The three options for fixing the variance on pump exhaustion

jasonrohrer · 2019-05-07 17:41:03

1. Sounding pole that is provided by the pump. Insert pole between pumpings to measure how much water is left. Pole itself tracks pump lifespan. Maybe it can only be stored right near the pump somehow, or too heavy to carry long distances, or some other thing that prevents fresh sounding poles from being imported.

2. Big water jug or barrel. Inserted into the pump before pumping. After pumping, it removes all the water in one action. After that, barrel can transfer water into the cistern. Then pump can track it's own lifespan via a use counter (instead of using the use counter to track number of buckets of water removed). Lifespan visible as moss on the stones.

3. "Milking contest" style thing where the pump runs for a short time, and there's infinite water available during that time. Lots of buckets and cisterns, and a group effort to fill em all up while the pump runs. Maybe some "fishing" aspect, where it takes two seconds for one person to fill a bucket, and they're stuck there while it fills, so it's not a total APS click-fest, but instead a question of how many people you have helping. Again, the pump can track its own lifespan, as visible by moss on the stones.

There were other ideas floated, but many of them were similar in spirit to one of these three ideas, or they would require too many engine changes to implement.

2 and 3 are the most straight forward to implement with the current engine, with no holes in the design.

I'm not totally thrilled with 2, because having yet another water storage object around isn't great, but it will have the least weird side-effects. Having it only transferable into a cistern will keep it from being redundant (you can't just bring the barrel to a farm and take buckets out of it... the hole in the barrel is too small for a bucket). Another weird thing is carrying a huge barrel of water around. Probably need to have it only moveable by empty cart (which becomes a water cart, temporarily). Another complication is that each more advanced pump currently produces more water per pumping.... that means that there will have to be separate logical barrel types for each pump (and separate logical water carts, if I go that route), but that's doable.

Milking contest would have tons of side-effects, but is also pretty clean otherwise, really easy to implement, and it makes logical sense (pump is running!), and it might be kinda cool as a group-effort (which will require leadership and communication), and cistern layout around the pump could be cool. The advanced pumps would run longer per firing, obviously, which is again very easy to implement. Before building a better pump, you'd better have water storage ready to handle the extra water per pumping.

I'm currently pretty torn between these two.

Twisted · 2019-05-07 17:55:14

Looking at these three options I like #2 and #3 more than I like #1.

If you do option #2 (which I like most) you could make all the pumps give the same amount of water, but give them a smaller chance to stop working the more advanced the pump is. There could be a new water cart item that would basically be just a big water tank on a cart, and you could use it to transport water around town which would feel pretty cool IMO. If you make it require the Hand Cart with Tires that would also give us another reason to make the upgraded cart.

The recipe could go like this - Bow Saw + Boards = Large Wooden Disk, Two Large Wooden Disks + Big Stack of Boards + Two Steel Blade Blanks + Rope = Immobile Water Tank

It would also give us a reason to have a cistern near every farm which is a good thing in my eyes.

If you do option #3 I would be very grateful if you allowed us to stack four or five buckets in a pile.

Tarr · 2019-05-07 17:56:52

I personally like option three the most due to the coordination it requires from the community. Option 1 suffers from the chance that if the sounding pole was movable it could be hidden which is obviously bad. With option 2 adding a barrel would sort of devalue having more than a few cisterns and might just end up being replacements overall for some cisterns should they be added as suggested in option two so I don't think that's the right choice at all.

My only fear with option three is we see the same issues we had with cows: either the thing never turns off and people abuse it or the it accidentally just stops working all together. Other than that I think making players work together to collect water is a great choice as at the end of the day this game should be about teamwork an building together on at least some level.

Option three

Ari-ori-412 · 2019-05-07 17:57:34

You could make a poll and have people vote witch method they like best, i personally prefer 3 because it makes people work together a bit more.

Dodge · 2019-05-07 18:00:21

"Milking contest" idea sounds more fun and a group effort event kind off thing, it has to be well balanced though otherwise you could get 20+ buckets with just one firing of the pump.

Also it does sound logical because currently the water pump seems to pump water in the well but how does the water stay on the surface level of the well? Having the water flow down and not stay on the surface makes much more sense.

As for the water container of number 2 it could be fuel tanks used to hold water and the more advanced the pump the more fuel tanks you could attach to the pump.

jasonrohrer · 2019-05-07 18:04:02

This thread is the poll!

Booklat1 · 2019-05-07 18:05:14

2 seems best but i think tarr made important points.

if barrels/tanks exist they have to be worse as storage than cisterns, so costing some iron should be needed.

