should your high score in a game reflect the difficulty level you're playing at and your starting point in the game (IE easier vs harder starting points) ?

or should it just be a score, with the additional info of "i chose starting scenario X, and am playing at difficulty Y", or something like "i have a current score of almost 3 million with the sole survivor startup at 200% difficulty" ?

Depends :)

Generally the more casual the game the more I would lean towards "separate score boards for each difficulty level" while more hardcore & skillbased games (like roguelikes) work better with "score multiplier for difficulty level".

I agree, i think it would keep it simple to just have separate score boards for each level of difficulty. It'd also be easier for the player to look at when they're after feedback on their scores etc.

unfortunately, there are no simple difficulty levels.

difficulty is controlled by setting player hit points, rate of healing, monster encounter chances, and NPC encounter chances.

each of those can be set from 0% to as high as you want (within the limits of overflow).

so you might play with hit points at 200%, heal time at 10%, animal encounter rate at 100%, and NPC encounter rate at 150%.

and you can change the difficulty at any time.

so a player could turn difficulty way down, get a lot of stuff, then turn difficulty way up, thereby automatically increasing their survival score if difficulty is factored into the score.

an alternate would be to accumulate score, as opposed to calculate on the fly based on current game state. IE you get points when you acquire a stone knife, based on current difficulty. but when you lose that knife - you don't know how many points it was worth originally, unless you store that info. the difficulty might have changed. so you don't know how much to reduce the score.

i suspect it'll need to be on the players to not cheat. the game comes with all testing cheats enabled, so anyone can cheat their way to a high survival score.

If you think people are going to care about their score compared to others, the score would have to be modified to fit the difficulty. Otherwise, the highest score would go to someone playing on an easier difficulty, slowly grinding their way through easy stuff to get a high score.

I put score multipliers and other modification in my game Arcane Castle. It has a wide range of difficulty that adjust as you win or lose, but no difficulty settings. In endless mode, difficulty increases as you go, and starts at a difficulty relative to the last time you played. I took some time figuring out how to give out points so if you start endless mode at a high difficulty, and survive for a little while, you get all the points you would have if you got that far from a low difficulty. This way, you always get more points for playing at the highest difficulty you can handle.

Then again, I'm not sure how much people care about high scores in games.

Then again, I'm not sure how much people care about high scores in games.

well this is an open world sandbox survival game. the idea of a "survival score" or rating was a way for the player to judge their "progress". although progress is somewhat of a misnomer. the world constantly changes, stuff wears out, skills get rusty. so its more of a rating of how you're doing at the moment which will go up and down over time, as opposed to a traditional "score" which usually tends to go up, or only goes up.

then there's the whole question of how much each difficulty setting should affect the base score. its not a straightforward linear relationship like in many games. its not simply level 2 difficulty = level 1 difficulty with 2x hp and 2x exp.

does healing twice as fast make the overall game twice as easy?

does twice as many animal encounters make it twice as hard?

i think for caveman it will have to be just a base score given in the context of having played with a certain set of difficulty settings.

Given the number of options that you can tune on the fly, I think single high score table would be the best fit. Although a lot of playtesting would be necessary to tune exact multipliers of scoring. Some players will still try to game the system a bit (for example both -10% health and +10% monster spawn rate give the same bonus but health is easier to overcome so lets choose this) but it may not be that much of issue as better players will anyway get both :)

so a player could turn difficulty way down, get a lot of stuff, then turn difficulty way up, thereby automatically increasing their survival score if difficulty is factored into the score.

You may set the score multiplier to the easiest setting player ever set. So in your example it would make the game just harder for player without gaining more points.

As your game is survival so I think player is expected to restart a few times anyway.

Also I assume your game is not "infinite" and gets harder and harder over time to the point of killing the player anyway regardless of initial difficulty so the scoring makes any sense.

IMO a high score is pretty useless if its based on an algorithm that arbitrarily combines multiple statistics heavily influenced by randomness and arbitrary starting conditions.

Even if starting conditions were same, assuming this is a sandbox game, success is ultimately subjective, and thus its impossible for you to create a scoring algorithm.

Therefore I suggest just make a big page that displays a large variety of statistics of the playthrough. (days survived, grass stomped, fish eaten...) Players can then look at the statistics they care about for comparison.

If a competitive community happens to form around the game, they can come up with their own scoring methods.

