Scores

[Derakon]

Sure, let's have this discussion [again]. How should score be calculated? I believe right now it's simply based on player experience, which I think we all can agree is kind of silly.

Here's my suggestion: you should receive points when you do things that are difficult, when you win the game, and for making progress. Say, for example:

* You gain a point each time you reach a new depth for the first time.
* When you kill a unique, you gain a point if the unique is significantly out of depth for your character. Exact formula to be determined later.
* When you win, you gain 100 points.

That list is a little thin, so suggestions are welcome for accomplishments that are a) difficult, b) non-repeatable (no grinding for score!), c) not dependent on race/class/details of level generation, and d) reasonably easy to measure. Oh, and the reason for the "progression" score is to give players who don't win something to work towards. Obviously that portion of the score is going to be relatively easy for veterans to rack up, but they should be planning on winning anyway.

Then we can apply multipliers to your score based on the constraints you were operating under:

* Forced descent multiplies your score by 1.1
* No-artifacts multiplies your score by 1.3
* Playing as a half-troll multiplies your score by .9
* Playing as a mage multiplies your score by 1.15
* And so on and so forth.

The ideal goal is that your highest-scoring characters should be the ones that you made the most progress with under the most difficult conditions. One nice side-benefit of this is that we can display the score multipliers in-game and thus guide new players about what "game modes" are easy vs. difficult. We can also consider doing away with experience penalties (maybe keep humans as leveling relatively quickly, and then move everyone else to the same experience rate).

As a personal preference, I would like to keep the highest scores under 1,000. When you have huge numbers, it becomes difficult to see what impact any one accomplishment had on your score.

[Monkey Face]

Right now my Hobbit Priest is on level 93 with 7,182,356 experience, so I'm thinking I like the score based on experience :-).

It could be worse. The original rogue scores were based on gold.

[yyt16384]

* When you kill a unique, you gain a point if the unique is significantly out of depth for your character. Exact formula to be determined later.

How do you determine the "for your character" part? Character level is not good for this because it is not the main factor of character power after the early game.

And some other ideas:

* Add character level to score. This is better than experience because it stays mostly smooth for the whole game.
* For winning characters the turn count should be taken into account. I'm not sure what the formula should be. Probably something like a_big_number/turn_count will work.

[Ingwe Ingweron]

* For winning characters the turn count should be taken into account. I'm not sure what the formula should be. Probably something like a_big_number/turn_count will work.

How about Experience/Turncount? This is what it is for comps (although for non-winner comps, comps where there is a winner are usually determined solely by turncount).

[Rydel]

I agree the turn count should affect the score at winning - you should get more points for winning with a low turn turn and less with a high one, factoring in your class.

You could possibly also aware bonus points for winning with what NetHack dubs "Conducts" - doing or not doing certain things that the game doesn't enforce, but can measure. For instance, completing the game without drinking any potions, or without killing uniques other than Sauron and Morgoth

The score for killing a unique could also vary based on the unique, with monsters that you should never deal with at-depth, like Kavlax or the Terrasque, being worth more.

[Derakon]

How do you determine the "for your character" part? Character level is not good for this because it is not the main factor of character power after the early game.

It is inevitable that lucky characters will have higher scores than unlucky ones because they have higher power at lower levels. I can't think of a reasonable way to account for that; it's a "problem" for the competitions as well. I'd just make a formula that uses the character's level and the monster's level as its only inputs.

* Add character level to score. This is better than experience because it stays mostly smooth for the whole game.

On the other hand, we shouldn't encourage players to grind to level 50 just for the points. That's one reason to prefer depth achieved over character level achieved.

* For winning characters the turn count should be taken into account. I'm not sure what the formula should be. Probably something like a_big_number/turn_count will work.

How about Experience/Turncount? This is what it is for comps (although for non-winner comps, comps where there is a winner are usually determined solely by turncount).

I worry a little bit that this would be "enforcing" speedrunning of the game as the only way to get good scores. Should someone like Timo who takes his time but is nonetheless a good player be unable to achieve as high of a score as someone who relentlessly throws characters into the woodchipper until they get a supremely lucky setup? I know turn count is used for competitions, but I think that's mostly because it's one of the few measurements that's reliably available regardless of variant and competition conduct.

If we were to have a turncount-based score bonus, I'd prefer it be a flat bonus for coming in under some specific number of game turns, like 100k. Then the bonus basically is your "achievement" for getting a reasonably fast game.

You could possibly also aware bonus points for winning with what NetHack dubs "Conducts" - doing or not doing certain things that the game doesn't enforce, but can measure. For instance, completing the game without drinking any potions, or without killing uniques other than Sauron and Morgoth

The score for killing a unique could also vary based on the unique, with monsters that you should never deal with at-depth, like Kavlax or the Terrasque, being worth more.

If we can figure out an objective way to score the unique's "challenge rating" independent of their native depth, then I'm all in favor of having modifiers based on how nasty the unique is. I'd be fine with conducts as well -- those would basically be similar to the ironman rules, just not actively enforced by the game.

[AnonymousHero]

I'm surprised anyone cares about score since this is essentially a single-player game. (Modulo competitions, but those usually have their own criteria anyway.)

