a multiplayer game of parenting and civilization building
You are not logged in.
Upvote. Let cities self-destruct without bothering everyone.
It seems like lineages were indeed already dying out super quickly even with the Eve "exploit" and without the apocalypse to wipe out the towns.
Before the update, any time an eve died of old age the next eve would spawn in the most active town due to a bunch of bizarre mechanics interacting unintentionally. So a long lasting town was made of many eve lineages strung together.
Woah!
I'd love to help where possible.
I'm a noob at Python, but your code could help me/others learn and in turn help.
My current interests (as on your previous thread) lie in parent/litter statistics, and how to make that experience better.
Town survival is a super relevant statistic with the apocalypse 2.0 update. But with the eve-town-revival ban, I'd guess towns normally only have one lineage now.
My takeaways from the graph:
-wow! Steam surge acted like I thought. The active pop seems to be leveling back out at 120; triple the reach!
-Wondering if the AVG lineage length stays somewhat constant; Steam may have made the 24 hr lineage possible only because of the shear number of attempts.
-Looks like a large effect from player skill comparing the first few weeks of steam release.
If you already grabbed that data, cool.
CSV works as a standard
either PM me or a link (google drive or?) for everyone here
Bless.
I'd struggle to put that data together myself. Full of anticipation for your next insightful plot.
Speaking of sustainability,
Now that Eves can't revive cities, /die-Eves only hurts cities.
Every time someone /dies to become an eve, all other mothers get -5% fertility.
Of course, we can't prevent women from choosing a primitive life; they can leave town if they want to. But at least those moms will be close to a civilization center for a better chance at rediscovery. And they'll become a mom by earning it rather than /die-eve magic.
I propose /die-adams to keep the lineage ban intact.
Maybe I'm not understanding the play-style or intentions of /die-eves?
Looking at another cause of infertility; /die-ing to become an eve.
I wish I had data on the average ratio of birthers. It's the biggest unknown in the equation.
On server 1, out of a population of 60, a liberal 25% birthers would mean there're 15 mothers.
Everyone's birth rate drops by 6.25% each time a person /dies to become an eve.
[1/15 - 1/16] / (1/15) = 0.06250
By my current estimates, adding an eve to a full server is worse than facing nocturnal infertility on a sub-server.
Eves should only be made when all possible mothers are on birth cool down. -To keep the lineage ban alive, /die-eves should be an infertile variant.
(but still, the biggest problem is servers bouncing on/off)
The difficult-to-create item that Jason has actually made a requirement for the apocalypse is not the totem pole, or a car... it's a sustainable lineage. We can have the apocalypse just as soon as we figure out how to create a lineage that can survive for twenty-four hours. One of the big obstacles to that is global infertility, another one is male/female RNG, and we can't do anything about either of those besides come up with hacky work-arounds like having private servers or private lineages. But another obstacle is player incompetence, and while we can't exactly fix that, we can all do our part to better teach and train the newbies so that lineages are less likely to die because everyone forgot to water the f*ing bushes or don't know how to make a shovel when it goes missing.
Seconded.
I like that the Tower is low-tech; it presents the challenge purely as sustainability.
To be the ultimate griefer, you must help Tower-towns survive.
I'm not sure I'm understanding the proposed solution.
-How will the pipes be connected without "reading" other tiles?
-What is the "system as a whole"?
-How will the engine know which different animation to play?
-How will it water plants without being part of their object?
Compared to the major issue here of servers bouncing on/off, I'm not certain nocturnal infertility is a huge killer when I put numbers on it (but it can still be improved by de-concentrating the load)
Assuming as above, populations have ~20% Birthers. And pulling numbers from thundersen's graphs, population crashing window from 00:00 to 12:00 UTC is approximately 12 persons per hour.
AvgBirthRate = (-0.2 persons per minute)/(Pop*0.2partsBirther) + 0.4 births per minute (the constant-pop birth rate, from above calc)
The smallest %change in overall birth rate is when the population has just started dropping, the last two servers at ~60 share the burden.
