Sounds like you did not expect your viewers to be a bunch of nerds. Really? :) We watch 2d cars moving to 2d buildings while the dude at the steering wheel considers lots of tweaks in terms of nitty-gritty optimizations to their flow. That is a fun part of it, and we (well, I for sure) really appreciate the effort you put into it to like analyze whats really going on.
Love how this masterpiece of a comment has nearly 100 likes, and yet no comments… yet, for I have taken that spot. Also, you sure got me in you’re team. Nerds forever, mate.
In this video, Retroflux defends his dissertation on the M.O.V.E Principal of Mini Motorways by using statisical analysis to prove his shared trip hypothesis like the absolute genius he is. Well done!
I'm still waiting for someone to come by and refute my claim with even more scientific rigor and everyone will laugh at me and my crackpot theory... You know, just normal fears lmao
Me sitting here like “this video is awesome, let’s find part 2” and then realise that this video was only just uploaded and I’ll have to wait for the next one. I’m too used to Netflix’s “dump the entire season at once” structure haha
Me, after spending 20 hours trying to make content: "Phew, time for a brea-" My fans: "MORE" I love you guys, but chill LOL Hit the subscribe button and I swear you'll get the notification :')
AHHH I FUCKING KNEW IT! I had a hunch that destination demand would "overflow" into other destinations (of probably the same color?) since in my playthroughs and other let's plays I keep noticing that if one destination starts struggling other destinations of that same color start struggling. It makes sense as a snowball-y game mechanic to keep players on their toes, but I never figured we'd get concrete proof that this mechanic exists beyond dev notes. Thank you for all your hard work analyzing the game to prove and spot this!
Absolutely. It seems there is always one color in the game that becomes a problem about midway through an then ultimately ends the game. I would be interested to see if the spawn rate of house colors follows a predictable path based on the requirements of the shops.
I had an investigate of my own, looking through code, and you're absolutely right. If a destination is maxed out, the game tries to put the new pip on the nearest non-full destination of the same colour. If all destinations of that color are full, it puts it on the nearest non-full destination of any colour.
I'd wager that this is the result of the changes they made to punish players for blocking too much of the map and preventing destinations from spawning; the pin spawn rate per colour escalates over time based on how many destinations are *expected* to exist. If you allow destinations to spawn normally, then the number of destinations increases in tandem with the demand and pins are distributed among them in a manageable way. If you block destinations from spawning, then the demand at the ones that did spawn will become overwhelming.
mind blown.... I always thought it strange how small destinations randomly came under strain late game. They were being shafted pips from their struggling friends!
"We're an equal opportunity employer, we shaft all of our employees equally hard" "Yes Hans, the big pipe for the small structure" " What are you doing, step-store? " I could do this all night 😂
I think if you wanted to confirm the ”pips placed in other shop of the same color” hypothesis, it should be rather simple in the mid-game. Simply disconnect all but one of the shops of the same color and see what happens when they’re full. It should be quite easy to determine if the single shop remaining is getting all the pips or if they’re being distributed to other colors too. Great video, thanks for your effort! :)
Thanks for your hard work ! Of course, looking at the code would be easy enough, but I enjoy more the reverse engineering part, what hypothesis you come up with from observing the game behaviour and how you test it. I'm very interested in the two ones you tested and proved wrong. "It's about the journey, not the destination" if I dare say x')
So in conclusion, you need to anticipate square dests both expanding or becoming overflow ones when the round ones max out. Otherwise your scores will never break through that threshold.
Had to check if someone had already mentioned that 6:37 thing, but they did and you already acknowledged it, so yeah, this is just for the algorithm lol Also I just subscribed because I am addicted to this game and you are enabling me.
I’m incredibly bad at math and always wonder about this kind of stuff, it’s cool to see someone do all the calculations and get into technical stuff of games.
I love it, I have often spent too much time recording data for things to try and understand how it works so as to optimise but largely just to understand. Unfortunately, I don't have the ability to analyse the data and consolidate it into graphs and charts, as well as you do. If you maxed out the games that you counted, to what score did you count? It seems to max out at around 10000 but the real increases in pip generation might max out before I feel sure of it and the last part is the city balancing out, including me making final adjustments to any areas that are still in danger.
The max score that I had to work with was 11355 - houses were still spawning and ruining my day so I get the feeling it keeps going from there. Next time I see an infinite score video pop up, I'll see if I can use that as my footage... but it might be impossible to get infinite scores now if they changed the growth rate too much, so we'll see what shakes out. 10K for most maps is still hopefully possible! :)
This type of analysis has always been on my mind while playing this game. I searched google for this kind of in depth mathematical analysis of car rates and road optimisation, not just which thing is better, but by *how much.* Alas, I could not find any results. Please continue this series!
Very interesting, so often the colour going bad appears to be a result of cars getting lost, or accidental connections causing ridiculously long journeys, because they don't update when you fix things. Just today, I realised by not recording my games to play back, that I was having to guess at the cause of meltdowns. Sharing demand is something experienced players mention when considering blocking. Basically if you over be do it, demand ramps up on other dests and you haven't the houses to satisfy it. Thanks for looking be at this systematically!
