a multiplayer game of parenting and civilization building
You are not logged in.
JC, I will re-run the experiment with true randomness instead of cyclical pattern.
Wish I wasn't moving today, but if I get some time I'll try to help out with some sims too. My favorite tools for these things are jsfiddle.net and desmos.com for code and math respectively since they're both permalinkable
Think you could shed a little light on some of the desired attributes of the ideal algorithm? Or even if there were a few key contrived situations with desired consequences we could focus on. The current usage of it is just for tool slots or for eve selection also? Is eve selection something that should be really strict about only the best available players getting a chance to do or would some unpredictability about who gets chosen be desired?
For these kinds of things I usually like to have a core algorithm that's as deterministic as possible, and then layer on top of that any randomness I want so that I can scale the randomness easily and still always have the "truth" to refer to, in this case that might mean trying to differentiate the 95% people from the 90% people in a way that doesn't fluctuate unless their actual ability changes, ie, unless they start consistently doing worse after a long period of consistent success. That might not be possible but it's just an example of an ideal to work towards, or maybe that's not the ideal you had in mind at all. Hard to pick the right math without knowing that.
Thaulos wrote:I would guess a quick way of making sure variance isn't too bad but while still providing a quick "startup" would be to somehow factor in the number of lives you previously played.
So maybe first life it would be a big jump, second a big but smaller one, etc. Until you cap variance at the average of your past 10 lives (as an example).
For the score to mean anything it should be hard to drop and to climb. One should only really have high scores if they are consistent across their lives.
I do like the idea of having scores be more stable overtime, so it's less up to luck to be near the top of the leaderboard. An average score for your past 10 lives might indeed work really well, so /die'ing still affects it a lot but only extreme lives will REALLY alter the score.
Thing is, how do you calculate that score?
Is it like a 'grade' out of 60 (to keep the current number)? Does getting to age 60 and all your kids surviving to 60 as well mean you get full points and /die = 0? Then what does a score of 30 mean? Also: what % of the score is just YOUR life and not your kids'? Don't want to incentivice things like avoiding kids because you can always get perfect marks on just your own life without any kids...Or maybe you start at 30 and early deaths are - and surviving is +, same as now. But if that's the case, does having 8 surviving kids w/ another 2 dying mean you'll get a higher score than having only 2 kids and 0 deaths? Should it be a % of how many of your kids survive? Then what about your own life?
What would be a good way to calculate scores to even get an average from to begin with?
yeah chess elo traditionally (but iirc not necessarily) does this too, there's an initially faster curve that settles down as you get more games in. Sounds like jason's implementation does this to a degree but dropping 10% from one death 10k births in seems too high still, unless im reading the numbers wrong.
jcwilk, the existing algorithm is indeed modeled very closely on Chess Elo rankings.
How much your score jumps after a game of Chess depends on how surprising the result is. If a grandmaster is beaten by the novice, that's a very surprising result, and the novice's score jumps a lot, and the grandmaster's score drops a lot. However, if one grandmaster beats another, that's not a huge surprise, so their scores barely change. The winner goes up a tiny bit, an the loser goes down a tiny bit. Elo has the nice property of converging over time in such a way that it can be used to estimate chances of winning against a given opponent. If we have equal Elo, we are expected to beat each other 50% of the time. A grandmaster (2800) is expected to beat a novice (700) 99.9994% of the time.
How much your score jumps after an offspring dies in OHOL depends on how surprising the result is. Given that your OHOL genetic score is a moving estimate of the average lifespan of your offspring so far, if an offspring exceeds that averge, your score goes up. If an offspring dies much younger than that average, your score goes down.
The top players in OHOL with high scores are like grandmasters in chess. They have nowhere to go but down. If your Elo is 2800, there's no one you can play that will make your ranking go up very much, and a huge number of people you can play (everyone) who will make your ranking go down if you lose to them.
So if your genetic score is 55, that means that on average, you and your offspring have been living 55 years. Living to 60 isn't that surprising, so your score only goes up a little when an offspring lives to 60. But dying at 0 is a HUGE surprise, so your score drops a lot when that happens.
