On Randomness and Game Longevity

[Dave]

Dave, I'm not sure why you're using D&D referance to a battle system as most computer games do not use a system based on 5% intervals.

Mostly because I wanted to illustrate how Wesnoth really isn't all that random compared to many other games, and d20-type battle systems are probably the ones people are most familiar with.

And you sound half like you want to give up on wesnoth all together and half like you want to re-invent it. Wesnoth is a good game. The stats and % are all very good, terrain and movement and traits and factions and att types....it's all good. Some people are just getting burned out from playing too much.

Actually I don't want to 'give up' on Wesnoth so much as I want to say that I think it is basically complete. There might be some technical improvements that can be made, but as a game I think it is complete. I haven't contributed to it in any serious way for a very long time, and that's because I'm happy and satisfied with it.

I don't want to 're-invent' it -- rather, I would like to see more FLOSS strategy games that use different concepts and which have different goals. That is actually my main point. Wesnoth was never designed to be all things to all people. It's designed to be one certain type of game, and it has fairly modest goals. But I wish people would develop other types of games that used different concepts. If they want to, they could re-use many aspects of Wesnoth to do this.

---

I agree with some comments that randomness doesn't play as big a factor as many suggest. Even when players are of similar skill levels, one or the other is likely to use a better strategy in any particular game. I actually think the bigger 'problem' with the randomness is simply that some people find it frustrating. But many people love it for the same reason. This is why the best solution is to simply have different games with different systems to appeal to different people.

I also completely disagree that almost anyone sees Wesnoth as being like a 'casino' and enjoys simply 'rolling the dice' and see how they come out. Some people like board games or pen-and-paper RPGs because they get to roll dice, and that is somewhat entertaining in itself (complete with people kissing the dice....making a big roll when it's an important one, and so forth). It's not as entertaining as when there is money at stake, but it's still somewhat fun. Doing it on a computer is no fun at all. I don't think anyone at all enjoys Wesnoth for that reason. They enjoy Wesnoth because they can strategically position their forces to give themselves as high a chance as possible for success.

David

[torangan]

I agree with Dave. Wesnoth is a well defined game and if it's not exactly what people want, they need to start a new project which fits their needs better. One game can't be for everyone and people like different kinds of games. For example, I just tried out Glest. Nice graphics, a few nice concepts but to me it's a boring game so I removed it again. I simply don't like RTS much, I prefer to have time to consider my moves.
I think most progress on Wesnoth should rather be on the campaign side. I'd like to see support for even more RPG like campaigns and multiplayer campaigns could be very nice as well. Of course, technical aspects like resource usage, graphics, music, sound, translations etc. should hopefully continue to improve at the rather high rate seen so far.

[Jetrel]

The fundamental problem with randomness in wesnoth is NOT the actuality of things - as Dave has pointed out, many other, very popular game systems have an equally random setup (*cough* Civilization).

The problem lies in the perception of this randomness.


The challenge for me, in making any future games, is to make a game that, based on some knowledge and intuition of human psychology, is designed to make the player feel more in control of things.

[Temuchin Khan]

The challenge for me, in making any future games, is to make a game that, based on some knowledge and intuition of human psychology, is designed to make the player feel more in control of things.

That would be hard to do. According to what I've read, it is a matter of human psychology that when we are told we have a 50-50 chance, we expect to have the equivalent of a 70-30 chance. Wishful thinking, I guess.

Anyway, under those circumstances, you will always, inevitably, have at least some people who think they are being treated unfairly by the RNG even if/when they are being treated perfectly fairly.

[Sauron]

The challenge for me, in making any future games, is to make a game that, based on some knowledge and intuition of human psychology, is designed to make the player feel more in control of things.

That would be hard to do. According to what I've read, it is a matter of human psychology that when we are told we have a 50-50 chance, we expect to have the equivalent of a 70-30 chance. Wishful thinking, I guess.

Anyway, under those circumstances, you will always, inevitably, have at least some people who think they are being treated unfairly by the RNG even if/when they are being treated perfectly fairly.