0.2 / (120*0.2) = 0.0083 less births per minute
a 2.1% change
Likewise
The largest %change in birth rate is just before server1 starts sharing the load,
4.2% fewer births
Then when Server1 starts sharing the load, its back to
2.1% fewer births
Sharing the loads between servers equally would cut these numbers in half, and the game's difficulty wouldn't depend on which server we get assigned to.
One idea, regarding fertility, is that towns are supposed to be competing for babies. If all the fertile women huddle near the fire, they will swamp the heat weighting and "attract" babies. I will also add yum bonuses into the mix here (the women with the highest yum bonus will also have the highest chance of having a baby).
Warmth is neither a challenging nor rewarding way to increase birth rate.
Biome warmth (read: baby attraction) 0 - 0.5 scale:
Grassland 0% ins.: 0.0705
Grassland 40% ins.: 0.101 (max ins to keep mosquito bite non-lethal)
Desert 0% ins.: 0.158
Jungle 0% ins.: 0.456
We can increase our birth rate up to* 6 times (grassland vs jungle) or 2x (grassland vs desert) by being warm. *(but usually we hit the birth cool down cap first)
For standing in a jungle infested with mosquitos. Or standing idle on a desert corner eating 1 berry each year. I'm not going to be nearly as warm while farming or baking or doing anything productive for my baby's future.
If I work hard and clothe myself fully, my warmth becomes only 0.25, and I risk a mosquito overheating me, in turn killing my baby.
Mainly the jungle update made this mechanic obsolete. We deserve more challenging and rewarding ways to affect our birth rate. Which might mean just leaving it to our parenting skills.
Waiting for years restricted to warm tiles is not fun gameplay. At least, I'd like to see a game where if I want my kids to live, I spend my life preparing clothes for them rather than being glued to warm tiles. That is, equal birth rates and let our tech speak for itself.
NOOOOO don't add yum bonus into the mix PLEASE. Rip Eve camps if that happens! And I chain yum to be able to travel without devouring foods, not to get more kids! I already try to keep myself cold when I'm out to get rabbits so I don't pop out kids when I'm out of the camp.
I agree.
Likewise with the warmth mechanic, in a productive (not waiting by fire) life I find myself warmest when exploring for resources because I make my path along desert borders. Definitely NOT when I want a baby.
Warmth does not relate to my intent to have a baby, sans metagaming.
But yes, lineage bans are a problem here too during nighttime hours. Ideally, a small group of friends would be able to keep an important town alive through the night, but lineage bans clearly prevent that (though the lineage bans only kick in when server population is large enough). Lineage bans are really important the rest of the time, though..... so it's hard to reconcile these two things.
Maybe I'm misunderstanding, but this sounds like you DO want that group of friends to replay in a city? I'd thought cities were meant to have mainly fresh faces passing through.
Lineage bans seem to be smartly balanced for all server loads to keep people from replaying towns; as I understand, 30 minutes for every 13.33 people on the server (min zero at 10 players and max 1.5 hrs at 50 players). I thought it was working as intended.
Regarding nighttime infertility:
The law of large numbers (resistance to RNG) is weakest when population is low.
But less intuitively, infertile hours are from 4pm till 4am Pacific Time (00:00 - 12:00 UTC) based on thundersen's graphs, while the player count transitions from high to low. Getting a baby becomes harder (~3% harder, but only because the drop is concentrated on the last two servers) as soon as the player count has finished peaking for the day.
But more importantly, servers are split or are removed too frequently, during which birth rates plummet regardless of player action.
Hey guys, here's a new link to a more recent version of my plot. Not exactly sure what happened, but somehow plotly is redirecting my uploads to this.
It sounds like thundersen's been pouring over more detailed life-log data, but based on just these graphs I'm seeing predictable population surges.
During the largest peak, 3 servers of ~100 could hold everyone
During the lowest trough, 3 servers would drop to only ~20.