Your example is a bit misleading though, because you're not matching off against other players you're just competing against yourself and your past performance, so it would be like newbie vs newbie or gm vs gm in your chess example. A gm losing to a gm is not a surprising result, it's just fractional ebb and flow of the ones digit, it doesn't throw them careening down to return to the midlevel range with their gm rating long in the past from one reasonable mishap.
Also your experiment seems a little unrealistic, it appears to converge on a certain number but that seems to only be because it's a perfectly cyclical pattern. I bet if you made it 10% failure rather than 9 successes and 1 failure then it would be jumping all over after the occasional 2 or 3 in a row, when in reality even the best player can suffer through having a few babies die due to circumstances outside their control. The big question is whether someone who has a 90% success rate should be consistently rated higher than someone who has 80 or even 70% success rate. With the described algorithm they probably spend more time rated higher than not, but I'd expect them to frequently not sit at #1 despite their superior performance. But maybe that's desired, we seem to be focusing too much on the math and not enough on the ideal behavior so it's difficult to discern what's better or worse.
In another thread I mentioned the possibility of context which is relevant to your example. What if you get born to a really low scoring mother (or even a long line of them, implying a shitty or nonexistent village) but end up being a super achiever as a mother yourself, that would be the equivalent to a newbie getting one over on a gm and deserving a huge boost.
It's a mix of luck and skill, even what happens after you die. It's about making decisions that marginally increase the chance of your family living, like getting clothes to your babies, making batches of pies or compost, and in general doing what you can to improve and stabilize the basic resources that are necessary for survival. It won't guarantee anything but on average your family will have a better chance of survival in a village if you keep it well maintained than if you don't. Things like runner babies and new players balance out in the long run, so while they're annoying at the time, everyone gets that about equally.
This would all be perfect if every starting scenario was about the same - those who were better at providing would build up a larger score on average. The issue is every scenario is not the same, eve camps are tougher to survive in so playing at that stage will tend to be detrimental to life score, while a developed town is probably more beneficial to gene score on average. Some dynamics of the game contradict the gene score dynamic as well. For example two days ago I was in a nice starting camp with a well and one other person. Things were coming along nicely but then I started to have babies, around six in 5 or 10 minutes. I had no yum chain and no clothing. We didn't have the food supply to support more than probably one ideally, maybe two. So at that point the correct decision to avoid exhausting the village of resources so that it can continue would be to let four or five of those babies starve and tank my gene score. Two of these were new players too. Now I have a hard time bringing myself to leave any babies to starve, so I fed them all. The other player I had been peacefully working with cursed me, and the food supply was exhausted shortly after and we all starved. I understand I was cursed because I had too many babies and tried to keep them all. And I guess this type of scenario also balances out.. is it a desirable dynamic though? And I think it illustrates too how much better for gene score developed towns are, here I was forced into a scenario where my gene score was going to go way down regardless of what I did, which is ok, but gene score treats all scenarios as the same, and a developed town you don't really have this dilemma. I don't like also how the correct move can be to starve your babies, especially new players being the best choice to starve, but I'm not sure what can realistically be done about that.
Good example. I wonder if there's some way to take the context into which you're born into account, like if you come from a long line of low scoring moms but you knock it out of the park and keep 4 babies alive that should be wildly rewarded. Likewise if you come from a line of moms that all kept 3 or 4 babies alive but you can't even keep one alive then you should be penalized pretty severely.
Some systems should not run out. The 'everything runs out' notion is not one that will appeal to a wide variety of players. Some vital systems running out can work if other vital systems can replace them.
My vote would be on all systems running out but it taking a potentially very long time if your tech level is well above your population level, eg, if you have a well and only 4 players then that should last a really long time but if you have 40 players then that'll run out pretty quick and you'll need to tech up or build additional wells or move. This makes more sense without the rift, of course. Eventually you might need a train to transport raw materials from far away mines and water holes and such, something like that. As the radius of influence of your town grows the area of stuff you have access to increases with a squared relationship so that should provide a lot of wiggle room to a growing city assuming they're industrious enough.
So it does look like it goes down and down over time. However, it eventually converges. Running this 10,000 times (for 100,000 BB) we get this as the last cycle:
51.709164 52.538248 53.284423 53.955981 54.560383 55.104344 55.593910 56.034519 56.431067 50.787960And like magic, if we take the average of these last 10 values, we get 54.0000
So this is working, to effectively make your genetic score converge to the actual average of your recent offspring's lifespans.