I do not agree. Your stats does not show WHEN miss occurred. You can be perfectly at EV, still miss all killing blows - result: opponent eliminated more units and grows in strength, while you're getting few in numbers and DO NOT ADVANCE! If my 2 orcish archers miss all 4 shots at ghost at 50% it is ... guess what? Bad luck? The feeling of being unlucky is not perception of 50% chance as 70% chance. It is plain mathematics - all done EX POST. I made extended stats, that is supposed to present the mentioned above problem... Still - the fight system I propose (average-based) is much better solution than any other so far proposed (even my some-damage-guaranteed one).

[Elvish_Pillager]

That would be hard to do. According to what I've read, it is a matter of human psychology that when we are told we have a 50-50 chance, we expect to have the equivalent of a 70-30 chance. Wishful thinking, I guess.

After years of playing Wesnoth, you can get a pretty accurate perception of a 50-50 chance.

[Temuchin Khan]

You can be perfectly at EV, still miss all killing blows - result: opponent eliminated more units and grows in strength, while you're getting few in numbers and DO NOT ADVANCE! If my 2 orcish archers miss all 4 shots at ghost at 50% it is ... guess what? Bad luck? The feeling of being unlucky is not perception of 50% chance as 70% chance. It is plain mathematics - all done EX POST.

Actually, it sounds to me like you're falling into precisely the error I mentioned above. When you hear "50-50" you think "hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss" and not "hit miss miss hit hit hit hit hit hit miss miss miss miss miss miss miss hit hit miss hit." When ex post facto you examine the result and see that it is not the former but the latter, you're verifying that something like the latter happens sometimes, while claiming that something like the former should be happening.

Anyway, that's what it sounds like to me.

[joking]
Maybe what we need is a "less hope" mod that renames all percentages to cancel out false expectations:

0% is 0%
1% is the new 10%
2% is the new 20%
3% is the new 30%
10% is the new 40%
20% is the new 50%
30% is the new 60%
40% is the new 70%
50% is the new 80%
60% is the new 90%
100% is 100%
[/joking]

[Dragon Master]

You can be perfectly at EV, still miss all killing blows - result: opponent eliminated more units and grows in strength, while you're getting few in numbers and DO NOT ADVANCE! If my 2 orcish archers miss all 4 shots at ghost at 50% it is ... guess what? Bad luck? The feeling of being unlucky is not perception of 50% chance as 70% chance. It is plain mathematics - all done EX POST.

Actually, it sounds to me like you're falling into precisely the error I mentioned above. When you hear "50-50" you think "hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss" and not "hit miss miss hit hit hit hit hit hit miss miss miss miss miss miss miss hit hit miss hit." When ex post facto you examine the result and see that it is not the former but the latter, you're verifying that something like the latter happens sometimes, while claiming that something like the former should be happening.

Actually, both your examples are wrong. a 50-50 chance does not garuntee that you will get an equal amount of hits and misses in a sample. With 50-50 it's possible you could get miss miss miss miss hit miss miss miss hit hit.

[Temuchin Khan]

Actually, it sounds to me like you're falling into precisely the error I mentioned above. When you hear "50-50" you think "hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss hit miss" and not "hit miss miss hit hit hit hit hit hit miss miss miss miss miss miss miss hit hit miss hit." When ex post facto you examine the result and see that it is not the former but the latter, you're verifying that something like the latter happens sometimes, while claiming that something like the former should be happening.

Actually, both your examples are wrong. a 50-50 chance does not garuntee that you will get an equal amount of hits and misses in a sample. With 50-50 it's possible you could get miss miss miss miss hit miss miss miss hit hit.

Very true.

[Vendanna]

Well I take it more from a statistics approach so four attacks at a 50% chance you get..

four hits in a row 0.5*0.5*0.5*0.5=0.0625, aka 6.25% chance
Three hits in a row 0.5*0.5*0.5=0.125 aka 12.5% chance to occur
two hits in a row 0.5*0.5= 0.25= 25% chance
1 hit = 50% chance

So yes, you have a big chance to get at least 1 hit in the opponent, but don't expect to be as lucky as to get 4 hits in a row (6.25%) if we are talking in similar ground from both units.

