Random numbers in attacks... and idea

General feedback and discussion of the game.

Moderators: Forum Moderators, Developers

User avatar
Unnheulu
Posts: 738
Joined: November 25th, 2007, 4:50 pm
Location: Cymru
Contact:

Re: Random numbers in attacks... and idea

Post by Unnheulu »

Chris NS wrote:It's a simple law of maths. If your mage is attacking the enemy leader with a 70% chance to hit and four attacks, and any one hit will kill him and win the scenario, the probability of missing all four times is 100%.
:lol2:

Dave
Founding Developer
Posts: 7071
Joined: August 17th, 2003, 5:07 am
Location: Seattle
Contact:

Re: Random numbers in attacks... and idea

Post by Dave »

Edward V Riley wrote:will do. Hope it doesn't take long with my dialup to post it though.
I look forward to it with bated breath.

Usually when I ask people to provide some kind of evidence of the chance factor of the game not working as intended, they enthusiastically promise it, and then I wait...and wait.....and guess what? Never here back.

So I'm looking forward to being surprised. :)

David
“At Gambling, the deadly sin is to mistake bad play for bad luck.” -- Ian Fleming

uzy5o
Posts: 52
Joined: March 9th, 2008, 2:39 am

Re: Random numbers in attacks... and idea

Post by uzy5o »

This is merely human nature: people are much more likely to remember instances of bad luck screwing them over than good luck saving their behinds.

zelurker
Posts: 4
Joined: April 18th, 2008, 1:31 am

Re: Random numbers in attacks... and idea

Post by zelurker »

Ok, just to show the fairness of the random numbers.
I don't post a replay, because actually I have quited after this turn, but I could record it (in this case you need just the last savegame, that's it ?).

Anyway, this is on an amd64 linux system
campaign : heir to the throne, misson siege at elensfar

This is not extremely bad luck, just a little bad luck compared to what I am used to, but nothing extraordinary...

1 of my archers tries a 11-5 attack at 60%, hits only once.
The opponent had a 10-2 counter attack at 60%, he hits twice (100% hit rate, wow !).

Then I try an entangle at 6-2 (60%) -> hits twice (100%)
the opponnent had a 10-3 at 60% -> hit 3 times at 100% too !!!

Then a knight : 15-2 at 60% -> misses twice (0%)
sorry I forgot to write down what his opponnent did, I thing he has hit 1 of his 2 hits, which is in the average for once.

Then a commander at 5-3 (60%) -> misses 3 times (0%) !!!
Then 10-4 (60%) -> hits 3 times (which is just slightly above the average)

A thief 5-3(80%) -> hits twice
Another one : 5-3(60%) -> hits only once, his opponent had a 8-3(40%) and he hits twice !

Then 10-4(60%) -> hits twice (which is 50%)
The opponent misses all his 3 shots for once, which balances a little things at this point (it was a 10-3(60%))

And I finish with a 7-4(60%) -> hits only once.

So...
If you take all the attacks including those of the opponents I can believe these are still really random numbers.
But if you look a the suites of numbers produced, it's really unfair at this point, and it's not exceptional at all, it happens all the time this kind of thing.

Ok, I have made a few tests and I have seen that if you improve the chance to hit of high level units, then the 3 level 1 thieves who appear in this mission have almost no chance to survive for very long !
Same thing if you improve chance to hit when the 1st hits are missed.
So I don't see any easy solution now, but at least don't tell me that these numbers are fair !

Anyway maybe the best option is to try to play in winxp for now, maybe the numbers get more random in xp, who knows ?!!!

megane
Art Contributor
Posts: 410
Joined: October 30th, 2006, 4:55 am
Location: The Big Ö (a.k.a. Austria)

Re: Random numbers in attacks... and idea

Post by megane »

zelurker wrote:but at least don't tell me that these numbers are fair !
Uh. Those numbers are fair. You had an unlucky turn; as everyone is constantly telling you, this is going to happen sometimes, and the results you posted aren't even that far off the average. Come on, seriously, this is just silly; if I roll a pair of dice and get snake eyes, should I be surprised? Surely, since the numbers are random, I should get exactly 7 every single time, right? Getting 2 this time clearly indicates that these dice are unfair! I guess I'll have to go find some other dice, maybe they'll be more fair. :roll:
that little girl's parents were attacked by ninjas - generic npc
hee hee! - little girl