6 points seems like a huge change for one baby dying from popping out while in the middle of getting griefed or something like that. Maybe it would be worth being really clear about what the goal of the algorithm is?
Do we want it to be possible for someone to remain #1 through being consistently better than everyone else at mothering? Or do we want randomness to give #2-5 brief moments in the sun at the #1 spot? Or does the #1 spot not matter as much as being at least in the top 10? Or does it only matter if they have a score of at least 50? Once the goals are clearly defined then it's not difficult to work backwards to what the math should be.
But yeah looking at things like how chess ratings work and the math behind how they do diminishing return over time might provide some inspiration and give a little more meaning to success rather than knowing it's short lived until you get born to a starving nomad and 4 kids pop out and you heroically manage to keep 3 alive but the 4th one destroys a week of progress and puts you back below a dozen other players who kill babies daily (contrived example, maybe daily/hourly rather than weekly/daily fits the described simulations better)
Went off nomading the other day and had two babies /die themselves when born when they saw I wasn't in a village. Why the hell can babies suicide anyways? It should auto-curse the baby from the mother when they /die prior to a certain age, especially since you can't even seem to curse babies in any way if they suicide before you can name them. Wasn't able to curse a baby I had named after they died either
Certain groups of players do not care at all about having more tools slots (griefers, roleplayers, others) and will be content with only being able to use 2-3 tools (knife, bow +other), these players will keep suiciding either to choose the life they want or simply to annoy other players.
I can already imagine the /die spam and baby bones everywhere.
As someone who mostly plays this game out of curiosity in the direction it's going and for exploring and nomading, and occasionally stirring up some shit and doesn't really want to do work in a game, can confirm. 2 tool slots is plenty for a little minimum effort help around the village and basic nomading.
I think dangling being an eve for really high performers would be a really strong motivator. Even for people who just like causing trouble the draw of being an eve is strong since it means you have the chance to start a new troublemaking family, and it would probably mean that they'd actually behave more so that they could cause a higher tier of chaos later ?
Now that im thinking about it, we could claim objects like we do property fences and if someone outside our family took them we could declare a war.
Yeah but then that's a way for some 10 minute old character to start a war between families, maybe the claimed objects can't be taken and property fences can't be broken through until a war is declared... something like that, and then obviously there would need to be a way to coerce a family into war
+1, having more interesting pvp and raiding elements would draw in a more diverse set of players too... Having supervillains makes a game like this very interesting ;P seems boring that there's no way to break through a property fence. If we're trying to model real life then there should be no 100% effective defense. In a vaguely similar game they implemented a very difficult to build "trial by fire" kind of thing outside the base that would make a base penetrable after a certain amount of time if the defenders didn't take it down before then (the raiders would have to defend the trial by fire during that time) perhaps there could be an offscreen notifier kind of like the combat exclamation mark
So if you are having babies away from town, these may be griefer babies.
My new life mission - start a black sheep offshoot of a family off in the deep woods, call ourselves sparta, retake the lands with blood
I had my kid telling me that we should use 1 soil. Soil is cheap I told him, it is now he said. Well, that's dumb as fuck
I have nothing to contribute here except that I wanted to make sure no one skimmed past this gleaming gem
Haha I was one of those babies, that mother was a champ she kept like 6 babies alive pretty successfully. Unfortunately there were a good 10+ babies trying to stay alive xD
Yeah the decision on whether the tribe kills the baby or not are more interesting and nuanced, the one I'm more concerned about is whether the baby suicides to get the more influential role and to avoid the danger of being the last man standing (which may be no fault of theirs, maybe everyone else ages out or dies out of carelessness or unluckiness) which basically turns it into a single player game. As a female, as long as you yourself are playing optimally, you're never certain that you're stuck playing single player until you hit old age regardless of the actions of everyone around you.
I'm not suggesting that change of course, since males shouldn't have babies, but just worried that the advantage of not having unexpected babies is too small a compensation and that it would basically punish players for not suiciding until they get a female, which would lead to player behavior that's detrimental to the flow of the game overall.
