Ridiculous math

[JMC]

I really like this game, BUT:

The stated "odds of striking" an opponent are ludicrous in the higher difficulty settings. Clearly ludicrous. I've been playing this game for weeks on every difficulty setting and it's absolutely apparent that the "difficulty setting" is merely a matter of adjusting the odds in the computer's favour. Time and time again I see him strike successfully four times in a row at 30% to hit, while I struggle to get one at 70%.

I wish the developers would find a different way to adjust the difficulty setting. Between the timelines for many missions, and the recalculated odds of hitting any enemy, most higher-level difficulty missions become an enormous save-load festival of retardation.

[Gambit]

Er. Where do I begin...

The odds to hit are dependent on terrain and only terrain. The difficulty setting only changes the amount of gold you and the enemies get, and if you don't believe me, you can have a look at the source code.

'Nuff said.

[zookeeper]

The stated "odds of striking" an opponent are ludicrous in the higher difficulty settings. Clearly ludicrous. I've been playing this game for weeks on every difficulty setting and it's absolutely apparent that the "difficulty setting" is merely a matter of adjusting the odds in the computer's favour.

No it's not, nothing adjusts the chances to hit in anyone's favour. You're absolutely wrong.

Time and time again I see him strike successfully four times in a row at 30% to hit, while I struggle to get one at 70%.

Post a replay.

[HomerJ]

Gambit, can you post that cute troll picture again?


Greetz
HomerJ

[Rya]

Despite from what the devs say, the AI still cheats with the RNG. At critical moment it always hits. :p
The source code doesn't reveal it, but I've seen the AI owning a player often simply by doing reckless things and luckily getting the best result from it. I suspect the AI secretly rewrites the rand() function.

[thespaceinvader]

No. It doesn't. Conspiracy theories aside, this is a very clear example of the bias inherent in peoples' understanding. You think the RNG is biased. You therefore remember the times when the AI gets lucky, and hits four times in a row against your thief on a village, who was the linchpin of your strategy and had to live (for instance). You don't remember ALL the thousands of times when it DIDN'T. You don't remember all the times when YOU killed that critical unit on which your strategy hinged, and which you had maybe a 3% chance of killing.

If you seriously think the RNG is biased, there is a whole SLEW of replays on which you can perform statistical analyses to prove your point at replays.wesnoth.org.

When you can't prove it, shut up about it. You are WRONG.

[artisticdude]

Where have I seen this thread before?

I agree 100% with tsi. Would you be happier if the ai always had the short end of the stick, and you hit 99% of the time? I know I wouldn't be. But then there's still a 1% chance that you will miss, and even though the other 99% of the time you hit, you'd still think the ai was cheating because you missed one time out of one hundred.

The ai has been carefully developed over a long time, and I can tell you sincerely and truthfully that I have never suspected it of cheating. Never.

In short, if only the ai was random and you always hit, the ai would always be underpowered and you would always win no matter what, which would make for boring gameplay. The current random system gives you a different battle every time, and you must adapt your strategy to the current situation, regardless of what transpired in earlier battles. That's what makes Wesnoth fun: the replay value (which is practically limitless).

[pauxlo]

If you seriously think the RNG is biased, there is a whole SLEW of replays on which you can perform statistical analyses to prove your point at replays.wesnoth.org.

Nah, these are multiplayer matches, not those against the AI in single player. On multiplayer matches, the random numbers are generated by the server, so these don't count to prove or refute the theory "the AI secretly rewrites the rand() function".

(I still don't think this theory is right, but replay analysis is not the right tool here.)

[Gambit]

Note: Rya and pauxlo are joking about the AI rewriting a C++ function to cheat.
Or at least I hope they are, because that is the most rediculous thing I've ever heard.

[cretin]

if only sarcasm could translate over the internet better life would be so much sweeter.

[Daxion]

Eventhough I totally agree with TSI, I would just like to point out that it is possible to give
the AI a better hit% than normally. This can be done via WML (don't remember the command though). However every time I looked into a scenario to for one reason or another, this was never used.
Hmmm sometime I wonder why the dev don't simply change the rnd() simply not to have to put up with all the complains...

[jb]

The RNG is not broken.

Thread locked.

[Jetrel]

and if you don't believe me, you can have a look at the source code.

Emphasis mine. Posting after the lock for a PSA.

OUR CODE IS NOT A SECRET.
Unlike practically any other game, you can actually look at it.