Turnipseed · 2019-05-07 18:08:39

I like 1&2

Jaona · 2019-05-07 18:11:37

I like number 3 the most, then 2.

Spoonwood · 2019-05-07 18:24:46

It turns out that the update has sort of altered the status of cisterns as to whether you want them or not on bigserver2. With families that import water from shallow or deep wells, they are good. With families that throw water in them from a charcoal pump and consequently run the charcoal pump more, they are bad. Tarr doesn't seem to have picked up that the barrel would only interact with a cistern.

The barrel idea motivates cisterns more, so I like it. I don't understand why we would need a barrel to be separate for each pump. Couldn't the barrel just remove two buckets of water from a pump, and the charcoal pump now produce 4 buckets of water, the kerosone pump still produce 6 buckets of water, and the diesel water pump produces 8 buckets of water? Or maybe it removes three bukcets of water and the charcoal pump produces 3 buckets of water, the kerosone pump produces 6 buckets of water, and the diesel water pump produces 9 buckets of water? Also, you're still allowing a bucket of water to get put into a dry charcoal or kerosone pump, correct?

I prefer one over three also, but I pick two, the mighty barrel.

I'm completely against three. Personally, I've often enjoyed games like civ III and RimWorld where you can pause or let the game sit there while you doing something else. Bigserver2 OHOL has enough time pressure as it stands, and I don't like the idea of it having more time pressure. Also, multiple cisterns in one location isn't the best call for farming or a village in general. Cisterns, or water buckets, ideally exist around each farming section, one near the oven, and one near the smithy (your family needed to rush the diesel water pump so you had no time to get one up and I understand that... but if someone wants to make a car or plane or radio after a diesel water pump, a full cistern one spot right of the engine can help them). Idea 3 puts all the cisterns in one location making towns worse in terms of potential quality, because they just had to make a cistern cluster to maximize water output.

Thaulos · 2019-05-07 18:25:55

No chance of having the interconnected items?

Maybe just the requirement of a third party presence in the area within x tiles? The pipe could just be a measuring station that goes down one charge every time the pump is fired.

Thaulos · 2019-05-07 18:28:44

I don't particularly like either of the options tbh. I would take the exhausted well refilling again after a certain amount of time (30 minutes/1 hour?)

Amon · 2019-05-07 18:31:32

I think this will be torn again because most people like 2 and 3. They're good ideas, but I think I lean more to the barrel thing because it's more foolproof.

Water barrel ideally should only be movable by a cart and rolled slowly by hand. Just imagine how much water the thing is holding!

futurebird · 2019-05-07 18:51:45

I like anything that gets more people involved with pump tech. Even if it's just helping empty a bucket in to a cistern so it can be refilled. Even if people go nuts and make tons of buckets and cisterns you are limited by population as to how much water you can get. Let's say 1/5 people participates (that's being really generous) you'd need a much larger population (and more buckets and cisterns to get more water.)

And it's still going to wear out. So, yeah, you got 15 buckets, but you probably had a ton of people and the water will go much faster anyway. And after 5-8 such bonanzas the pump will still break, it's still 1/20, just it would never break on the first use and almost never on the second.

It'd be neat too, maybe the first time you run it you have just three buckets and no cistern so you can't make the most of it, so you are motivated to upgrade water storage and make buckets so the next run is even better... I could really get in to that.

I also find mass storage tempting, but I wish more things in the game were like baking pies or making plates and bowls with a few people. You always say "nice!" after you get it all done and if you filled two cisterns you would be super proud of your work.

One more though. Each time you run the pump the run is a little shorter. So 30 seconds, 25, 20 15 etc. That could keep it in check without having the pump grab the bucket and hold it or something like that.

The other super fun thing in the game is pulling iron. It's nice to have something that can go a number of ways, with a little jackpot potential. Water is kinda boring as it is now, and this would make it more exciting.

This wouldn't alter the diesel pump either.

wondible · 2019-05-07 19:05:31

Spoonwood wrote:

Couldn't the barrel just remove two buckets of water from a pump, and the charcoal pump now produce 4 buckets of water, the kerosone pump still produce 6 buckets of water, and the diesel water pump produces 8 buckets of water?

Engine wise, counting how many barrels a pump holds isn't any different than counting how many buckets it holds. The point of tank/barrel mechanic (really, all these mechanics) is that the pump doesn't need to count anymore.

---

I like the communal idea of #3, though doubtless first time users will get one or no water from it - just like everybody burns their first rabbit.

I also like #2 setting up tank/barrel tech, up until the multiple different sizes part, which seems excessive. Possibly awkward as well. I suppose they would have to be used on an empty cistern, but fill it different levels?