It's not obvious or provable enough to be more than a concern at this point. This is probably something that will require the open alpha to confirm or deny, and there will likely be changes that affect it prior to then anyways.
Though if that's "the way Ultima" did it... wow, that screen shot is pretty overloaded! I can't make heads or tails of it...
Are you aware that up-arrow works to cycle through previous messages that you typed? So if someone doesn't notice you talking, you can chase them and press up arrow instead of typing it again.
Haha well this is an extreme example just to show you how the positioning works. IIRC there's a hotkey you can press which makes everyone's name appear as if they said it as a message, so it looks like they just spammed that 3 times and took a screenshot with dozens of people stuffed into the screen. Normally it's far more sane. The feature I'm most begging for though is the pulling-it-back-onto-the-screen from a reasonable distance. If you don't think they should stack that's cool, but even with spamming it with up-arrow you still have to get onto their tiny little screen and you have no idea whether they saw it or not. If it pulls it onto the screen from a reasonable margin then it would be way easier to get someone's attention and relay quick messages without having to chase them down quite as much. I'm thinking like... 2-3 tiles off-screen? Maybe just 2. Basically, if you can hear them running then it follows you should be able to see them talking.
Okay, regarding the whole male thing.... life is very different as a male in this game. I don't know if I'd call it an advantage, but you can go out on long journey's and such without having to worry about being stricken with a baby along the way. You can also chose to be a hermit, whereas a female character cannot truly do that.
Also, the choice about which babies to keep and which to discard can be a little more complicated than "always keep females." It's true that they were first on the lifeboats, but in certain cultures, female children are sometimes seen as a burden. I think that's because of the potential reproductive load that they represent. The farm just can't handle a bunch of extra mouths to feed, and that's what a female child can mean in the long run, potentially.
So, which one you will favor may change depending on the situation. If you're barely getting your farm off the ground, you may prefer a male child for now, because that will delay the influx of additional population growth more.
And every baby that suicides wastes resources that bring the whole civilization down.
Yeah but having extra females vs running out of females is the difference between the colony continuing and it irreversibly ending. Having to eat another carrot or two every now and then due to a kid that you quickly kill off is a small inconvenience, especially since more often than not you probably would want that child unless you were very early game. Every male is basically taking food out of a female's mouth, and is no more capable of protecting them than a female.
And actually I was suggesting that seasoned players would probably kill themselves, not get killed by their parents (but maybe that too, but as you said there's a food cost to the colony that isn't suffered by the respawning child) because being a male will always be less contribution to the continuation of the colony than being a female and you get a fresh start every time so there's no real disadvantage to killing yourself anyways. I'm assuming that continuing a colony is the main factor driving gameplay, or that it would be for the more serious players at least who are going to be putting in the most lives.
You mentioned the bad mother penalty a bit ago, but that's basically just turning the female into a male and only after checking many times whether or not they want children and only at the expense of some food. And that food expense needs to stay low/reasonable since otherwise normal birthing is too dramatic/hazardous.
Maybe you could think of some more serious costs to letting a child die? Like maybe it would give a "morale" type of debuff to the mother and would make her move more slowly and be less responsive for some time, something like that? Maybe she's less willing to eat and her max hunger bars drop, perhaps permanently or for a while at least. If child death actually mattered then I could see your argument being stronger for the absence of births being a sometimes-advantage. As is it seems like once you get to farming or are near plenty of berry bushes it's not something that would slow you down much.
Anyways I'm making loads of assumptions here, it's probably better to just keep things simple and wait and see what happens when there's a more substantial population squirming around. I'm really looking forward to seeing that, even with our relatively small test groups it's interesting how quickly things can appear unbalanced when they seemed fine before. Hard to imagine how crazy things will get with swarms of people and the kinds of things that none of us will have considered.
My memory isn't enough for a life-by-life recount, so I'll instead try to make some observations:
Birthing is much more sane than it was previously, and birthing is much less likely to kill the mother. Out of about 10 births I only had to abandon about 2, another 2 or 3 wandered off aimlessly, I think I had about 5 successful births where they got to the point of being able to feed themselves.