I also suggest you have a few "default" configurations for the starting conditions and settings. If people can just freely change the settings it will be very difficult to not feel like youre not getting the intended experience, or are "cheating" by using too easy settings. Youll also make it easier for the people who want to compare scores, they can just say they played "hard mode" instead of providing the exact configuration they were using.

You may set the score multiplier to the easiest setting player ever set.

yes, that would truly reflect difficulty settings...

As your game is survival so I think player is expected to restart a few times anyway.

i've been playtesting the long term playtest game over the last few days. 10 band members with huts and storage pits and bows. they've been alive for almost one game year now. over the last couple days of testing, about 4 or five game days has passed, during which time at least one band member died on about a dozen occasions. on two or three of those occasions, the entire band was wiped out in a single encounter. so, yes you die a lot. typically 2-3 times or more in an evening's play for an experienced player. but you can save a new game with automatic file naming with a single hotkey at any time. it saves the entire game world state, right down to missiles in mid-flight.

Also I assume your game is not "infinite" and gets harder and harder over time to the point of killing the player anyway regardless of initial difficulty so the scoring makes any sense.

incorrect. this is a simulation, not a game. so the game does not get harder as you become more powerful. becoming more powerful means you can take on and survive greater challenges. for example, the playtest band (Grog's band) is only now getting to the point where they can take on big game. They got their first giant sloth kill ever yesterday.

http://www.gamedev.net/gallery/album/946-caveman-the-adventures-of-grog/

when you first start out you can't really take on anything. then you can only take on small game with basic weapons. then you can take on bigger game and even survive predator and hostile caveman attacks. and then you can take on large game, and engage in inter-band combat. in skyrim terms its would be like having to run from everything except level 1 skeevers at level1, and finally being able to take on multiple level 100 dragons at level 100+. but the world is full of all monster types, from level 1 skeevers to level 100 dragons, at all times. so as you get stronger, you die less and have to run away less often. but the threats in the world remain at the same difficulty.

however - the game is based on tabletop classic D&D, and tabletop traveller. one of the D&D influences is "number appearing adjusted for party strength". so in that respect, as you get more band members and followers and pets and weapons and armor, the game will become more difficult. but there's no leveling of monster stats, such as fighting level 1 skeevers when you're first level in skyrim, then fighting level 10 skeevers when you're level 10. hit points are basically a function of body weight. weapon damage is all balanced to real world conditions as to what it takes to kill a rabbit, deer, human, saber tooth, mammoth, etc. and there are some real monsters in the game too. besides large mammoths, saber tooths, etc, there's an 18' giant lizard, and a 20 foot boa (yet to be implemented). it was definitely a dangerous place back then. and like traveller, hit points never go up. and there's no magic books to raise your stats. having just a handful of hit points seems to make it all so much more intense. combat can only be described as fast and furious. perhaps more intense than any other game that comes to mind. this wasn't necessarily expected or intended. i think it may be what some would call "emergent behavior", but its not really, its just that when you build a complex sim, and try to make each part accurate (a week was spent on the physics model for projectile flight), eventually all the bits start coming together and working in an almost synergistic manner. which typically leads to a few of these "emergent behavior" things per project. but its not really what i would call "emergent". self programming AI would be emergent.

as for infinite - yes and no. the world is 2500 miles x 2500 miles in size, so its not infinite in size. as for time - the game ends when your last band member dies. band members are player characters the human player can control. the human player can tab through band members to control any one they want, others are controlled by AI when the player is not controlling them. sort of like having multiple followers in skyrim, all with full stats just like the player's character, and being able to tab between them and control any one you want, with AI taking over the others. you can recruit new band members at any time (i just kicked the max band members up to 20). the game models ageing due to background radiation, so band members get old and die of old age if nothing else gets them first. i've figured out the rules for courtship, mating, and offspring, but have yet to implement them - lost of new models and animations required. so while individual band mebers may come and go, the band can go on for generations - infinitely long up to the limits of overflow of game time/date based variables - of which i don't think there are any. nothing based on a full date that could overflow. just stuff like current second, minute, hour and day from the game clock. so it would blow up when day overflowed at 4 billion or whatever (signed 32 bit int - what is that 2 billion?). and it might not really blow up then, just congratulate you every year for being alive for some negative number of years. lets just say behavior when game day overflows to a negative value is undefined. <g>

IMO a high score is pretty useless if its based on an algorithm that arbitrarily combines multiple statistics heavily influenced by randomness and arbitrary starting conditions.

i think "high score" carries the wrong connotations - "current score" probably conveys the concept better.

