helps optimize and start using it, that's not okay? Does the answer depend on whether I tell other people about this script or whether I make the script publicly available?

The way I see it is that a scoring system such as we have is all about what players can do themselves, playing DROD. It's about humans competing against humans, at DROD. It's not a programming contest. I mean, by the same reasoning I should be able to enter chess tournaments with a computer under my arm, provided I programmed it myself. Computers do enter chess tournaments, but the programmers don't get credit as players. The program itself gets a rating etc. The point is that DROD, like chess, is a game meant to be played by humans. How computers do at those games is a completely different issue. If you want to do AI research by way of solving DROD rooms, the highscore table isn't the place to publish your results.

I wouldn't mind if someone did start a research project on solving DROD by computer, but if they uploaded scores it should go under something like "halyavin's bot", not "halyavin". (Btw, from a theoretical perspective it would be far more interesting to see how "halyavin's bot" does, than "halyavin and his bot".) I would also then like the option to hide known bots from the table.

So, we don't have an official rule that the highscore table isn't a programming contest, but I thought it kind of goes without saying. Guess we badly need some official rules. Which would turn it into a find-the-loophole contest. Unless of course, as I thought, we could rely on trust, and people at least checked with the forum before using something that may very well be debatable.

I was absolutely befuddled that a significant number of optimizers didn't want UU in DROD at all.

I never actually got around to saying anything in that thread after the announcement that UU won't be available after all. Originally I was all for it, but now I'm glad it won't be included. The mistake I made was thinking only of a relative handful of rooms where it could save a lot of time, while forgetting that far too many rooms will become too easy to optimise. I don't want a situation where for most rooms the first one through gets an unbeatable #1. Not with a scoring system where the first one to get the best score gets more points, and #1s are listed separately. I'd rather put up with the tedium of optimising some rooms than have the highscore system become a realtime race.

...maybe it's time for an official policy about rules for high-score demos.

Yes, can we please have an official ruling?

NiroZ
Level: Smitemaster
Rank Points: 1302
Registered: 02-12-2006
IP: Logged

michthro wrote:I'd rather put up with the tedium of optimising some rooms than have the highscore system become a realtime race.

From what I have seen, it is already somewhat of a race already.

michthro
Level: Smitemaestro
Rank Points: 679
Registered: 05-01-2005
IP: Logged

Yes, there are already too many such "race" rooms. I don't want a whole lot more.