Unless the %fluctuation in active playerbase grows, there should always be some ideal number of servers for any future state of the game's popularity, without needing to start/stop servers so frequently. Basically what CrazyEddie had said previously.
Me throwing numbers around:
A server splitting to grow a new server is just as severe as shutting a server down.
In both cases, ~30 people leave the server in ~30 minutes
Which translates (see above eq.) to -1 births per minute/ 80players*0.2partsBirthers = - 0.06 births per minute, granted this only lasts for 30 minutes.
a 15% drop in fertility when a server starts splitting
and 60% drop in fertility when the server is being shut down.
hmm... I mean these are just estimates.
But the net result affects everyone, because the same villages are kept alive even after they die, removing the challenge of reaching higher tech or longer-term projects (like the bell tower).
This appears to be the only given reason for making a change. Plus the overall vision for the game of course.
There's been discussion over what makes the game of long-lasting civs challenging. It's not iron, or charcoal, or labor. It's instead lack of baby girls, caused by RNG and nocturnal infertility -- things no player can ever master. Removing eve-save-revivals will exacerbate those challenges that are outside of player control.
Others here have mentioned population crashes too, and they're supported by thundersen's graphs of server populations over time.
I'm pretty sure that I will change the Eve spawning rule in (1) to only apply if you are the only player on the server, or maybe if the server is below some population cap.
Maybe the rule is this: if there are any mothers on the server of fertile age (whether or not you qualify to be the baby of one), then if you are Eve, you are placed in the wilds instead of your saved Eve spawn location.
Those would both weaken the playability of the game when player count is low (i.e. high# servers, and the distant future).
Reason being that, since active player count fluctuates, the last player on the server will keep being pushed out of their village whenever enough people rejoin.
I think you made all the mechanics already, they just got connected unintentionally through /die-ing to become an eve.
Eves should spawn only when there are no fertile mothers (controlled as intended by your birth cooldown mechanic).
Eves should revive the last town only if they are the first eve on the server.
Whenever a player denies all current lineages, they become a male in the *wild.
*So if they want to be reborn to an old line (to grief or oppose the one-life vision), they still have to serve their time in some other life. Lineage ban doesn't affect small pop servers already, so those will still be prone to grief, but won't be ruined by this mechanic.
The only functionalities this takes away is an artificial bonus from old eve death, and choosing into a fertile eve life. Shouldn't those be rare and special, anyway?
I'd gotten that notion reading the July 2017 thread, and the reflector code has comments for it.
[...]
I just realized that you can currently "game" the placement alg at the 50/50 stage by suiciding over and over until you finally land on the server that you want to play on.I can fix it by seeding the random number generator with your supplied email address. However, [...]
So, I guess I just need something that will work for the average player who knows nothing about this, to prevent repeat-suicide until you end up in the "old world," because if everyone is doing that, the load balancing essentially doesn't work.
[...]
The base fluctuation is unavoidable, but it seems like the handling of it could be improved. Is there any good reason that the minority server HAS to be starved off artificially? Why not just let that line receive its fair probability of children and give all lines on all servers a fair chance of dying off?
Your analysis looks right to me.
I'd add another con could be that servers with populations under ~20 are more likely to give replayers lineage bans from all current lines, causing a few extra eve spawns overall.
Though I'm not sure what connection magic forces replayers to stay on their original server.
I would love to hear Jason's take on this.
Me too! No matter how awesome and revealing your plots are (super cool, thanks again!), he's the one with the power.
I've no luck finding info surrounding the Nov 20 reflector commit. But the problem still existed in other forms before then.
This old July 2017 thread has a lot of statements about the issue back in the day. The only changes since then were to the server cap as lag was managed, and the Nov20 commit, afaik.
Here's my grabs:
[...]
But once the balanced populations get too low, we switch out of "spreading mode" and stop sending new players to the secondary server entirely, and it becomes a "Children of Men" scenario while the last generation of people live out their lives there.This works similarly when the population is big enough for three servers. We keep spreading them evenly until the spread population drops below the 20 threshold, and then suddenly stop spreading them.