Also, I probably would have needed to count the chance to get a miss because the chance to get only 1 hit its probably lower than getting two hits land on the opponent, since that 50% is the chance to at least have 1 hit on the opponent

[Sauron]

Well I take it more from a statistics approach so four attacks at a 50% chance you get..

four hits in a row 0.5*0.5*0.5*0.5=0.0625, aka 6.25% chance
Three hits in a row 0.5*0.5*0.5=0.125 aka 12.5% chance to occur
two hits in a row 0.5*0.5= 0.25= 25% chance
1 hit = 50% chance

So yes, you have a big chance to get at least 1 hit in the opponent, but don't expect to be as lucky as to get 4 hits in a row (6.25%) if we are talking in similar ground from both units.

Also, I probably would have needed to count the chance to get a miss because the chance to get only 1 hit its probably lower than getting two hits land on the opponent, since that 50% is the chance to at least have 1 hit on the opponent

Actually not true, the real estimates are as follows:

Code: Select all

	p(hit)	p(miss)
hits	0,5	0,5
0	0,0625	
1	0,25	
2	0,375	
3	0,25	
4	0,0625	

so as you can see the result 2 hits is MOST PROBABLE one. That is why all who see other outcome feel they're lucky/unlucky. And the perception is correct. 4 misses are really bad luck while 4 hits are really irish luck. 1 or 3 hit(s) result is still acceptable.

And for the gameplay the result of single fight may be very often pivotal - so situation of miss miss miss miss hit hit hit hit is not acceptable even though it is still according to 50% chance. THat is why the MOMENT when the EV is reached is more important than it is admitted to me.

Now time to try my mod and see HOW the gamaplay is changed IF you are close to EV all the time. It really does not hurt.

[Vendanna]

Glad to see you taken into account the chance for missing into the equation, as I said above I was sure I was wrong for not taking it into account.

However, those are in the case both units had a 0.5 chance to impact (aka in a terrain with 50% defense) since the values in others places may vary accord to it.

And as I said above the chance to get two hits was probably higher than one at all, so after looking at your table I was right

Thank you for your time.

[Dio Cassius]

Sounds like time for a Dilbert quote: http://www.pen.k12.va.us/Div/Winchester ... random.jpg

I think Glowing Fish is right to point out there's a difference between campaigns and one off battles when it comes to how acceptible randomness is.

A narrow victory in a one off battle is still a victory, so even with some bad luck, or even the occasional small mistake, the superior general should still win.

But the way most campaigns are set-up in Wesnoth, it's not enough to narrowly win all the battles - you have to win all them by a big margin if you're going to progress. So even the smallest bit of bad luck is far more likely to derail the campaign. Lose a key character, or one of your best units, and you're dead.

So personally I find that progressing in a campaign involves a lot of going back to previous saved games. Personally it's not the randomness I dislike - it's the save/restore cycle that winds me up. Wesnoth even goes as far as enshrining the save/restore cycle in the user interface, so central is it to normal game play.

On the subject of randomness:

If I'm attacking with a 4/2 attack at 50% chance, then on average I should cause 4 points of damage. So it's possible to imagine a completely non-random system where such an attack will always cause 4 points of damage.

Such a system can be considered one end of a spectrum from the current system, but in reality you can combine different proportions of the two systems to get any point along that line between the two extremes. Every point along the line will have the same average damage, but with different amounts of randomness.

If the user was allowed to select where along this line they wanted to place this slider for a campaign game, would it make the campaigns more or less difficult, given the average damage would still be the same?

[Becephalus]

I have always thought one elegant solution would be as follows. A "bonus" system. Now obviously this owuld only solve the problems to a certain extant, but take a game like game 5 between DK and I.

I was 160 dmg under EV (very bad) and DK was 20 over (normal). Now such a game is not fun for either of us. DK consumate professional he is spent most of the match apologizing for the good luck.

Now one way to solve this would to take a straightforward % under EV calc and then add that or a fraction fo that (say 1/2) to you CTK. This way the overall EV for both sides will naturally trend to where they should be without changing the overall amount fo luck or game dynamic at all.

Obviously this doesn't fix situational luck where you are really good at wounding everything but never kill anything.

But it does help with the main problem which in my mind is the whole course of games getting ruined. I can live with occasionally getting a completely screwy result form an attack. A game where all of my attacks fail just drives me nuts though.

Anyway just an idea, not expectign many people to think its good.

EDIT: LOL AFter reading the thread it looks lik e Soliton said exactly the same thing.

[Tomsik]

EDIT: LOL AFter reading the thread it looks lik e Soliton said exactly the same thing.

Well, I did also.