weighing the value of stats to determine the score will be difficult. all inventory objects have a quality and value which can be used. large world objects (huts, rafts, bedding, temp shelters, landmarks, etc) do not, but everything in the game can be crafted, so anything can be assigned a value based on tools, parts, skills, and time required. but then comes the less straight forward stuff... how much is a band member worth? a companion? a pet? knowledge of a nearby water source? having a cave or rock shelter under your control? how much is having an alliance with another band worth? the worth of band members, companions, and pets can be based on stats/skills. the value of knowledge of resources and control of natural shelters is a bit more fuzzy.

there will be no randomness involved at all, just look at all the current stats and such and add up a current score.

right now there's only one starting condition - random location - no gear - no skills - nothing. multiple starting scenarios are planned. easier starting conditions would incur a penalty to "current survival score". alternate starts will in effect give you additional gear and skills - making the penalty easy to calculate - make it same as the bonus they get from having that gear and those skills.

that just leaves the difficulty settings as the only remaining "arbitrary starting condition". so it seems to boil down to diff lvl vs current score.

Even if starting conditions were same, assuming this is a sandbox game, success is ultimately subjective, and thus its impossible for you to create a scoring algorithm.

it shouldn't be too hard to come up with an overall "how powerful are you right now" score. but as you say, how much different things should influence that score will have to be a subjective judgement call at best, as opposed to an objective decision. despite that, with a reasonable weighting of the influences, it should at least provide a quick snapshot for the player - even though they may not entriely agree with the weight distributions. i was thinking that a breakdown of how many points for what might address that issue somewhat. then you could see the weights and take the score in that context - "well i only have a score of 900 some odd, but it doesn't give me a lot of credit for" - hmm... it would have to be something subjective, so it cant be an object, stat or skill, the game already defines the relative value of those. that leaves stuff like caves, rock shelters, and knowledge of nearby resources. so "my score is only 900, but it doesn't give me credit for known resources farther than 3 days travel" - the player may think it should be 5 days travel or something. also, i think that by not playing favoriets with respect to the vale of one object or skill to another, especially with respect to combat vs not combat things, will allow one to sandbox away without undue influence on score one way or the other. in the end, all player activities (if successful) will result in an increase in possessions, exp, or temporary stats (sleep, food, water, god relations, etc). everything has a cost in time, skills, tools, and parts. all parts and tools have a basic monetary trading value (trinkets are the "money" in the game). the value of skills can be expressed in terms of the cost in time, tools, parts, and any prerequisite skills. skills with no prerequisites can be express in terms of just the time and parts and tools cost. so just like the ral world, there are only two fundamental commodities: time, and money (trinkets in caveman). so time would be the only real subjective value there. if a skill requires 2 trinkets worth of stuff and four hours to learn, and 1 current survival score point = 1 trinket, how many points is that four hours worth?

Therefore I suggest just make a big page that displays a large variety of statistics of the playthrough. (days survived, grass stomped, fish eaten...) Players can then look at the statistics they care about for comparison.

got this, bit its just small for now: cavemen killed, animals killed, days alive, that's about it so far. full blown stats (number of painted hides crafted, etc) are planned. unlike the sims, bladder is not modeled, so no "dumps taken" stat. <g>.

If a competitive community happens to form around the game, they can come up with their own scoring methods.

yeah - i'm doing this more for the single player's benefit. online stuff wasn't an influence in the idea, although it might fit well. that's why i'm not really looking at this from an online stuff point of view, more just for personal use only - at least to start.

I also suggest you have a few "default" configurations for the starting conditions and settings. If people can just freely change the settings it will be very difficult to not feel like youre not getting the intended experience, or are "cheating" by using too easy settings. Youll also make it easier for the people who want to compare scores, they can just say they played "hard mode" instead of providing the exact configuration they were using.

yeah, the jury is still out on that one. its looking like straight 100% across the board, with either 1% or 5% or 10% for heal time, as the default difficulty setting. maybe 200% hit points, but i'm not so sure about that. 1% heal rate keeps you in the game without too much downtime, just enough. at 100% it takes 6 months to heal up from almost dead. so the sweet spot will probably be somewhere between 1% and 10%. i wanted to keep away from a few preset difficulty levels to avoid the issue of one level too easy and the next too hard: silent hunter 4, top two diff levels. if you could toggle different stuff off you could custom tailor the difficulty like in red baron 2 and other flight sims. BTW, toggling settings affects realism % which affects campaign points earned in red baron 2. so they chose to have difficulty affect score - experience points, actually,