Actually, the spread isn't even with three or more servers. Because of the multiplying fractions, only 25% end up on the second and third servers. With 5 servers, the fractions are: 50%, 25%,12.5%, 6.25%, and 6.25%
Of course, this means that in a really heavy population boom, the main server will still reach 95% faster than the rest, and so on down the line.
Maybe it's okay... but the algorithm can be improved to make it more even, someday.
[...]
I do want the majority of players to have a "full server" experience, and have the "desolate server" be a more rare experience.
In your last case, where the population falls to 513, we have 72% of the player population on very full servers, 14% on an almost-half-full server, and 12% of players on desolate servers.
An even spread across all 7 servers would put 100% of players on an almost-half-full server. But trying to maintain an even spread no matter what would mean that sometimes the entire population of players experiences more and more desolation. In the case of a dip in population, we really want to consolidate somehow and bring the remaining people together onto the same server.
[...]
Can you run your sim with various parameter settings and take some measurements?Like, number of lone eves? Number of times a server gets cut off from new babies and dies off completely?
Currently, the reflector is using 50% and 10%. Maybe other settings are better.
Also, if we count the number of servers that are currently being spread to, we can spread with different splits along the way (instead of 50/50 at each step). Like with 7 servers, the first server should send 6/7 to the next server, and that server should send 5/6 to the next server and so on.... I that should put 1/7 on each.
Yes, the front server starts with a higher pop, so during a boom, it will keep a higher pop, but only for an hour max, until those people die. After that, each server will have 1/7 of the population.
We could still keep the stopSpreadingFraction, and once the 1/7 amount dips below a threshold, we start letting extra servers die off and focus new players onto the remaining servers.
But I'm not sure if that's better.... maybe there's something good about keeping the top servers more full.
We can either give most players the better experience, at the expense of some players having a less good experience, or give everyone the same mediocre experience.
[...]
Well, the problem with 0% as the lower threshold is that the population will remain spread, forever, no matter what, after it starts spreading. So if it dies down to only 5 players, there will be one on each of 5 servers. At that point, they should obviously be together on one server.
I'm trying to make a system that doesn't require any manual babysitting.
"all about mosquitos" is outdated
The mechanics have changed (now 8 heat, for 35 seconds, and 30% food) and the rest of the comments are conjecture.
There's scattered discussion about mosquitos, but no thorough guide I've seen. Pein's had practice trapping them, but still no definitives.
Hate to be the resident necromancer, but this thread is interesting. Has anything been done to combat this? I think someone said keeping your satiation bar full increases your chances of being picked for a baby when a player joins. Is that a myth? (I think that was a comment on one of HoneyBunnyGames' videos.) Maybe not a strong effect, but it could give a person some control. I would hazard a guess it's not strong enough to defeat this cyclical problem, though. [...]
TY! I've been linking this in discussions since the OP realized this effect really early on.
For those interested in controlling their birth rate, the current mechanic is warmth...
(server code)(and it PEEVES me!)
I (and anyone) can increase my birth rate up to 6 times (grassland vs jungle) or usually 2x (grassland vs desert) by being warm. If there's 15 mothers half as warm as me on the server, I cut their birth rates down by ~6%. For standing in a jungle infested with mosquitos. Or standing idle on a desert corner eating 1 berry each year.
If I work hard and clothe myself fully, my temperature reaches only 50% perfect, and I risk a mosquito overheating me, in turn killing my baby.
Mainly the jungle update made this mechanic obsolete. We deserve more challenging and rewarding ways to affect our birth rate. Which might mean just leaving it to our parenting skills.