Sangel
Moderator Emeritus
Posts: 2232
Joined: March 26th, 2004, 10:58 pm
Location: New York, New York

Re: Random numbers in attacks... and idea

Post by Sangel »

:hmm: Do we have a "random numbers are random" game tip that can show up by the main menu? If not, we should probably consider it.

Incidentally, one of my favourite RNG moments was when I had a slightly injured Mage about to level. So, I attacked a badly injured skeleton archer with him, making sure to attack from mountainous terrain. One hit was all I needed to kill him. Three hits were needed for him to kill me. Sure enough, my Mage missed with all three attacks, and the skeleton hit with all three. This was back before the Damage Calculations screen, but the odds of that were pretty low. Dead Mage, happy Skelly. :D
"Pure logic is the ruin of the spirit." - Antoine de Saint-Exupéry

User avatar
Dovolente
Posts: 140
Joined: April 17th, 2007, 9:02 pm
Location: Southeastern US
Contact:

Re: Random numbers in attacks... and idea

Post by Dovolente »

I remember when someone made the add-on to do away with randomness, but I really didn't like how pre-calculated the battle was. There was no chance of the enemy missing 3 magical strikes on a unit with 5 HP, and the outcome of each battle could be predicted quite easily. And judging by the number of people who whined and complained about luck, look where that guy's work got him. Nobody was interested in playing after awhile, and it just died into obliviousness.
I find it a little ridiculous when people bring up Sauron's Wesnoth mod up to in order to essentially say: "People don't *really* want the option to play with less randomness--see? See? SEE?"

As if making a new and distinct version of the game (it wasn't just an add-on, but work with new source code, right?), one that would compete successfully with established Wesnoth and the momentum of its community of developers and players, one that would require coding skill and someone(s) dedicated enough to the game to maintain the code, continue its development, provide servers--as if all that is the necessary mark of a good idea, or shows what options players would like.

Right. :(

User avatar
Chris NS
Posts: 475
Joined: May 6th, 2006, 3:22 pm
Location: Where the Queen lives

Re: Random numbers in attacks... and idea

Post by Chris NS »

Code: Select all

(it wasn't just an add-on, but work with new source code, right?)
More accurately, a small modification of the existing source code. Which of course anyone can legally do as it's under GPL.

Code: Select all

one that would compete successfully with established Wesnoth
Why not? If that's what people prefer, that's what they'll switch to. Openoffice and Firefox succeeded against established brands under far more skewed conditions.

Code: Select all

one that would require coding skill and someone(s) dedicated enough to the game to maintain the code, continue its development, 
Simple matter of applying the code modification every time a new version of Wesnoth comes out. Which shouldn't be much more work than the new binaries of Wesnoth people produce every time a new version is released.

Code: Select all

provide servers
That's what the wesnoth_server package is for.

Stop moaning. It can be done quite easily if someone wants to do it. You never know, if you ask the devs nicely, they might tell you which bit of source code to change.

User avatar
Dovolente
Posts: 140
Joined: April 17th, 2007, 9:02 pm
Location: Southeastern US
Contact:

Re: Random numbers in attacks... and idea

Post by Dovolente »

Moaning? Hmmm...

*BRAAAAAINS. I WAAAAAANT TO EAAAAAAAT YOURRRR BRAAAAAAAINS.* :)

Seriously though, you are really downplaying of the knowledge, skill, time and effort that developing and maintaining a separate Wesnoth variant would entail. Perhaps it would be cake to you, but it's pie-in-the-sky to me (and, I'd guess, to most Wesnoth players). If a dev pointed pointed me to the spot in the code to change, it would do ZERO good; I strain neurons--good, innocent neurons--just to rehash other people's WML. :) Do you really think most Wesnoth players could jump into C++, start releasing builds for different OSes, maintain a separate MP server, & etc?

