It is not true that there is no such thing as ASIC-resistance.

Algorithms can be formulated that are not simple to manufacture ASICs for, like the Cuckoo Cycle. The reasoning goes that the more difficult it is to make an ASIC, the less of an advantage it will have against general consumer-grade computers. On the flipside, the complication raises the barrier to entry of competing ASIC manufacturers, thereby potentially centralizing the ASIC suppliers.

More importantly, it misses the forest for the trees. ASICs are not the core reason for centralization of mining, it doesn't even appear to be a significant factor. The most significant cause is comparative advantage in energy costs, which is just as advantageous even if you had an ASIC-proof algorithm.

While it's strictly true that running a node is not rewarded, rewarding them would not aid in decentralization. After all, it's trivial to run more than one node in a centralized manner, and a reward would only encourage that behavior.

Centralization is evidence that operating a node is non-trivial and that centralization offers a significant advantage and reward.

Ideally, we would like to see the cost of mining exceed the block reward in order to drive out centralized high-overhead competitors, while the charitable miners remain, while simultaneously hoping that the Work is high enough to secure the blockchain. When the problem is phrased like that, it's apparent that ASIC-resistance and node reward won't get you there.

In order for this to happen, casual users must be willing and able to operate a mining node while investing trivial time and resources to do so. In this regard, ASIC-resistance could be useful because it could negate the need to purchase specialized hardware. On the other hand consumer-grade ASICs could potentially make POW easier to offload from their devices. Regardless, just like they aren't the most significant factor leading to mining centralization, ASICs are not the most significant reason nodes are not trivial to operate.

The elephant in the room is the size of the blockchain. For a casual user, it's slow and tedious to download and verify, and it takes up enough space to be annoying. There is also the problem that the software is still not settled and prone to potentially consensus breaking behavior, meaning a node cannot be left to run without any user intervention.

1.) Running a node and mining should be unprofitable.
    a.) Mining should be done charitably.
        a.) Providing work should be trivial and easy to get started.
            a.) Bitcoin's consensus code needs to be settled indefinitely.
            b.) The blockchain needs to become easier and cheaper to download.
        b.) Some sort of religion. (Lottery?)

ASIC-resistance, higher inflation, POW substitutes, and larger block sizes have all been proposed as ways to achieve a higher quantity of decentralized nodes. ASIC-resistance is at best an incomplete solution. The rest make the problem worse.

That's my 2 cents, anyway.