Warmth per biome (naked) 0 - 0.5:
Grassland: 0.0705
Desert: 0.158
Jungle: 0.456
A player joins and is directed to a server (or connects to one manually). [reflector code]
Once connected to a server, an eligible mother on that server is chosen at random [weighted by their wamth, server.cpp 4400] and the player spawns as that mother's child. Eligible mothers are fertile females that aren't in birth cooldown and whose lineage the new player is not on lineage ban from. [4167]If there are no eligible mothers than the player spawns as an Eve [4288]and the spawn location is some combination of "spawning at an old Eve camp" and "spawning on the next spot on the spiral Eve spawn chain". [+Eve 'spawned next to random player' 4390]Each time a female gives birth they're put on a birth cooldown timer chosen randomly from 0 to 5 minutes. [560]Children who /die don't trigger the cooldown (or, rather, they untrigger it when they /die, resetting it back to zero).[8027]
Living more than thirty minutes bans you from your current lineage for 90 minutes (clock time, not play time, I think) on a multiplier based on ServerPop]. Murdering or being murdered does the same. Using /die does the same. Eves who die of old age are not lineage banned and so have a chance to respawn as their own descendant sooner than the 90 minute timer would otherwise allow.
If you're cursed then only mothers in Donkey Town are eligible. If you're not cursed then only mothers not in Donkey Town are eligible. If you are cursed and there are no eligible mothers in Donkey Town then you'll spawn as an Eve in Donkey Town.
^Sounds right. Verified where I could.
---
As I understand you, AvgBirthRate is capped at ~1 per 2.5 min
if AvgBirthRate > 1 per 2.5 min,
eve spawns, increasing %Birthers until AvgBirthRate =< 1 per 2.5 min. (thats up to 10 babies in an average birthing window!)
Got me wondering if a constant pop server normally is capping their birthrate
Average lifespan was found to be ~15 for women (excluding eves) and ~10 for men (higher infant mortality, and other factors) from here
If I had to estimate %Birthers I'd say out of a pop of 100, 66 are female. 33 make it past 15, and 20 are Birthers. (close to your guess) (real age of death stats here)
So that makes AvgBirthRate on a constant pop server about 1/(12.5*0.2) = 0.4 per minute, which happens to be exactly the birth cool-down cap of 10 kids in a birthing window. But these are just estimates.
That 2.5 average birth cool down has a delicate job of limiting maximum birth rate, but also making sure to not excessively produce eves since they'll make infertile hours even worse. (also in the last link topic, total eve spawn rate was 10%)
As mentioned in the 'Having 0 daughters thread' recently, each person /dying to become an eve during infertile hours cuts into everyone's fertility a tiny bit.
but regardless, the server problems still stand, making some lives harder to grandma in.
Really, the math doesn't reveal anything, it's just me wrapping my head around things.
Your graphs tell the story best.
The main problems visible; player's offspring experience varies greatly (and unnecessarily) depending on their server's situation.
But the solution also has to fit into Jason's goals/ vision.
I'm curious how eves might help balance things too, but I don't yet understand their spawn mechanic.
TY for this thread. I think its the most applicable solution.
After turning my attention to being a good mom, I did the same as lionon, and got the same results.
I imagine the OPs method working in villages. I've reminded many farm-learners to eat! We've been in their shoes. Guide them down a learning path thats safe, and they'll live to repay our efforts. If not in this life, the next.
The point of this post will be to show how server login routing affects the in-game experience. So I’m linking Logins to BirthRate.
As always, I could have missed something in this model.
Assumptions:
-Only automated server choices
-No delayed effects, e.g. QuitRatio is constant
Note: I’m calling fertile women ‘birthers’ because its shorter and sounds less medical.