Yep! I figured there had to be something underlying it all, so I went to work figuring it out... unfortunately what didn't get into the video was me testing three other hypotheses that turned out to be incorrect... ten hours wasted, but it led to this so all is good! I get the feeling that post-patch, the demand curve is even steeper, that's the only way that I can see them capping infinite runs... but that's content for another video :^)
I botched so many runs because I connected two sections to have one or two houses add to a mall but end up accidentally creating a larger networking that makes me lose. One might think I have learned from those but nope, not yet I guess >_>
@@SETHthegodofchaos It's easily done,. Yesterday a very promising map with good colour segregation and many motorways broke down when I had to find 5 houses for a new round dest that appeared without more houses spawning. I was short on that colour because it was round only AND I'd had the housing groups all used in some way. I made an accidental connection in one area, then by hooking up some different colour houses, which would be needed soon, it created ridiculous multi motorway routes. As I cut road to fix, the deletion bug hit and cars of multiple colours became lost and went sight seeing. Now we know the lost cars created extra demand on unrelated dests, which meant everywhere was short of cars and with nothing in garages, there was nothing but crossing fingers to fix. It's really frustrating when you're high in the Weekly challenge but want to up the score to be top 100 at the end of the week, because such games aren't quick endeavours, with interruptions a lot of effort had gone into it. I was very near being able to fill the map, there was room to give me more houses but it has a way of getting nasty!
Thanks for the sub! Hopefully you enjoy the latest video, it's a nother deep dive! Less graphs though, since I was just playing with raw numbers and frame counting for this one. The next one should have a bit of graphing, though! :)
Loosing on level one buildings that looked fine definitely did puzzle me, but this explanation makes sense! Maybe that is also the reason why cars from really far away try to go to the small one if at all possible even though it'd take a tremendous amount of time? It could be a car that was supposed to go to a big one but chooses another destination because it is full.
Fascinating! Looking at the end of game screenshot I took of my last run, I'm pretty sure I can see this pip sharing in action. Thank you for including subtitles too!
:D I try and CC all of my talking videos, the hour-long ones are just not feasible to do manually so I pray the autogenerated ones are good enough most of the time :p Glad you liked the video!
I've noticed the lack of control as to where cars end up in interconnected systems. They don't necessarily go to the most urgent spot which is a strange effect and I'd be curious to figure out why that happens. It makes it difficult in critical scenarios to predict whether a suplimental source of cars will have the desired effect. Looking forward to your next video!
As best as I can tell, pips are assigned to the closest available car. And once a car takes off on a journey to a specific pip, it will continue no matter what. If you have a really urgent building and connect houses close to it, you'll see instances where cars drive right past them, on a fruitless journey across the map. That's because the pips for that angry building already have cars on the way to get them (and the cars for the close houses are across the map, completing journeys of their own).
yes!! finally! someone is taking this game this seriously! havent even played but watched tons of videos and i bet using methods you used today, you could get a perfect game!
It's still speculation! But I think it makes sense, and it gives me a new angle to play the game with :) once I'm done reverse engineering, I'm gonna stream the source code and try and make sense of it all
@@Retroflux Completely agree, I think you're on to something! I played the weekly (dubai) after watching the video and managed a PB of 10.2k. I tried to go for maximum efficency for the level 2 buildings, min 6 houses each and then just maximum amount of "backup" cars for the level 1s whilst not worrying too much about efficiency (route length) and it seemed to help. Alas, a level 1 did fail me in the end though but I'm sure like you said it's because the L2s got overrun and loaded up the L1s
Earned my subscription. I'm excited to see where this goes! I personally haven't done any math (obviously), but I have gotten to 6000+ on LA by minimizing crossroads at all costs, even in neighborhoods. Since the big change to traffic at the end of August, I've instead been minimizing the total number of intersections, as cars slow down about the same for a 4-way as a 3-way intersection, not including the roads connected directly to houses.
This was an amazingly in depth analysis of a simplistic 2d video game about roads, and I loved every second of it. Can't wait for the rest of the series
Interesting video, couldn't help but notice at 6:30-6:40 when you were explaining your maths, you said the sum was 4(1.4935(26 weeks) + 9.2863)) however in the video its labelled as 4.....+0.2863) could this have also affected your mathematical reasoning on the sum, as it would have made your count 165 instead of 156, closer to the 162 manual count you had.
Hi hi! The written equation is correct, it's just a enlarged, worked version of the equation on the graphs. Someone mentioned that annotating might be a good idea, but I think that feature has been removed. 156 was correct.
This is great, really getting into the nitty-gritty of the games stats and such (I do something similar) and I'm glad to see someone looking at the math of it all. 👍
By cars movement through intersections I may assume this recordings from previous patch. In new patch they drastically changed pins distribution and more important how anger system works. Now in late game some colors can produce just a few pins while others might be flooded. On such colors you physically cannot fulfill the demand. Because of anger. It accumulates more quickly. That's why no infinite runs on TH-cam on last two weeks. So, your math is good, but it might be outdated.
The pip counting is pre-patch yes, but not all the footage is... I think. I'll definitely have to follow-up if it's different, though I would be surprised if they did a major overhaul to their base system functionality - maybe just a faster growth slope, for example. I'll have to get another 10K run to double check though, so maybe that's up next!
I do think the pin distribution system was changed drastically. When the game hits 3K or 4K trips, it now feels impossible to meet all of the demand by the stores. But I don’t believe it’s just a feeling, especially when I have most of the cup de sacs have private motorways to the entrance of the stores (if not already really close to the stores) and still, they are not satisfied. Either way, Retroflux I thoroughly enjoy this video and can’t wait to see more analysis on other aspects of the game too.
Amazing data, thanks for the work so far! I'd love to set up a discord to talk about this games meta so we could push scores higher and higher. I'd gladly lend a helping hand in any data needed to be collected!
The pain you went through is why I hit sub without even thinking twice. That's not to say I wish to see you in more math pain but I'm not saying that I don't wish to see you in math pain...
It seems like every house is connected and regardless of color. It is easy to see when blocking spawning points so you have only one destiantion - this destiantion is so busy that you can't keep up. When you allow another destination to spawn not matter the color the it becomes much easier.