So I stand by my point: it's ridiculous to suggest that because people don't (as most of us *can't*) build a separate variant, we aren't interested in a less-random MP option. And yes, I know, it isn't going to happen. It's like my vote for Ron Paul: he wasn't going to win the republican nomination, but he rocks so I voted for him anyway.

Please don't say that I should run for president myself... :P

megane
Art Contributor
Posts: 410
Joined: October 30th, 2006, 4:55 am
Location: The Big Ö (a.k.a. Austria)

Re: Random numbers in attacks... and idea

Post by megane »

Dovolente wrote:So I stand by my point: it's ridiculous to suggest that because people don't (as most of us *can't*) build a separate variant, we aren't interested in a less-random MP option. And yes, I know, it isn't going to happen. It's like my vote for Ron Paul: he wasn't going to win the republican nomination, but he rocks so I voted for him anyway.
I think you're misunderstanding. How hard it would be to make such a mod is irrelevant, because someone did, in fact, make it. What is important is that, though he had put in the hard work required to do so, very few people actually played it, because those who complain about the RNG, as far as I can tell, simply don't understand what they're asking for; Sauron's mod said, "here, this is what you demanded, aren't you happy?" and they replied with an overwhelming, "meh." At which point he decided it was pointless to continue.

In fact, Dave recently allowed me to do an accuracy mod, which you can play in trunk. It's certainly enjoyable, and, again, basically exactly what you guys say you want, but whether it'll be popular enough to warrant my continued attention remains to be seen.
that little girl's parents were attacked by ninjas - generic npc
hee hee! - little girl

Sangel
Moderator Emeritus
Posts: 2232
Joined: March 26th, 2004, 10:58 pm
Location: New York, New York

Re: Random numbers in attacks... and idea

Post by Sangel »

I think you're both right. It is a big effort to code and maintain a Deterministic Wesnoth Variant. Not unconquerable, as has been demonstrated, but keeping up with "normal" Wesnoth would take time and effort. On the other hand, Sauron attempted it once and it never took off. I suspect that this is a sign that those who want a less-random Wesnoth are significantly outnumbered by those who don't.

It's really about effort vs reward. Dovolente's suggesting that it's a lot of effort in code and maintenance. Megane's suggesting that the rewards is low because the result wouldn't be hugely popular. These are compatible statements, and I believe they're both correct - and when taken together provide the best explanation of why we're not likely to see a Deterministic Wesnoth Variant take off.
"Pure logic is the ruin of the spirit." - Antoine de Saint-Exupéry

freddf
Posts: 17
Joined: March 20th, 2007, 9:03 am

Re: Random numbers in attacks... and idea

Post by freddf »

megane wrote: In fact, Dave recently allowed me to do an accuracy mod, which you can play in trunk. It's certainly enjoyable, and, again, basically exactly what you guys say you want, but whether it'll be popular enough to warrant my continued attention remains to be seen.
Can you give some more details about what exactly this mod do?

About the Sauron's mod: actually people and the author were interested, but the RL precluded him from continuing (last post on this page http://forum.wesnoth.org/viewtopic.php? ... a&start=15 )

megane
Art Contributor
Posts: 410
Joined: October 30th, 2006, 4:55 am
Location: The Big Ö (a.k.a. Austria)

Re: Random numbers in attacks... and idea

Post by megane »

freddf wrote:Can you give some more details about what exactly this mod do?
http://forum.wesnoth.org/viewtopic.php?f=8&t=20997
that little girl's parents were attacked by ninjas - generic npc
hee hee! - little girl

freddf
Posts: 17
Joined: March 20th, 2007, 9:03 am

Re: Random numbers in attacks... and idea

Post by freddf »

Thanks for the link, megane, interesting reading indeed, and coming from the top, none the less. :wink:
I wander if it would be possible to use to create a HoMM-like attacks like described in this post pt1 http://www.wesnoth.org/forum/viewtopic. ... 38#p293038
or even better, a skewed probability distributions as units levels up.

Post Reply