On any given server,
PopChange = Logins - Logoffs (replaying folks are recycled) [eq. 1]
noting
Logoffs = ServerPop*QuitRatio (QuitRatio is a constant, related to the average playtime per person) [eq. 1a]
AvgBirthRate = PopChange / NumBirthers + ReplayingP / NumBirthers [eq. 2]
noting
ReplayingP = ServerPop / AvgLifespan [eq. 2a]
NumBirthers = ServerPop * %Birthers [eq. 2b]
Putting it together,
AvgBirthRate = PopChange / NumBirthers + 1 / (AvgLifespan * %Birthers) [eq. 2,2a,2b]
AvgBirthRate = (Logins - ServerPop*QuitRatio) / (ServerPop*%Birthers) + 1 / (AvgLifespan * %Birthers)When the server pop isn’t changing,
Logins = Logoffs, (meaning PopChange = 0) [eq. 1b]
AvgBirthRate = 1/ (AvgLifespan * %Birthers) [eq. 1b,2,2a,2b]
= ReplacementFertility?, the ReplacementFertility is some constant (independent of ServerPop) related to player skill, similar to real life litter sizes. (In human terms each mother needs ~2.1 babies to become a grandmother, in OHOL it's more)
Rethinking whether it actually equals the AvgBirthRate of a constant-population server. Maybe it's lower and players on those servers are struggling, and eves are making up for lost lineages? Maybe those servers have it super easy? (probably the latter; see post below)
Fun gameplay is when we’re close to ReplacementFertility - just enough babies to keep our lineage going on average. If we’re skilled, our lineages tend to grow, and if we’re bad moms our lineages tend to fade.
IF we want all players to feel like they're in a constant-pop server, then PopChange needs to be very small
compared to NumBirthers on a given server, but how?
We can't control Logins, but applying the Logins across the whole population would keep PopChange/NumBirthers small.
One way to change %Birthers is to increase the number of eves, as is protocol if all birthers are on cooldown.
QuitRatio is not likely to change.
I’m still mulling over in my head what the best approach(es) may be.
Right now, there are two ways population crashes occur:
1) When the active player count is falling, the last 2-3 servers are handling all of the player fluctuation while the other servers sit happily at constant population.
I’d rather have all servers bear the load equally, as was the case before the Nov10 reflector commit (sans when total active player count > 2* server max.)
Better yet, send Logins to each server proportionally to how many birthers are on the server. Every single woman will have a near equal opportunity for grandkids. Maybe one server will dominate because its tech is thriving and will reach the server cap. But the lone woman who finds the perfect spot on server 4 will have a chance to make a thriving town.
2) Servers rising cuts into the second-to-last server. And risks reversing the process soon after.
CrazyEddie’s suggestion would help, and is easy to implement.
WOW! Neat stuff. Thanks for putting that together!
Irrelevant:
Server4 just popped back up, so I'm going to log onto it once server3 hits its minimum (10% I think).
(was just [~00:00 UTC Dec16] Johny http://lineage.onehouronelife.com/serve … id=2406576 on server 3 while the population dipped from ~40-23. I was in a large town with lots of women, so I can't say there were problems. The lineage data shows more than what I could see in a life.)
(missed the killing of server4. happened pretty quick <30min, just before 08:00 UTC Dec16)
...In addition to that slowdown, there is also a devastation which will be visited upon the inhabitants of the highest-numbered servers....
True about server ending/starting, as you describe for stagnation, but also when growing as seen for server 2 in thundersen's graph when starting to spread into server 3. The latter seems more easily avoidable. (like at least taking players equally from servers 1&2. Or ideally, if the active playerbase is growing fast enough, keeping the previous servers constant and skimming their surplus onto the new server)
- - - - -
I had made my conjecture looking at the following optimistic simulation. I wasn't looking at the real world where we have ~150 active players at a time.
When the total player number is somewhat greater than that of 2 maxed servers, then the main server will still have enough players joining to stay maxed, even in a population decline; which would in turn push the infertility burden onto the secondary servers.
Edit: outdated, as thundersen points out. But still applies after one server has capped (at 1/2 its max).
...I've run a rough simulation of your server system in which I had a max of around 1000 players and a min of around 500 over a day.
What I found happened was that during the times with the most players 7 servers were populated with a distribution something like:
191 191 191 190 100 50 50
So the first 4 stayed 95% full 5th around 50% and the next 2 25%During the times with the least number of players the last 2 servers would go extinct and the distribution would be something like:
189 184 74 33 33 0 0...
I'm still confounded as to why thundersen's graph shows server 1 remaining at its 50% mark, even though its secondary servers have more than its population. (shouldn't that mean there's enough players joining to grow the main server's population? unless artificially capped there)