Points system for completing levels

[Fleech]

However, it would appear that the formula you proposed would generate mostly negative scores

I've done a little testing, and I think you may have read the formula wrong :winktounge:

As far as I can see the lowest save requirement on any level is 10%. Even if you took 9 minutes to complete it you'd still need to use at least 47 skills to go below 0 points;

(10*100) - (47*10) - (540) = -10

Unless of course you're referring to attempts where you fail and only save say, 1/80, in which case an automatic zero for failing would solve that.

If you applied scoring to custom levels though, then I guess there might be a problem in some extreme cases. But the normal levels in both Lemmings and ONML all work fine.

It does seem disproportionate giving out points for lemmings which have to be saved though...Would it be possible to just use the fraction extra?

The trouble with this is the same as with skills and time - many early levels have stupidly low save requirements (even some Mayhem levels have requirements that are far too low - you can kill up to 20 in The Fast Food Kitchen for example...).

On the other hand there are levels that require a high %, but can be done 100% (e.g. Taxing 1). You'd recieve very few points on levels such as these if you only looked at lemmings saved over the required amount. Sure, it'd be the same for everyone playing the level - and the points for saving required lems are worthless - but IMO you'd run into more problems by removing these points than by including them.



EDIT: didn't see your post Shvegait 

[covox]

... why didn't it occur to me 11 hours ago that the percentage could be a number from 1-100 rather than a decimal  :mikeXD: That's what happens when you study engineering for too many evenings in a row, erodes your common sense.

Maybe some sort of common benchmark is needed to compare how "nice" scores look. Ideally you shouldn't be able to pass a level while gaining a negative/zero score, but there should be a reasonable spread between low end (i.e. barely passing) and high end (i.e. master player).

I think SgB's subtractive theory is looking good at the moment, all it needs really is some sort of foolproof mechanism of avoiding negative/zero scores -at any cost-. (bonus points if it doesn't involve adding another number)

[Proxima]

I think SgB's subtractive theory is looking good at the moment, all it needs really is some sort of foolproof mechanism of avoiding negative/zero scores -at any cost-. (bonus points if it doesn't involve adding another number)

The only levels with a 10% save requirement are 1, 2, 3 and 5 Fun and Cascade. Of these, 5 Fun has the most skills (50) and it has five minutes. The lowest possible score would be 1000 - 500 - 300 = 200.

The only level on which negative scores would be possible is 4 Tricky: 20% required, 320 skills. But let's face it, realistically if you pass that level at all there's no way you actually would save only 20%. I suggest you simply have a default score -- perhaps 200, as the lowest possible on any other level, would be appropriate, or maybe 1000 -- and if you pass a level but score less than the default you get the default instead.

[ccexplore]

The only level on which negative scores would be possible is 4 Tricky: 20% required, 320 skills. But let's face it, realistically if you pass that level at all there's no way you actually would save only 20%. I suggest you simply have a default score -- perhaps 200, as the lowest possible on any other level, would be appropriate, or maybe 1000 -- and if you pass a level but score less than the default you get the default instead.

Or, we can simply bump up the weight for number saved so that instead of 100:10:1, it's something like 200:10:1 (leading to 40000 - 32000 - 480).̆ This also makes it even more unlikely that someone can tradeoff number of lemmings saved with skills used or time (not that it's a realistic option for most levels anyway).

[Proxima]

Hmm, that raises a thought. What is the current record for biggest difference between number of skills needed to save 100% and number of skills needed to save all but one? My first thought is Mary Poppins' Land but I don't know the figures.

The other alternative would be to decrease the points for number of skills; I already said I think the "1 skill = 10 seconds" equation overvalues skills somewhat.

[ccexplore]

Hmm, that raises a thought. What is the current record for biggest difference between number of skills needed to save 100% and number of skills needed to save all but one? My first thought is Mary Poppins' Land but I don't know the figures.

I have to do some calculations, but levels worth examing in this regard include:

- The Boiler Room (Mayhem 2)  [especially since the 100% also takes up more time, in order to lump the lemmings in the crowd very close together]
- Turn around Young Lemmings!  (Tricky 22)

Still, with a 20:1 ratio (under my proposal) in the number of lemmings saved vs. skills trade-off, and the fact that all-but-1 entails a minimum of 2 percentage points in % saved (79/80 -> 98%), we are talking about requiring 41 less skills used (or some significant amount of saved time) to make up the loss of one lemming.

[Proxima]

Still, with a 20:1 ratio (under my proposal) in the number of lemmings saved vs. skills trade-off, and the fact that all-but-1 entails a minimum of 2 percentage points in % saved (79/80 -> 98%), we are talking about requiring 41 less skills used (or some significant amount of saved time) to make up the loss of one lemming.

That's true. Still, it would be interesting to know from the theoretical point of view of "how many skills is saving one extra lemming worth?"

[LemSteven]

That's true. Still, it would be interesting to know from the theoretical point of view of "how many skills is saving one extra lemming worth?"