One birth caused me to nearly immediately die as I wasn't near food and it brought me down to 4 bars while I was carrying something. Suddenly losing a large portion of your food supply is pretty lethal in the early game, but this is a little unusual since everyone is in the early game so 100% of births were to someone in the early game... Normally that would be a precarious but somewhat rare scenario I think, especially if current food stores affect birth. Maybe temperature should affect birth too? As in for two mothers who are full, the baby goes to the one closer to neutral temp.
As many noted, milkweed was a very difficult resource. Personally I didn't really realize how much milkweed is necessary to get to a rabbit skin water pouch and kept planning poorly with how I was using my milkweed. I'll try to list out the things necessary for reference:
- stone hatchet (4 milkweed)
- snare (4 milkweed)
- fire bow drill (4 milkweed)
- needle and thread (2 milkweed)
I think that's it... so 14 milkweed if you don't waste anything. If you go the bowl route you can skip the snare and thread, so 8 milkweed. This is all while shivering without any clothing too, a full set of clothing is 6 pieces so 12 milkweed plus another rope or two for snares for efficiency, so let's say ~17 milkweed per clothing set.
So currently the best thing to do is probably to rush to clay bowls, then rush to milkweed farming and carrot farming and making clothing as you're able depending on where your resources are, but probably not prior to putting down crops. So yeah, I guess since you need so much milkweed before you can make any yourself I agree its spawn frequency ought to be bumped up a bit. That way rushing to milkweed is optional, not the only thing that makes sense.
There were a couple weird issues presumably due to latency, like I picked up a baby and it died in my arms and I got killed by a wolf but no wolves got less than 3 tiles away from me. I'm guessing both of those things happen entirely on the server and being an open source game, not really a way to avoid that I guess.
Back to babies, I don't think I ever had a successful mother-baby interaction about feeding. As a baby no matter what I did they would ignore my A/Z and just keep picking me up (I thought Z was universal for "I'm full, stop" by now) though I could mostly get their attention to feed me with "A". IIRC you said breastfeeding only 1 bar was inefficient right? Whenever I tried telling babies to tell me "A" when they're low they just kept spamming "A" over and over so I just ignored them and tried to time it from memory xD
With multiple people running around and such I've been noticing that it's extremely difficult to communicate. Maybe you could make text bubbles snap to the edge of the screen when they're within a few tiles off screen? It's really annoying when I can see someone's full body but they're at the top of my screen so if they're saying anything I have no idea. Here's a screenshot of how Ultima Online did it - https://uo.com/wp-content/grand-media/image/Hokuto.jpg you can tell that speech that's too close to the edge or slightly off-screen is being brought fully onscreen and aligned with the edge of the screen. Speech from multiple people off the same corner will just overlap. Personally I'd prefer the speech to stack from each person too like it does in the screenshot, like if they say "M" "O" "M" then it'd stay onto 3 rows and each row would fade after X seconds (something very short like 1-2s). I'm sure there's some aspect of communication needing coordination or something like that but a lot of the time you can't even get the person to be able to tell you're talking at all, I think it's too far in the extreme end of the balance currently.
And yeah, with no advantage to being a male if I wasn't testing I'd almost certainly just suicide once I noticed my gender until I got female. IMO you should give males some consolation prize to not being able to participate in spawning... maybe they can carry 4 items in a basket and 5 in a backpack, maybe they're less likely to die from a wolf, not sure but it just seems strange to have two character types and give one of them a huge feature and the other zero. You mentioned you wanted to compare about lifeboats and drafts and such but if you give one character type advantages and the other character type nothing then of course the one with advantages will be protected, it doesn't matter what the advantages are, it doesn't really say anything about how valuable birthing is presently IMO, it just makes the game disappointing for folks who get stuck with males as it's most optimal to suicide until you get a female.
It's getting there though, the gameplay was for the most part pretty reasonable. Once some more content gets in there and maybe some diverging tech paths and some more biomes it'll really start to get interesting.
Sorry... I thought that 10a would work for everyone. I thought it would be 10p or something, being on the other side of the world, but I guess you're closer to California than I thought.