Hypothetically: Imagine a level 2 building with 10 houses very close and efficiently (roundabouts) connected. Would this help another house of the same colour from every fully capping out? Rephrased: You spoke of pip sharing as detrimental to other houses of the same colour, but is it possible to utilise the pip sharing mechanic to focus more resources into a single building of the same colour to become the designated pip-sharer?
Assuming the efficiently connected level 2 building has a car turnover rate that exceeds its current MOVE criteria, yes it would in theory start pulling from other structures, increasing its efficiency curve. The problem of the other structures still remains and can still end the game, though - remember the timer is for too many pips, so if you never clear them... You're going to lose eventually. But tl;dr yes that's exactly what I mean by pip sharing
Damn from watching your DD no deaths to this... you are definitely an underrated channel for sure! So I'm glad to see you grow from about 2k subs from when I started watching to over 6k now
Great video Retro! It's interesting to learn that the destinations share the load, can't wait for the followup to see what else we learn about this charming puzzle game.
Ever since I came across this game, I've always wanted to construct some sort of mathematical model based on a system of (more likely than not) coupled differential equations. I've been thinking that I could try to model the destinations or maybe the roadways as a capacitor of some sort given that there will be some sort of car flux associated with each tile. I was never really able to figure it out as it's a hell of a wild system to even attempt to model - especially given that we're also dealing with a weird metric space considering the form of any of the maps. I look forward to other videos in this installment.
yaaas this is the kind of systemic analysis I've been hoping to see. From my own testing, it does seem that the rate of pip generation is global and scales up over time (IDK if this is a global value per colour but I think probably not, at least to some extent). I tested this with the usage of diagonal road preventing new destinations from spawning. The houses simply failed to keep up with the ramping demand, however, when I "allowed" a new business to spawn (ie; I missed a gap and let one slip in), it seemed that it was more able to distribute the pips, lessening the load on the other structures. The way in which this happened seemed to me like there was a hidden pip-generation penalty value when a business was unable to spawn to deter this method specifically, but this would need more thorough testing to confirm. It might be possible that this hidden penalty is what causes the game to end in a loss eventually, since it becomes inevitable that no new business will be able to spawn when enough houses and roads exist. The best way to test this theory would be to inject a custom map into the game that is effectively the same as any other map, but has much more playable space, to determine if the destinations have a hard limit (either globally, or per map), or if it is an implied limit as an emergent property of how the game works. I look forward to seeing more deep analysis, consider me subbed
The rate of pip generation could be color dependent and tied to the total number of corresponding houses, and then cycled through the associated businesses. Once no more house can spawn the game can run for ever provided the demand can be met. House spawn happens on business spawn or upgrade and continuously so blocking businesses altogether is counter productive. Also, there is a hidden zoning on maps that dictates where houses and businesses will spawn more frequently, there is a video of the dev talking about the development process that show that, maybe it could be abused to control house spawn. Houses tend to flock together, so manipulating the area around neighborhoods with roads to control house development patterns could help.
Just found out about this game tonight. This is the most satisfying and calming shit to just put on that I've seen in a long time. Granted, I'm not the one playing. Haha
It is pretty deep, for a generally simple concept! Glad that my anxiety doesn't come through and it's calming for you LOL, the game can be pretty unnerving at times if you're the one playing it!
Bless you for your algorithm boost, regardless! And remember a dead subscriber is still helpful in their own way! Some people don't take small channels seriously, especially companies >:(
it *is* within colour group by the way u can see it in maps with 2 of a certain colour spawning constantly, when only one spawns then it will almost definitely be your doom
oh wonderful! really good work. so i bet the code for this is literally when increasing demand in a building, "if this building is currently full, pick a different building". i wonder how the different building is chosen, if that's always one with the lowest load perhaps
given the cyclic behaviour seen in the early game, it could be a linked list, where you pick the first item in the linked list that isn't full, and move it to the end
Yeah, that's my first thought too, something that could easily be added to that maintains an order, sounds like a LL or a derivative like a queue! What might also exist is a "colour counter" variable that keeps track of excess, and then dumps it if space opens up.
would you ever want to help me collect data and equip this game with a machine learning optimization program to see what the theoretically optimal strategy is?
Having a shared spawnrate among the same color does make sense. It would be a very simple way to load balance the system. While hte game does have a fail condition, it would be very frustrating if you lost because one building *randomly* got 16 drive requests in a few seconds. Or if htere was ever a journey that takes too long to finsh. Consider two lists that are itterated over. If you pass the last entry, restart at front. One list has the houses. One list has the destinations. The game tries to match the current house to the current target. If the house has no cars, it has to skip the house. If the target is too far from the house or the target is already full, it has to skip the target. If there is no valid target for this house (isolated with no empty target), it skips the house. Otherwise, you get a car journey. For bonus points, every X cycles of either list you get a new house and/or a new target.
@@Retroflux Another thought crossed my mind: Maybe there is a spawnrate shared between all colors as well? Like on top of itterating through houses and target buildings, is also itterating through colors would mean that one extra color do not adds (many) extra spawns - just more complex pathfinding. But the time between spawnings might also decrease over time. Maybe starting at something like 5000 game ticks, counting down by 1 for each new house spawned.