Twisted · 2019-05-07 19:13:34

Spoonwood wrote:

Couldn't the barrel just remove two buckets of water from a pump, and the charcoal pump now produce 4 buckets of water, the kerosone pump still produce 6 buckets of water, and the diesel water pump produces 8 buckets of water?

The entire point of the change is to remove charges from pump, so no, it can't work that way.

Spoonwood wrote:

Personally, I've often enjoyed games like civ III and RimWorld where you can pause or let the game sit there while you doing something else.

I'm not sure if I'm getting your point here - you do realize that OHOL is primarily a multiplayer online game, right? You can't pause those.

Dodge · 2019-05-07 19:17:42

In case this ends in a draw, why not both?

If you put the water container on the pump you get a fixed ammount of water and if you dont you get to play the "Milking contest" and potentially get more.

Where it could be interesting is if it's really well balanced between the two and doing the "Milking contest" you could potentially get more water but also you could get less.

Also this way it wouldnt penalize if there is only one person at the pump.

BladeWoods · 2019-05-07 20:22:19

With option 3 water griefers could run the pump when people aren't ready, effectively destroying the towns water. And I don't really like the idea of every person in town stopping what they're doing to milk the water.

Between these three options, I would vote for 2.

Can make the water filled tank super slow to move or maybe unmovable to prevent people from effectively using it for water transportation. The recipes for making the different sized tanks could maybe involve combining a different number of buckets with the smaller tanks upgradeable to the larger tanks.

Potjeh · 2019-05-07 20:24:36

Yeah 2 seems best.

FeignedSanity · 2019-05-07 20:57:11

Option 2 has my vote

wio · 2019-05-07 21:50:29

I don't understand how 1 or 2 affect variance. All they seem to do is give insight as to how much water is remaining.

I wish I understood what the goal here was. I've read the variance thread but even then I found the goal to be a bit unclear. The variance thread looked like there was a desire to entertain probability solutions, but the goal wasn't clear there either.

Option 2 seems to be the most reliable solution provided, but I don't know what it is solving.

Twisted · 2019-05-07 22:04:02

wio wrote:

I don't understand how 1 or 2 affect variance. All they seem to do is give insight as to how much water is remaining.
I wish I understood what the goal here was. I've read the variance thread but even then I found the goal to be a bit unclear. The variance thread looked like there was a desire to entertain probability solutions, but the goal wasn't clear there either.

The goal is to make pumps run out as to necessitate upgrading them to a more advanced level. Currently, the Newcomen pump has a 1/20 chance to exhaust on use. This gives it 20 uses on average, but every now and then it will break down on first use, which is a really feelsbadmoment and can completely kill off the village. Every tile can have charges (number of times it can be used before transitioning to the next stage), so ideally we'd use those (similar to steel tools or seed bowls), but that's currently not possible as the charge system is already used for taking out buckets of water - every time you fire up the engine it gets X charges of bucket uses.

The idea is to find a way to make the pump tile not use charges for bucket uses so we can use those charges to set the number of uses.

Thaulos · 2019-05-07 22:13:46

If you make it so that the exhausted regenerates after a long time it would still be enough incentive to upgrade to diesel asap but still provide some couching for odd occurrences like the 1 in 20.

Alternatively you could just make the process of gathering water longer. ie, increasing the firing time from 15s to 60s or even 120s. Would still provide some water but in a very slow rate.

The above would also encourage the diesel burner option as it would still take a long time but would increase the yield during that time.

Last edited by Thaulos (2019-05-07 22:19:09)

futurebird · 2019-05-07 22:24:27

I like the slowing down idea, Thaulos! It's better than it just breaking IMO.

jasonrohrer · 2019-05-07 23:10:48

Imagine you have a D24 die. You have something special happen if you ever roll a 1 (like a pump exhausts).

You ask "How many rolls do I expect, on average, before I roll that 1?"

The answer is 24. That's the Expected Value, or the Mean, or the Average. That tells you something important about this probability distribution.

But that only gives you one part of the picture. The Variance is how spread out the distribution is around that Expected Value, essentially "what is the average distance away from the expected value."

In the case of the "how many rolls until 1" question with the D24, the variance is really high: 552. It's more useful to talk about the standard deviation, which is the square root of the variance. In this case, 23.5. I.e, the entire range of results is within one standard deviation.

We can also ask, "How many people can we have start rolling the D24 in parallel before we expect one of them to roll 1 on their first roll?" The answer to that is also 24 people.

So while each person in this group will on average roll 24 times before seeing 1, in the worst case, we expect someone to roll 1 on their very first roll. The variance is huge, after all.