That said... how about just recording some stats and letting the player decide how to sort the hall of fame? Some candidates for stats: Turn count, OoDness of uniques killed, real time played, monsters killed, races killed out of all of the races (that is, 100% would be "you've killed at least one of every type of monster), number of turns without rBase, etc. etc.

[Carnivean]

Should someone like Timo who takes his time but is nonetheless a good player be unable to achieve as high of a score as someone who relentlessly throws characters into the woodchipper until they get a supremely lucky setup?

Timo is about control of the game, therefore it is likely that he suffers less damage than other players. How about damage taken being negative to total score somehow?

[Ingwe Ingweron]

I worry a little bit that this would be "enforcing" speedrunning of the game as the only way to get good scores. Should someone like Timo who takes his time but is nonetheless a good player be unable to achieve as high of a score as someone who relentlessly throws characters into the woodchipper until they get a supremely lucky setup? I know turn count is used for competitions, but I think that's mostly because it's one of the few measurements that's reliably available regardless of variant and competition conduct.

If we were to have a turncount-based score bonus, I'd prefer it be a flat bonus for coming in under some specific number of game turns, like 100k. Then the bonus basically is your "achievement" for getting a reasonably fast game.

Good points. Plus, in a comp, every player is starting with the same setup, stats, race and class. If it were purely turncount/experience on the ladder across every race/class combination, it wouldn't make as much sense. A fragile race mage must play more slowly than a robust race paladin, for example.

In fact, without some way to make an equivalent comparison for each race/class combination against any other race/class combination, all other things being equal, then the whole idea of score is a lot more tenuous.

[Nick]

I'm surprised anyone cares about score since this is essentially a single-player game.

Indeed.

I used to look at my scores when I first started playing, but I barely notice them now. I would suggest that the main use of scores is to compare one's own characters' progress, and mainly for newish players. So maybe something simple like character level + max dungeon level + number of different monsters killed.

[Egavactip]

Meaningful scores are desirable to measure one's own accomplishments against one's earlier accomplishments or those of others. So I am definitely in favor of score reform.

Leaving aside actual victory, I think score should reflect the ability to achieve as much as possible in as little time as possible.

So let's say winning gives you a score of 100,000 base.

To that, there should be negative modifiers for every X turns past Y (modified further as need be for race/class combinations). This may also sufficiently take into account the need to penalize non-humans accordingly; if not, then there should be an additional penalty for that.

There should be positive modifiers for each unique defeated, modified still further if defeated at a dungeon level higher up than where they normally reside or by something like character level vs unique dungeon level.

There should be smaller & various positive modifiers for each defeat of certain other designated monsters or monster combinations.

There should be small positive modifiers for reaching milestones of 1,000,000gp, 2,000,000gp, 5,000,000 or 10,000,000gp at the time Morgoth is defeated.

There should be a tiny penalty for killing Farmer Maggot or Santa Claus.

Other than that, I don't think there need to be other modifiers.

[krugar]

I'm personally not very interested on scores of any kind in a game where a character success is dictated by so many distinct factors, with a great prevalence on randomization.

I would prefer a table of statistics.

Character Name, Race, Class
Character Level achieved
Depth Achieved
Total Number of Turns Used
Win (Yes,No)
No. Uniques killed
No. of Artifacts/Randarts found
Total Number of Monsters Killed

Score shouldn't really be an important statistic. Frankly no formula you can come up with with be entirely satisfying and will most certainly exclude some corner cases, resulting in constant debates on the forums. Meanwhile I would be more interested in those statistics above and a column sorting feature. That would allow me to compare characters on real data, not on some reductionist mathematical hash we tend to call Score.

[Ingwe Ingweron]

An algorithm to make race/class combinations comparable with other race/class combinations is way above my pay-grade, but without such an algorithm I doubt "score" will be very helpful for comparing characters, even one's own characters.

[Derakon]

I'm personally not very interested on scores of any kind in a game where a character success is dictated by so many distinct factors, with a great prevalence on randomization.

I would prefer a table of statistics.

Character Name, Race, Class
Character Level achieved
Depth Achieved
Total Number of Turns Used
Win (Yes,No)
No. Uniques killed
No. of Artifacts/Randarts found
Total Number of Monsters Killed

Y'know what? I like this a lot. Let's just collect a ton of statistics on each game and then show the stats to the player (not just at end-of-game, but also during the game, of course).

[Galen]

As a new player, I feel that 'the deepest dungeon level reached' is already a perfectly fine score.

The goal of the game is to descend Angband and kill Sauron and Morgoth... if a score is necessary, how about that: score = deepest dungeon level seen +1 for killing Sauron + 1 for killing Morgoth.

But (I think ) players don't need a computer to calculate this score for them And veteran players don't need a score to tell how well their or someone else's character is doing.

I like the table of statistics idea. I actually think new players would benefit more from statistics like total number of common/unique monsters generated/detected/killed, number of 'important' items (artefacts, stat potions, speed items etc) generated/found, than from a score.

Another idea for statistics is 'monster memory completion'. After you learned everything about every single monster in Angband, it's 100% complete.