Each house generates some number of pips per minute The pip is placed on a destination of the same color that is a) not full and b) is the destination that has not received a pip in the longest duration There are some variations that were too complicated to pick up on without detailed analysis, but I do know that trip generation rates will skyrocket if a new destination cannot be placed for a while
@@Retroflux yaas! if houses are the ones generating pips it could also be "Jim? they haven't opened a new Redmall™ in three days!" "What? No!" "I'll call the rioters..."
if counting pips is something that interests people you can use cheatengine to slow the game down (speedhack, set it to 0.5 or 0.25 or whatever) - then when you play it back you'll have double or quadruple the frames for each pip event or whatever. Yes, it's a slog to play, but this is about math, dangit.
your data of this game also canbe used to know how hard the devs do math to make this thing hapens and we can also know how all ai was coded in this game
Hopefully they followed Occam's Razor and just did the simplest solution, though there's lots of little things here and there... The more I dig, the less I think it's simple.
That’s an interesting question. In my experience with the game, I had much more success not sharing loads solely because it can cause too many intersections. I think the downsides of the intersection outweigh any benefits of sharing load, but I’m not sure on this.
Everything in this world is unnecessary if you really look into it too much, but that's where the joy of life is! Enjoy the unnecessary, it's the nectar of life :)
Thanks for putting in the work but ah linked spreadsheets would be nice to get a fuller understanding of your findings. I think others may find other conclusions in the stats as well. From what I've gathered they are connected by colour but does colour have a specific amount? I doubt it although some colours are certainly more active than others.
Some day in the future I may stream! But for now, I am humbled by the love of everyone so far, so I want to take no donations. Your subscription and gospel of my channel is more than enough for me, 100%. Money comes and goes, but what matters at the pearly gates is what you did with your life.
![[Weekly] I'll Never Make Fun of Traffic Lights Again - Mini Motorways](http://i.ytimg.com/vi/TRRwm_EuqrY/mqdefault.jpg)
![[Weekly] I'll Never Make Fun of Traffic Lights Again - Mini Motorways](/img/tr.png)
![[Top 2%] Conquering Mexico City's Challenge on First Attempt: Tunnels and Turnarounds](http://i.ytimg.com/vi/_Kh57JLMJPM/mqdefault.jpg)
Sounds like you did not expect your viewers to be a bunch of nerds. Really? :) We watch 2d cars moving to 2d buildings while the dude at the steering wheel considers lots of tweaks in terms of nitty-gritty optimizations to their flow. That is a fun part of it, and we (well, I for sure) really appreciate the effort you put into it to like analyze whats really going on.
Love how this masterpiece of a comment has nearly 100 likes, and yet no comments… yet, for I have taken that spot. Also, you sure got me in you’re team. Nerds forever, mate.
Look I’m not a nerd. I’m just a data scientist who really likes optimization games.
Ah yes, my favorite topic in games: MATH & STATISTICS
(for real, I love those in-depth analysis types of videos about games)
Ok matpat
Jk I actually love that as well
@@orangegaming9562 Yeah, that's also why I like GT/FT by Mat. It's nerding on a whole nother level. :D
Who'd have thought it... The destination does not request the visitor, the visitor requests the destination.
Wow thanks for summarizing all of my work into 16 words hahahaha... oof.
@@Retroflux just the conclusion, not the work. The work is the fountain :)
In this video, Retroflux defends his dissertation on the M.O.V.E Principal of Mini Motorways by using statisical analysis to prove his shared trip hypothesis like the absolute genius he is. Well done!
I'm still waiting for someone to come by and refute my claim with even more scientific rigor and everyone will laugh at me and my crackpot theory...
You know, just normal fears lmao
@@Retroflux At least you moved the conversation forward intelligently. Bravo, sir.
This is the deep dive I hoped the algorithm would eventually serve up!
Me sitting here like “this video is awesome, let’s find part 2” and then realise that this video was only just uploaded and I’ll have to wait for the next one. I’m too used to Netflix’s “dump the entire season at once” structure haha
Me, after spending 20 hours trying to make content: "Phew, time for a brea-"
My fans: "MORE"
I love you guys, but chill LOL Hit the subscribe button and I swear you'll get the notification :')
@@Retroflux All good! Take your time, I know this kind of content isn't easy to make.
This is why strategy/puzzle gamers scare me. They will stop at nothing to find the optimal exploits of they’re favorite systems
Puzzle gamers rise up
AHHH I FUCKING KNEW IT! I had a hunch that destination demand would "overflow" into other destinations (of probably the same color?) since in my playthroughs and other let's plays I keep noticing that if one destination starts struggling other destinations of that same color start struggling. It makes sense as a snowball-y game mechanic to keep players on their toes, but I never figured we'd get concrete proof that this mechanic exists beyond dev notes.
Thank you for all your hard work analyzing the game to prove and spot this!
Absolutely. It seems there is always one color in the game that becomes a problem about midway through an then ultimately ends the game. I would be interested to see if the spawn rate of house colors follows a predictable path based on the requirements of the shops.
@@amasterofone about 3000+ this trend starts
I had an investigate of my own, looking through code, and you're absolutely right.
If a destination is maxed out, the game tries to put the new pip on the nearest non-full destination of the same colour. If all destinations of that color are full, it puts it on the nearest non-full destination of any colour.
Stop, my ego can only get so big
How do you look at the code?
I'd wager that this is the result of the changes they made to punish players for blocking too much of the map and preventing destinations from spawning; the pin spawn rate per colour escalates over time based on how many destinations are *expected* to exist. If you allow destinations to spawn normally, then the number of destinations increases in tandem with the demand and pins are distributed among them in a manageable way. If you block destinations from spawning, then the demand at the ones that did spawn will become overwhelming.
mind blown.... I always thought it strange how small destinations randomly came under strain late game. They were being shafted pips from their struggling friends!
"We're an equal opportunity employer, we shaft all of our employees equally hard"
"Yes Hans, the big pipe for the small structure"
" What are you doing, step-store? "
I could do this all night 😂
@@Retroflux oh that became much more graphic than necessary, I feel you've violated all my future gameplay in mini motorways.
