Theoretically speaking all algorithms are able to be optimized by ASICs. ASIC resistance is not about making ASICs impossible, it's about making them expensive. But any secure global PoW currency is going to have enough mining reward to drive ASIC development.
The really ASIC resistant algorithms will have an R&D overhead of hundreds of millions. Which means if one player can get in really and keep their R&D secret or patented, they can amortize costs and more easily block newcomers. Overall, even without intentional barriers, you will have a much more centralized coin in the long run.
For any coin intending to become worth billions or trillions, you really want to embrace ASICs and design a PoW that makes it as easy as possible for newcomers to profitably develop and manufacture ASICs.
It's a weird concept to accept but actually the really interesting and useful property of PoW is that you can provide a proof that you destroyed physical resources (provably wasting electricity) which are committed to one and exactly one interpretation of history.
Stake based systems lock or burn or manage coins, but there's no provable tether from those coins (defined only within the system) to the real world.
PoW has an undeniable tether to the real world in the form of waste.
Thought experiment:
Say there's a PoW currency A, and PoS currency B.
Alice mines A.
Now Bob wants currency A, but he can't afford to mine currency A, because he needs special hardware.
So Bob exchanges currency B for currency A with Alice. Alice now holds some currency B, and Bob is happy with A. Now the question is, would Alice see that she "destroyed physical resources" for currency B? (Yes, to Alice she had to destroy physical resources)
Later, Alice wants to transact with Charlie, who also holds currency B. She transfers her B to Charlie, that was bought using A. Would Charlie care that Alice had to "destroy physical resources" to get B? (No, that information is useless in a PoS system)
Some time in the future, currency B becomes very popular (as it's just as secure and not wasting resources like currency A). Alice wants more B too, so she exchanges her B with the masses. Eventually, currency B takes over A in popularity, and by that time, Alice has exchanged all her A for B. The question is, would Alice now view that to get B, she had to "destroy physical resources"? Would that fact matter if she continues to transact with Charlie?
There is no reason for Alice to want B, a less secure money than A which they already hold. If Alice were perfect mathematician, this exchange would never occur.
Oh, and then there were cryptocurrencies C, D and F too, but that's a different thought experiment.
If a transaction does not do this, it can be guaranteed to be insecure.
>Faster block times means she'll get her coffee sooner
Maybe. Or maybe the merchant doesn't get paid for coffee at all (because faster blocktime means more orphans)!
Sorry, didn't mean to bring out the ASIC debate so strongly - my point was mainly that the research is not exclusive, and can be shared between projects, as in the litecoin using Segwit example.
It's not like the research is exclusive. Note that any research done on bitcoin can also be shared and used on other projects. Litecoin for example, which is a fork of bitcoin, is now benefiting from bitcoin's R&D by using Segwit first.
Likewise, other blockchains can teach a lesson for bitcoin. For example sha256 proved not to be a good hashing algo because the reliance on ASICs, which centralizes mining. Another important area in research is the development of other greener forms of blockchain security such as Proof of Stake, which will most likely replace Proof Of Work once perfected.
In a way, when I think of bitcoin, I think of it as this whole massive umbrella project that includes all the alternative blockchain projects too, that can all share their research / experiences / ideas with each-other. This is the beauty of open source software.