It's cool! 10am pacific time on sunday is generally the safest bet globally
Haha deal. Hm one hour further... you in japan or something? im over in singapore
2-4 am monday morning for me :'( oh well I can pop in for the first hour at least
Need to make a kiln for firing clay things https://onehouronelife.gamepedia.com/Recipes
AFAIK you need to cook the rabbit, so yes you need fire for both. For cooking rabbits you have to make sure to cook them on coals not on an actual fire, so after you put kindling on a fire you have to wait for it to go down to coals otherwise it'll burn and I don't think you can get bones from a burned rabbit (kind of weird, but yeah).
Fire is a bit tricky but not that hard once you get the hang of it.
- use the fire bow drill on a long straight shaft to make it start smoking
- use a leaf (plucked from the skinny bright green tree) on the smoking shaft to get an ember leaf
- use the ember leaf on tinder (from the bunny biome trees) to start a fire
- use kindling (any branch can be turned into kindling with the stone hatchet) on the started fire to turn it into a small fast fire
You can stop there, or you can add firewood to make it a big slow fire that will last much longer. If you're trying to cook rabbits you'll want to stop after you add kindling so it'll turn to coals sooner. After you add kindling you can use the long straight shaft on it to make a firebrand (or whatever it's called) basically it lights the shaft on fire and you use that on the kiln to light it.
Most of this stuff you can figure out with common sense and looking through the tooltips. I figured out pretty much everything just by looking at tooltips though there were a couple items I got stuck on like milkweed seeds, for that you can reference the recipe list.
Without there being a finite resource players would just sit tight and everyone would keep spawning there. I think making resources finite is necessary to keep players moving and expanding and keeping the villages chaotic organic things rather than just tedious repetition generation after generation.
Hopefully he'll add rare spawns or biomes that give incentives to continuing to move beyond just depleting nearby resources... but i think forcing movement periodically one way or another is a cool combo with the constantly shifting populace. It means the flow will be very unpredictable and organic and unplanned
It is interesting that two people can carry some stuff but an exodus would be far easier with more people to transport tools, foods, utensils, etc...
Yeah seems like two people with two carts and two backpacks can currently transport most of everything required to kickstart a new colony... eg, all the metal tools and enough food to last through a very long stretch of no berry bushes. One person could do well enough since most of the metal tools aren't very essential aside from the steel axe.
Kind of funny leaving behind a colony that's stripped of resources... Anyone who finds it will probably be stoked until they start panicking and die of starvation since everything around them is stripped bare xD
Do goose ponds eventually replenish themselves?
IIUC the "respawn near where you died if you die of old age" feature is mostly just a band-aid for the server being mostly empty for now, so maybe this isn't important, but figured I'd note it anyways.
I died of old age and came back a day or two later (after closing the client of course) and when I spawned I became an eve but it was not in my previous homestead, it was in an area I'd never seen before.
The interesting thing is that I spawned right in the middle of someone else's homestead. It made me wonder if it somehow mixed up my last spawn area and someone else's.
It would be really cool if the respawn-nearby-on-old-age-death feature was reliable though... It's difficult/impossible to get to the endgame when we keep spawning in different areas even when we're not starving or getting killed. It's also a lot easier to test multiplayer aspects of the game when the same people can keep spawning in the same place.
It's kind of a tricky thing to test because I can only check it once per hour that I'm playing, but maybe it had to do with the fact that I closed the client before spawning again? I assumed that it doesn't matter since being able to close the client is kind of essential unless you have a 5 or 6 hour block open to be able to make progress across the board.
Let me know if you need more information.
Oops yeah you're right, thanks for the correction. IMO it'd still be a really valuable change for "pausing" reasons, but yeah as is backpacks aren't quite as cumbersome as I described.
What if your character automatically ate food that was stored in their backpack when their hunger gets down to 2 remaining squares? If you put 4 carrots into the backpack then you could idle for quite awhile. Even longer if you had higher quality foods. Would also be nice for end of life, you could stick a few berries in there to make it a little less stressful to make it to old age death. Would make the backpack more useful and less tedious (kind of a pain to take it off, drop it, take something out, eat it, pick it up, put it back on).
Maybe it could also be the case for a basket you're holding so that early game players could still idle, albeit less conveniently. Also makes exploring easier and less tedious.