@@Retroflux 😂
Damn, math, smashed both the buttons to honour the work!
This is the kind of thing I've been sitting on doing since I started playing. Thank you for putting the work in.
I think if you wanted to confirm the ”pips placed in other shop of the same color” hypothesis, it should be rather simple in the mid-game. Simply disconnect all but one of the shops of the same color and see what happens when they’re full. It should be quite easy to determine if the single shop remaining is getting all the pips or if they’re being distributed to other colors too. Great video, thanks for your effort! :)
You have more than earned your like and subscribe, probably going to end up watching more of your content :)
That's the dream! Welcome to the pond :)
Thanks for your hard work ! Of course, looking at the code would be easy enough, but I enjoy more the reverse engineering part, what hypothesis you come up with from observing the game behaviour and how you test it. I'm very interested in the two ones you tested and proved wrong. "It's about the journey, not the destination" if I dare say x')
So in conclusion, you need to anticipate square dests both expanding or becoming overflow ones when the round ones max out.
Otherwise your scores will never break through that threshold.
Bingo bango bongo, give this man a prize
Maybe it's because I do math at university and love it, maybe I am just slowly going insane, but: I LOVED THIS VIDEO :p
Por que no los dos? (why not both?) :^)
Had to check if someone had already mentioned that 6:37 thing, but they did and you already acknowledged it, so yeah, this is just for the algorithm lol Also I just subscribed because I am addicted to this game and you are enabling me.
Yep! Idk why I said "9", literally the graph right beside it has the formula...
This is just what I needed thank you so much 😊
Glad you enjoyed it! ☺️
I’m incredibly bad at math and always wonder about this kind of stuff, it’s cool to see someone do all the calculations and get into technical stuff of games.
I am also incredibly bad at math, but I'm more persistent than bad so it works out lol
I love it, I have often spent too much time recording data for things to try and understand how it works so as to optimise but largely just to understand. Unfortunately, I don't have the ability to analyse the data and consolidate it into graphs and charts, as well as you do.
If you maxed out the games that you counted, to what score did you count?
It seems to max out at around 10000 but the real increases in pip generation might max out before I feel sure of it and the last part is the city balancing out, including me making final adjustments to any areas that are still in danger.
The max score that I had to work with was 11355 - houses were still spawning and ruining my day so I get the feeling it keeps going from there. Next time I see an infinite score video pop up, I'll see if I can use that as my footage... but it might be impossible to get infinite scores now if they changed the growth rate too much, so we'll see what shakes out. 10K for most maps is still hopefully possible! :)
This type of analysis has always been on my mind while playing this game. I searched google for this kind of in depth mathematical analysis of car rates and road optimisation, not just which thing is better, but by *how much.* Alas, I could not find any results. Please continue this series!
Very interesting, so often the colour going bad appears to be a result of cars getting lost, or accidental connections causing ridiculously long journeys, because they don't update when you fix things.
Just today, I realised by not recording my games to play back, that I was having to guess at the cause of meltdowns.
Sharing demand is something experienced players mention when considering blocking. Basically if you over be do it, demand ramps up on other dests and you haven't the houses to satisfy it.
Thanks for looking be at this systematically!
Yep! I figured there had to be something underlying it all, so I went to work figuring it out... unfortunately what didn't get into the video was me testing three other hypotheses that turned out to be incorrect... ten hours wasted, but it led to this so all is good! I get the feeling that post-patch, the demand curve is even steeper, that's the only way that I can see them capping infinite runs... but that's content for another video :^)
@@Retroflux Now testing the other hypotheses and showing why you reject your idea in the face of facts, sounds like a vid the world sorely needs!
@@Retroflux 10 hours just to confirm that the hypothesese were wrong. oof. I guess thats the life of a scientist/reverse engineer :D
I botched so many runs because I connected two sections to have one or two houses add to a mall but end up accidentally creating a larger networking that makes me lose. One might think I have learned from those but nope, not yet I guess >_>
@@SETHthegodofchaos It's easily done,. Yesterday a very promising map with good colour segregation and many motorways broke down when I had to find 5 houses for a new round dest that appeared without more houses spawning. I was short on that colour because it was round only AND I'd had the housing groups all used in some way.
I made an accidental connection in one area, then by hooking up some different colour houses, which would be needed soon, it created ridiculous multi motorway routes.
As I cut road to fix, the deletion bug hit and cars of multiple colours became lost and went sight seeing.
Now we know the lost cars created extra demand on unrelated dests, which meant everywhere was short of cars and with nothing in garages, there was nothing but crossing fingers to fix.
It's really frustrating when you're high in the Weekly challenge but want to up the score to be top 100 at the end of the week, because such games aren't quick endeavours, with interruptions a lot of effort had gone into it. I was very near being able to fill the map, there was room to give me more houses but it has a way of getting nasty!
just subbed, looking forward to your analysis of data and models you create. would love to see some more graphs/data representation on screen tho
Thanks for the sub! Hopefully you enjoy the latest video, it's a nother deep dive! Less graphs though, since I was just playing with raw numbers and frame counting for this one. The next one should have a bit of graphing, though! :)
I play this game too much, my first thought was to move purple to the right entrance and green to the left to save roads lol
Haha, true. Also to have more room for spawns and other roads :D
Same…
Loosing on level one buildings that looked fine definitely did puzzle me, but this explanation makes sense! Maybe that is also the reason why cars from really far away try to go to the small one if at all possible even though it'd take a tremendous amount of time? It could be a car that was supposed to go to a big one but chooses another destination because it is full.
