I'm relatively conflicted by the changes, but overall am cautiously optimistic that this is a good beginning, and with further development might retroactively be seen as a great change.
Having read many of the comments so far, a common theme is that this completely removes any complications in fluids. Obviously we don't have the full details yet, let alone any demo to play with, but one obvious (but relatively minor) complication I haven't heard anyone mention: the new fluid system throttles throughput to each consumer according to the fill status. So for each consumer to work optimally, the pipe system needs to be full. But if the pipe system is full, producers are capped by what the consumers use each tick. Fine in the steady state, but presents issues in the 'consistent producers, spiking consumers' case.
Not a very hard problem to solve, but I imagine there might be a number of such minor complications that we are not aware of, that while different to the current problem set, nevertheless might make the new system interesting.
As an aside, I find it amusing that the new fluid system was heavily derided in a previous FFF as something Wube would never stoop to implementing.
~~~ Overly long rumination: ~~~
For the change itself, on the one hand, from an engineering/challenge for challenge sake perspective, its certainly disappointing. It feels like giving up. Instead of building something bigger, better, higher, the choice has been made to retreat to something simpler and apparently more primitive. And aesthetically, the knowledge that the game is actually simulating the fluids at some level adds a certain je ne sais quoi of immersion. Many other things in the game have what some may call unnecessary levels of simulation (inserter swinging, belts, trains), and this feels like a departure.
On the other hand, to be honest, this is already how I treat pipe networks. Each one gets rated for a certain flow rate, then as long as I make sure everything connected to it stays below that magic number, I treat it as if the fluid is just teleported around. Admittedly the highest flow I had to deal with was 8000/s for a nuclear power plant, so not very much compared to some megabases. However I didn't find the puzzle of squeezing that much flow down some pipes very challenging. Its just 4x2k, splitting into 8x1k. Honestly, pretty trivial. Train station were a bit more fun, but once you have a solution, there's not that much point in varying it, so the only a small amount of puzzle is lost with the new system.
On the other other hand, factorio seems to be all about giving the player a bunch of problems, and having them do the simplifying for themselves. Making ever more useful bluprints, increasing their own efficiency and effectiveness through clever design. This change feels like a case of 'the devs are playing for you', with something that used to be a problem now solved without any player engagement or thought.
On the other other other hand, the general vibe of problem solving in factorio is that the game is composed of relatively simple systems, the workings of which in isolation are easy to grasp. The complexity comes from the interactions of all the systems together. It is emergent. And fluids were an exception to that, with complexity build into the base layer of the system. The rules for the system are inscrutable. Experienced players have many rules of thumbs, and need to consult tables. Engineers have posted on this forum, showing many graphs and using python simulations (!!!) to explain how the system works. Very, very few people have a total understanding. It is quite easy to just completely break the system (try connect ~35 EDIT: heat exchangers in a row. The first 25 have water available, and the last 5. I'm the middle, 5 are just...skipped. This is understandable given the algorithm, but completely broken from a realism and gameplay point of view).
And I imagine its also not easy on the devs: if they want to make ANY changes to the fluid system, they have to consider a relatively complicated algorithm while doing so. Seems harder than it needs to be.
Whereas the new system is much more in keeping with the rest of the game. Easy to understand at its basic level and performant to boot. But it also makes it much easier to add complications on top, in a way that's easier for both players to understand and developers to implement. The question then becomes, what happens next? How will the new system be used by the developers to create interesting challenges for the players? I don't think we have enough information at the moment to judge.
Just to brainstorm some ideas of varying levels of awful:
- Total output from a segment is limited.
- The limit can correlate with the number of entities in the segment.
- Budgets can be separate for consumers and pumps.
- Segments are limited in size, either by number of connected consumers/producers, number of constituent entities, or by the geographical space which they occupy (e.g the area of the smallest square to fully contain the segment is limited).
- This might require direct segment to segment connections without a pump in between, but the algorithm would be dealing with entities with much larger total volumes and thus a lot of momentum, and would need to be invoked many less times, so could be allowed to do a better job.
- The amount of flow through any one pipe can be arbitrarily set
- This would probably require an extra algorithm on top: e.g drawing connections between a producer and all possible consumers, designating nodes in the segment where these connections converge, and ensuring that all nodes sit below the desired threshold. Such an algorithm may or may not be easy to deploy, but would be able to replicate most of the interesting behaviour of the current system
Of these, (1) seems most promising and trivially easy to implement. It can mimic most of the current behaviour: simply set maximum output as a function of the number of pipes/storage tanks. Ta daa!
(3) seems more and more like either nightmare fuel or a fascinating algorithm problem.
So basically, the new system leaves plenty of room for Wube/modders to add in plenty of complications to create fun and interesting puzzles to solve. The only question is to what extent will they do so.