And currently, when we want to decide whether the pump becomes exhausted, we roll a D20 and look for a 1. This means the pump will run 20 times on average before exhausting, but the variance is huge. We expect one out of every 20 pumps to fail after only one pumping.

One nice method for reducing variance is to split the single random variable up into several smaller random variables and sum them.

So, instead of rolling one D24, we can use four D6 dice. We start with the first die, and roll it until we see a 1. Then we move onto the next die and roll until that one sees a 1. Keep going until all four D6 dice have rolled a 1.

We expect it to take 6 rolls before we see a 1 each time, and all the rolls are independent, so we expect 24 rolls total before we see four 1s.

Thus, the expected value is the same: 24 rolls.

But the variance is way smaller. The variance for each die, in the number of rolls before 1, is 30 (standard deviation 5.47). To get the variance of all four, we sum them, so the total variance is 120 (standard deviation 10.95). This is way smaller than our D24 sd of 23.5.

And obviously, no one hits the target condition after their first roll. It takes four rolls minimum, so that's better right off the bat.

But how many people do we need rolling dice in parallel before we expect one of them to roll the target condition after their first four rolls? The worst case... that's (1/6)^4

We need 1296 people before we expect one to roll the worst case of four 1's in a row. Compare that to 24 people when rolling a single D24.

Figuring out exactly what these probability distributions look like is a little more complicated. The number of rolls of a single D24 before you see a 1 is a geometric distribution. Summing geometric distributions together gives us a "negative binomial" distribution. You can see the result of that here, in the animated gif:

https://en.wikipedia.org/wiki/Negative_ … stribution

The r=1 version of negative binomial is geometric. As r goes up, you can think about rolling more dice, one at a time, until each die sees a 1. You can see how the variance decreases, hugging the expected value tighter and tighter.

Anyway, reducing the variance like this for a pump requires that we have multiple tiers of the pump, and roll a smaller die until each tier exhausts. But the "tiers" are currently being used to track water taken out of the pump, instead of overall pump age.

(FYI, posts like this are mostly for me to clarify my understanding of the topic.... I'm not necessarily trying to explain this to anyone. I didn't know that the sum was a negative binomial distribution until just now).

One Hour One Life Forums

#1 2019-05-07 17:41:03

The three options for fixing the variance on pump exhaustion

#2 2019-05-07 17:55:14

Re: The three options for fixing the variance on pump exhaustion

#3 2019-05-07 17:56:52

Re: The three options for fixing the variance on pump exhaustion

#4 2019-05-07 17:57:34

Re: The three options for fixing the variance on pump exhaustion

#5 2019-05-07 18:00:21

Re: The three options for fixing the variance on pump exhaustion

#6 2019-05-07 18:04:02

Re: The three options for fixing the variance on pump exhaustion

#7 2019-05-07 18:05:14

Re: The three options for fixing the variance on pump exhaustion

#8 2019-05-07 18:08:39

Re: The three options for fixing the variance on pump exhaustion

#9 2019-05-07 18:11:37

Re: The three options for fixing the variance on pump exhaustion

#10 2019-05-07 18:24:46

Re: The three options for fixing the variance on pump exhaustion

#11 2019-05-07 18:25:55

Re: The three options for fixing the variance on pump exhaustion

#12 2019-05-07 18:28:44

Re: The three options for fixing the variance on pump exhaustion

#13 2019-05-07 18:31:32

Re: The three options for fixing the variance on pump exhaustion

#14 2019-05-07 18:51:45

Re: The three options for fixing the variance on pump exhaustion

#15 2019-05-07 19:05:31

Re: The three options for fixing the variance on pump exhaustion

#16 2019-05-07 19:13:34

Re: The three options for fixing the variance on pump exhaustion

#17 2019-05-07 19:17:42

Re: The three options for fixing the variance on pump exhaustion

#18 2019-05-07 20:22:19

Re: The three options for fixing the variance on pump exhaustion

#19 2019-05-07 20:24:36

Re: The three options for fixing the variance on pump exhaustion

#20 2019-05-07 20:57:11

Re: The three options for fixing the variance on pump exhaustion

#21 2019-05-07 21:50:29

Re: The three options for fixing the variance on pump exhaustion

#22 2019-05-07 22:04:02

Re: The three options for fixing the variance on pump exhaustion

#23 2019-05-07 22:13:46

Re: The three options for fixing the variance on pump exhaustion

#24 2019-05-07 22:24:27

Re: The three options for fixing the variance on pump exhaustion

#25 2019-05-07 23:10:48

Re: The three options for fixing the variance on pump exhaustion

Board footer