Fascinating! Looking at the end of game screenshot I took of my last run, I'm pretty sure I can see this pip sharing in action. Thank you for including subtitles too!
:D I try and CC all of my talking videos, the hour-long ones are just not feasible to do manually so I pray the autogenerated ones are good enough most of the time :p Glad you liked the video!
I subscribed when he said “if you can’t M.O.V.E it, you lose it.”
I've noticed the lack of control as to where cars end up in interconnected systems. They don't necessarily go to the most urgent spot which is a strange effect and I'd be curious to figure out why that happens. It makes it difficult in critical scenarios to predict whether a suplimental source of cars will have the desired effect. Looking forward to your next video!
As best as I can tell, pips are assigned to the closest available car. And once a car takes off on a journey to a specific pip, it will continue no matter what. If you have a really urgent building and connect houses close to it, you'll see instances where cars drive right past them, on a fruitless journey across the map. That's because the pips for that angry building already have cars on the way to get them (and the cars for the close houses are across the map, completing journeys of their own).
yes!! finally! someone is taking this game this seriously! havent even played but watched tons of videos and i bet using methods you used today, you could get a perfect game!
That's the basis for my 10K on each map long-term goal!
I highly appreciate the effort you put into this video! it was awesome! :)
Thank you so much!
Thank you for doing this work! I have honestly lost sleep thinking about why my square shops keep failing me!!
It's still speculation! But I think it makes sense, and it gives me a new angle to play the game with :) once I'm done reverse engineering, I'm gonna stream the source code and try and make sense of it all
@@Retroflux Completely agree, I think you're on to something! I played the weekly (dubai) after watching the video and managed a PB of 10.2k. I tried to go for maximum efficency for the level 2 buildings, min 6 houses each and then just maximum amount of "backup" cars for the level 1s whilst not worrying too much about efficiency (route length) and it seemed to help. Alas, a level 1 did fail me in the end though but I'm sure like you said it's because the L2s got overrun and loaded up the L1s
Earned my subscription. I'm excited to see where this goes! I personally haven't done any math (obviously), but I have gotten to 6000+ on LA by minimizing crossroads at all costs, even in neighborhoods. Since the big change to traffic at the end of August, I've instead been minimizing the total number of intersections, as cars slow down about the same for a 4-way as a 3-way intersection, not including the roads connected directly to houses.
This was an amazingly in depth analysis of a simplistic 2d video game about roads, and I loved every second of it. Can't wait for the rest of the series
When you're playing spreadsheet instead of the game itself
The first video on this channel was spreadsheets, I've come full circle at this point
@@Retroflux Made me look. Damn.
Interesting video, couldn't help but notice at 6:30-6:40 when you were explaining your maths, you said the sum was 4(1.4935(26 weeks) + 9.2863)) however in the video its labelled as 4.....+0.2863) could this have also affected your mathematical reasoning on the sum, as it would have made your count 165 instead of 156, closer to the 162 manual count you had.
Hi hi! The written equation is correct, it's just a enlarged, worked version of the equation on the graphs. Someone mentioned that annotating might be a good idea, but I think that feature has been removed. 156 was correct.
This is great, really getting into the nitty-gritty of the games stats and such (I do something similar) and I'm glad to see someone looking at the math of it all. 👍
Boy, I’m glad you did this math for me.
By cars movement through intersections I may assume this recordings from previous patch. In new patch they drastically changed pins distribution and more important how anger system works. Now in late game some colors can produce just a few pins while others might be flooded. On such colors you physically cannot fulfill the demand. Because of anger. It accumulates more quickly. That's why no infinite runs on TH-cam on last two weeks.
So, your math is good, but it might be outdated.
The pip counting is pre-patch yes, but not all the footage is... I think. I'll definitely have to follow-up if it's different, though I would be surprised if they did a major overhaul to their base system functionality - maybe just a faster growth slope, for example. I'll have to get another 10K run to double check though, so maybe that's up next!
Good luck with that. I'd really like to look at 10k+ run. I've had a number of infinite runs before, but now 10k is like space high for me.
I do think the pin distribution system was changed drastically. When the game hits 3K or 4K trips, it now feels impossible to meet all of the demand by the stores. But I don’t believe it’s just a feeling, especially when I have most of the cup de sacs have private motorways to the entrance of the stores (if not already really close to the stores) and still, they are not satisfied.
Either way, Retroflux I thoroughly enjoy this video and can’t wait to see more analysis on other aspects of the game too.
Amazing data, thanks for the work so far! I'd love to set up a discord to talk about this games meta so we could push scores higher and higher. I'd gladly lend a helping hand in any data needed to be collected!
You are such a mathematician, I love it!
Fun fact: I hate doing math, I'm just good at it lol
The pain you went through is why I hit sub without even thinking twice. That's not to say I wish to see you in more math pain but I'm not saying that I don't wish to see you in math pain...
It seems like every house is connected and regardless of color. It is easy to see when blocking spawning points so you have only one destiantion - this destiantion is so busy that you can't keep up. When you allow another destination to spawn not matter the color the it becomes much easier.
Hypothetically:
Imagine a level 2 building with 10 houses very close and efficiently (roundabouts) connected.
Would this help another house of the same colour from every fully capping out?
Rephrased: You spoke of pip sharing as detrimental to other houses of the same colour, but is it possible to utilise the pip sharing mechanic to focus more resources into a single building of the same colour to become the designated pip-sharer?
Assuming the efficiently connected level 2 building has a car turnover rate that exceeds its current MOVE criteria, yes it would in theory start pulling from other structures, increasing its efficiency curve. The problem of the other structures still remains and can still end the game, though - remember the timer is for too many pips, so if you never clear them... You're going to lose eventually.