That exactly was what I was considering when I presented my idea for a points system (which, incidentally, also has a 200:10:1 ratio).  Ideally, I would want to make a 100% solution score more than a 98% solution, regardless of how many extra skills or time is used; however, this would require an enormous save bonus which would effectively render the other bonuses insignificant.

As for the skills to time ratio, I wouldn't object to lowering the value of skills.  As long as skills are worth something meaningful, I don't really care.

[Fleech]

The other alternative would be to decrease the points for number of skills; I already said I think the "1 skill = 10 seconds" equation overvalues skills somewhat.

That all depends on how covox wants the scoring system to work and how a level performance should be judged. My idea was based firstly on lemmings saved obviously, and if that's equal then skills, if still equal then the completion time is compared. In this sense 10:1 might not even be big enough.

A smaller ratio should ensure a little more variety and force you to be creative,  but how much time would you suggest 1 skill is worth? It's much less cut-and-dried than the first option, and you'd need to know the workings of the formula in order to decide how to maximise your score.

I prefer the simpler option myself - since both will still result in high scores being determined by time taken once the 'best' solution has been found anyway - but I can see why the other might be considered better.

(LemSteven, you posted just after I'd finished typing that. Don't think I'm ignoring you :tongue:)

[covox]

Another idea about how to handle the skills problem: how about a logarithmic scale for determining the skills amount?

Normally with base-10 logs the value increases with the number of zeros ( so log(100) = 2  => 10^2 = 100 ,  log(1) = 0 => 10^0 = 1). However, the idea here would be to change the 10 to give us a reasonable range.

Let's pick an arbitrary skill threshold for "just too silly". It's highly unlikely that someone will use more than 40 skills in a level, so when we use 40 skills we want the penalty to be... 8.

x^8 = 40  => x = 1.585833

So the final formula will be: log([skills used])/log(1.585833)  (also known as the base-1.585833 logarithm of [skills used])

Under this system, using:
1 skill = 0
2 skills = 1.50
5 skills =  3.49
10 skills = 4.99
20 skills = 6.50
40 skills = 8.00

This value could then be used with another multiplier, perhaps 50 or something.

[Fleech]

I don't really think this would help to be honest. The main problem now seems to be deciding on the Saved:Skills:Time ratio, and using something like this would make it impossible to set because it would be changing every time you used a skill.

Even if this were to be used, it would over value skills on levels with few of them and undervalue them on levels where you have to use lots.

2 skills = (1.5*50) = 75
40 skills = (8*50) = 400

In theory that makes sense I suppose, but saving two builders on King Of The Castle is surely at least as valuable as saving two blockers on Fun 3? :tongue:

IMO before any final formula can be made, it needs to be decided whether;

* lems saved should always be worth more than any amount of saved time or skills. I think we're all agreeing this should be the case?

* skills used should always be the second decider if lems saved is equal, with time only coming into it when skills are also equal. This is what I believe it should be, but it seems Nuntar ,and maybe others, don't agree. Until this part is decided on I don't think you're really going to get anywhere, since people will want different ratios according to how they want the scoring to work.

Perhaps we should vote :winktounge:

[covox]

A weighting of [Lemmings saved] > [skills used] > [time used] is fine by me. Are there any objections?

[geoo]

[...]
geoo89: Oh my word. :mikehuh2: Isn't seven symbolic constants a little bit over the top for a score generator?   
[...]

Probably it is...it's just caused by my overly complicated thinking. :wink:

If everyone agrees on amount of Lemmings saved a ultimate decider, then why not simply make it two separate scores with the first one being the amount of lemmings saved and second one being the tie-breaker? If you look up the challenges thread, almost all challenges are about the amount of lemmings, almost only in case there's a definite maximum there are challenges considering the amount of skills used (mostly builders), and almost never time.
Then there's just to wonder about skills and time evaluation, an maybe the evaluation between the certain skills (e.g. is saving a builder worth more than saving something else?).

[Fleech]

(e.g. is saving a builder worth more than saving something else?).

On some levels yes, on others no, which is why this sort of thing should be left alone in my opinion. Ranking the skills in order of importance can only be done in the context of a single level, not the whole game, so assigning different points for each wouldn't work.

[ccexplore]

Even if this were to be used, it would over value skills on levels with few of them and undervalue them on levels where you have to use lots.

2 skills = (1.5*50) = 75
40 skills = (8*50) = 400

In theory that makes sense I suppose, but saving two builders on King Of The Castle is surely at least as valuable as saving two blockers on Fun 3? :tongue:

Is that really relevant?  I thought we mostly agreed that it isn't very meaningful to compare a score from one level with a score from another level, since it's just too hard to have a fixed formula that can generate such comparable scores across all levels (you'd have to know, essentially, what the "perfect" score is for each level in order to calibrate).

What could be relevant though is how this logarithmic scaling of skills would compare against the time penalty (ie. it is no longer simply 1 skill = 10 seconds; the number of seconds a skill is worth [as if that's even meaningful to speak of] varies with the level]).