ASIC Resistance : wtf is that?

Hello Everyone!

Maybe some of you have heard the claims of "ASIC resistance" amongst some proponents of this or that new proof-of-work algorithm, or even some non-work centralized algorithm.  "ASIC Proof" even.

Lets talk about what this might mean.

First off, it's worth pointing out that doing the appropriate hash function faster will make you more profit as a miner.  It's a computational problem, how fast can we compute these functions.

Consider some somewhat similar computational problem, that of factoring an integer or determining the discrete logarithm of some large number.  Here we have a nice history to look at, and we see that two areas have sped up this computation:  1)  Hardware    and     2)  Algorithm development (software).

There is no reason to suspect that hashing for proof-of-work public networks will be any different.  The algorithms and software WILL improve, and the hardware WILL improve.  This viewpoint might come from somebody who was alive during the peak stages of Moore's law, but still - that's really the way it is.  Software will improve, hardware will improve.

Provided of course that the network in question provides enough rewards to incentivize us to improve!

If the network hash remains worthless, then sure - we might not bother to get specially improved software or hardware.

So does "ASIC-Restistant" then mean that the coin will be worthless?  That would do the trick, but then again "resistance" implies that somebody was trying to break into the house, not that the materials visible through the windows were not worth bothering with.

That's probably not the type of ASIC resistance those who throw term about are going for.

But what else could it mean?  Several coins have now had proponents claim "ASIC Resistance" and have gone on to have promising futures :  Litecoin (and other Scrypt coins), Bytecoin (with Monero et al.),  Ethereum (Equihash et al.), and others.  However a quick search will show that dedicated mining hardware is available for all of these coins.

Is there any room for a valuable computable which cannot have a specific integrated circuit for its application?  I don't see any.  What am I missing?