But tl;dr yes that's exactly what I mean by pip sharing
Damn from watching your DD no deaths to this... you are definitely an underrated channel for sure!
So I'm glad to see you grow from about 2k subs from when I started watching to over 6k now
What I got from this video: trying to achieve 10k trips changes a person.
Jokes aside, really liked the video and look forward to more analysis:)
Stage 1: Euphoria
Stage 2: Depression
Stage 3: Min-Maxing with Spreadsheets
Stage 4: ???
Finally someone that brings real math into analyzing this game.
Great video Retro! It's interesting to learn that the destinations share the load, can't wait for the followup to see what else we learn about this charming puzzle game.
Ever since I came across this game, I've always wanted to construct some sort of mathematical model based on a system of (more likely than not) coupled differential equations. I've been thinking that I could try to model the destinations or maybe the roadways as a capacitor of some sort given that there will be some sort of car flux associated with each tile. I was never really able to figure it out as it's a hell of a wild system to even attempt to model - especially given that we're also dealing with a weird metric space considering the form of any of the maps. I look forward to other videos in this installment.
Liked, subbed and commented for the sheer amount of work put in :D
I subscribed and like the video cuz u spend 10 hours counting the beeps, thats dedication.
Somebody gotta
This is the only time I have ever commented on a video just because I like it. I don't know how to tell you we went this series enough.
Liked and subscribed looking forward to the follow up videos
NEEEEEERRRRRRRRRRD! that being said, thank you so very much from the bottom of my heart for this video
yaaas this is the kind of systemic analysis I've been hoping to see. From my own testing, it does seem that the rate of pip generation is global and scales up over time (IDK if this is a global value per colour but I think probably not, at least to some extent). I tested this with the usage of diagonal road preventing new destinations from spawning. The houses simply failed to keep up with the ramping demand, however, when I "allowed" a new business to spawn (ie; I missed a gap and let one slip in), it seemed that it was more able to distribute the pips, lessening the load on the other structures. The way in which this happened seemed to me like there was a hidden pip-generation penalty value when a business was unable to spawn to deter this method specifically, but this would need more thorough testing to confirm. It might be possible that this hidden penalty is what causes the game to end in a loss eventually, since it becomes inevitable that no new business will be able to spawn when enough houses and roads exist. The best way to test this theory would be to inject a custom map into the game that is effectively the same as any other map, but has much more playable space, to determine if the destinations have a hard limit (either globally, or per map), or if it is an implied limit as an emergent property of how the game works. I look forward to seeing more deep analysis, consider me subbed
The rate of pip generation could be color dependent and tied to the total number of corresponding houses, and then cycled through the associated businesses. Once no more house can spawn the game can run for ever provided the demand can be met. House spawn happens on business spawn or upgrade and continuously so blocking businesses altogether is counter productive.
Also, there is a hidden zoning on maps that dictates where houses and businesses will spawn more frequently, there is a video of the dev talking about the development process that show that, maybe it could be abused to control house spawn. Houses tend to flock together, so manipulating the area around neighborhoods with roads to control house development patterns could help.
Ooooh, that's just a cruel mechanics. Completely uncalled for
It's probably in response to all of the blocking strategies, now people can't go infinite
I liked the video, he went on his nah I’m not doing math, I hovered over the dislike, then he did the math. I guess I’ll sub.
IT WORKED MUAHAHAHA
err, I mean... welcome to the cult.
Wait that's not right either. I'll get back to you, we'll workshop it at evil HQ.
Appreciate you retro! Subscribed.
Appreciate you for appreciating me! Welcome to the pond... or something like that. We're still workshopping it.
Great Analysis!
Holy moly this changes everything...
Just found out about this game tonight. This is the most satisfying and calming shit to just put on that I've seen in a long time. Granted, I'm not the one playing. Haha
It is pretty deep, for a generally simple concept! Glad that my anxiety doesn't come through and it's calming for you LOL, the game can be pretty unnerving at times if you're the one playing it!
Duck mascot and statistics? You have my like and subscribe, my friend.
*quacking intensifies*
Im not gonna subscribe because i would just be a dead subscriber but have a comment and a like for the algorithm. Great video :D.
Bless you for your algorithm boost, regardless! And remember a dead subscriber is still helpful in their own way! Some people don't take small channels seriously, especially companies >:(
it *is* within colour group by the way
u can see it in maps with 2 of a certain colour spawning constantly, when only one spawns then it will almost definitely be your doom
WHAT!? I can't WAIT until the next one!
Great video. @6:37 you say 9.26 but it's written 0.26, might want to annotate that in some way if it's actually 9.26 or not.
Yes! The equation is right, I said the wrong thing. I will make sure to annotate soonTM
@@Retroflux no worries! Just trying to help, keep up the great videos :)
I've never even played this game, but damn all this math is intriguing me. Looking forward to more videos on this 👍
Why are there so many people here that have never played this game?!!? Hello, TH-cam? Glad you enjoyed it either way!
You should play, it's not expensive and it's available on Steam and Apple Arcade
"I refuse to do math for a game"
Stellaris and Victoria players:
oh wonderful! really good work. so i bet the code for this is literally when increasing demand in a building, "if this building is currently full, pick a different building". i wonder how the different building is chosen, if that's always one with the lowest load perhaps
given the cyclic behaviour seen in the early game, it could be a linked list, where you pick the first item in the linked list that isn't full, and move it to the end
Yeah, that's my first thought too, something that could easily be added to that maintains an order, sounds like a LL or a derivative like a queue! What might also exist is a "colour counter" variable that keeps track of excess, and then dumps it if space opens up.
This is amazing. Thank you!
You're very welcome!
would you ever want to help me collect data and equip this game with a machine learning optimization program to see what the theoretically optimal strategy is?
Email me? I'm super busy most days, but can at least offer guidance if you are interested in doing something like that.
Dude you my friend are crazy. And a fellow excel connoisseur :D
It would be interesting to count the number of houses, total demand could be tied to it.
Having a shared spawnrate among the same color does make sense. It would be a very simple way to load balance the system. While hte game does have a fail condition, it would be very frustrating if you lost because one building *randomly* got 16 drive requests in a few seconds. Or if htere was ever a journey that takes too long to finsh.
Consider two lists that are itterated over. If you pass the last entry, restart at front. One list has the houses. One list has the destinations.
The game tries to match the current house to the current target. If the house has no cars, it has to skip the house. If the target is too far from the house or the target is already full, it has to skip the target. If there is no valid target for this house (isolated with no empty target), it skips the house.
Otherwise, you get a car journey.
For bonus points, every X cycles of either list you get a new house and/or a new target.
I'd love to dissect that part of the game - the spawn rates and balance calculations must be pretty interesting to look at, if there even is any!
@@Retroflux I would look for some configuration file or settings.
Somehow I expect a simple/elegant design for the formulas. Nothing overly complex.
@@Retroflux Another thought crossed my mind: Maybe there is a spawnrate shared between all colors as well?
Like on top of itterating through houses and target buildings, is also itterating through colors would mean that one extra color do not adds (many) extra spawns - just more complex pathfinding.
But the time between spawnings might also decrease over time. Maybe starting at something like 5000 game ticks, counting down by 1 for each new house spawned.
Each house generates some number of pips per minute
The pip is placed on a destination of the same color that is a) not full and b) is the destination that has not received a pip in the longest duration
There are some variations that were too complicated to pick up on without detailed analysis, but I do know that trip generation rates will skyrocket if a new destination cannot be placed for a while
"Operator, yeah it's me again, they're blocking. --- I know, right, you'd think they'd learn, anyways triple the pip rate and take a lunch break"
@@Retroflux yaas!
if houses are the ones generating pips it could also be
"Jim? they haven't opened a new Redmall™ in three days!"
"What? No!"
"I'll call the rioters..."
I didn't know this going in but this is exactly the video I wanted. God I love nerds and charts and graphs. Thank you.
Very impressive, I wouldn't have the patience in the mind or the mathematical brain power to do that. Look forward to seeing more breakdowns 👌
Patience and stubbornness are two sides of the same coin, I'm very familiar with both 😂
Love this kind of analysis
Oh boi, scientific method applied properly. I love this! Thank you ❤️
if counting pips is something that interests people you can use cheatengine to slow the game down (speedhack, set it to 0.5 or 0.25 or whatever) - then when you play it back you'll have double or quadruple the frames for each pip event or whatever. Yes, it's a slog to play, but this is about math, dangit.
"I put in ten hours for a three-minute video"
Eight. An eight-minute video.
Yeah.... It kind of ballooned, like a lot
Great video, meep it up
Thanks, will definitely meep it up :^)
@@Retroflux 😂 meep meep
I loved that analysis and just subscribed. My husband, a stats nerd, was in heaven. 😄
glad someone did the math so we didn't have to ;P
Gift for the algorithm. Also, amazing analysis. Sorry you had to put in 10hrs counting pips to satisfy our curiosity!
your data of this game also canbe used to know how hard the devs do math to make this thing hapens and we can also know how all ai was coded in this game
Hopefully they followed Occam's Razor and just did the simplest solution, though there's lots of little things here and there... The more I dig, the less I think it's simple.
So does this mean that routes for the same coloured cars/buildings should be connected to share the load?
That’s an interesting question. In my experience with the game, I had much more success not sharing loads solely because it can cause too many intersections. I think the downsides of the intersection outweigh any benefits of sharing load, but I’m not sure on this.
You’ve got my sub!
Merci, Danke, Gracias, and thank you! 🙂
Actually seems like a nice series
Generally unnecessary, but intriguing none the less
Everything in this world is unnecessary if you really look into it too much, but that's where the joy of life is! Enjoy the unnecessary, it's the nectar of life :)
Awesome. Thank you!
Thanks for putting in the work but ah linked spreadsheets would be nice to get a fuller understanding of your findings. I think others may find other conclusions in the stats as well. From what I've gathered they are connected by colour but does colour have a specific amount? I doubt it although some colours are certainly more active than others.
Good video fam 🙇♂️
i NEEDED this video
"Tangent of the motorway arc" I thought tangent was only for right triangles. Is that the joke?
Technically it'd be the tangent of the slope of the arc, but yes I was definitely throwing very big math words all in a sentence for a joke as well
This is one of my favorite videos I've ever seen on TH-cam. Is there a way I can show my support with a donation? 😊
Some day in the future I may stream! But for now, I am humbled by the love of everyone so far, so I want to take no donations. Your subscription and gospel of my channel is more than enough for me, 100%. Money comes and goes, but what matters at the pearly gates is what you did with your life.
Don't particularly like the game, don't know the channel, but a big thumbs up for the work you put in! That's a feat of strength!
You don't know my channel? Well, now you do 🤔
@@Retroflux indeed, pretty good stuff!
Wow, i dont understand but great video! ;)
Me neither, hoping the comments will help me understand
damn that was actually really interesting
Ngl, I have that effect on most people when I talk
the devs seeing this video: "oh so that's how our game works. good to know"
If I can reverse engineer their whole game without looking at their code, I will be proud of myself