And what is everyone's end game behind bitcoin ?   As it is now, it is also to sell and hold everything back in fiat.  Although there IS for the moment a small amount of transactions (as compared to total volume) and HENCE a small part of the bitcoin market cap that is sustained from "buying goods and services", it is also quite minuscule, so this is NOT the finality of most bitcoin holders.  Most bitcoin holders are, probably like many monero holders, just speculating on "higher fiat value to cash out one day", NOT to "buy server time services with bitcoin".
As such, one can also say that "all of the bitcoin holders are really fiat holders, but they don't know it yet.  Why ?  Because there will be a time that they will convert their bitcoins back to fiat".

I'm not saying that "centralisation" is the solution.  That said, in a rigid set of rules, *whatever they are*, centralisation will happen, no matter what system of rules is used, because of economies of scale, and because of divergence of any formula which has something at stake.   

The thing is that calling PoW intrinsically centralized *because of storage and bandwidth* is, in my opinion, totally misguided, and proposing as a solution the hardcoded limit of functionality (in bitcoin, the number of transactions per second, as long as it is ridiculously small as compared with the potential market) is way, way worse than allowing the solution to scale according to technological limitations, which will remove themselves as technology advances.  No rigid set of rules should include hard limits on technology use, like Bill Gates did.

As such, of all reasons, PoW is NOT intrinsically centralized because of storage and bandwidth limits.  At the moment, bitcoin is centralizing, because of economies of scale which are SO HUGE with asics, that CPU/GPU mining is out of the question.

I agree with your statement of principle that any algorithm can be ASIC-ed, but the cost to do so, and the gains obtained can push this transition to the future.   The point is that this transition should be enough in the future for the coin to have a reasonable life time, and in the end, be obliged to fork and to become worthless.

My idea is that the only solution to avoiding centralisation, is modifying the rules regularly and having complex PoW mining, so that too big investments, for instance in ASICs, are too much at risk, so that the step to large economies of scale is always beyond the time line when the algorithms change.   There's a natural way of doing so, with continuously creating altcoins, that take over the "old" mature coins ; this could in a way be formalized by having finite block chains (that is, block chains that will only live a finite amount of time, say 10 years) and continuously forking.  The point being that a block chain that centralizes, is supposed to lose its value, because it loses its potential immutability.  The day that 60% of bitcoin mining is done by 1 entity, bitcoin will be worthless in principle.  So it is in a coin's long term viability that centralisation is to be avoided.
Another solution can be to have an inflationary pressure as a function of centralisation, in other words, a non-linear relationship between PoW and mining reward: this is a kind of transition from PoW to a diverging PoS.  The diverging PoS is necessary so that an inflationary bomb explodes when PoW gets too centralized ; in other words, the opposite of what happens in the ethereum time bomb.  As such, a centralizing coin destroys itself by hyper inflation, pushing people to a less centralized coin.  Instead of allowing wealth accumulation through centralisation, a coin should be designed so that it gets more and more worthless, the more it centralizes, instead of impossibly fighting centralisation.

The only solution to avoid centralisation, is regular "rule revolutions".  Like in real life, where war, revolutions, unforeseen events kill successful companies, and nations, so that centralisation has been avoided up to now.  We are leaving a dangerous period where centralisation in the USA was a danger, but as we see, the rules get overthrown, and the world is again diverging from centralisation, with enough wars, conflicts and uncertainty.

That's the fundamental of "greater fool theory".  Buy a collectible, with the idea that you'll find a greater fool that will buy it from you at a higher price.  There's still a lot of room on the up-side for greater fools in bitcoin, as most of the world population is still available to take on that role.

I don't call this "adoption".  It is "greater fool theory" at its best: people are buying it, to hope to sell it at a higher price.  There are still a whole world of "greater fools" to be captured, so there is still a lot of rising potential.  But it is not "adoption" in the sense of "I want to use it as a currency, to buy goods and services".  That's more or less dead now.

Nope, but I hate it.  This is increasing still more the speculative aspect of crypto, and killing off all together its usage part (as a currency, to trade goods and services).   I wouldn't even be surprised that big financial institutions are at it, to make it a purely speculative asset they don't mind (yet another derivative-like thing to have in portfolios, swaps, futures, options and other stuff), and get it out of the way of competing with fiat so that they stay master there.

Crypto booms in the speculative domain, and is hence dead as an alternative for fiat.  It is absolutely clear that this is NOT about adoption (as a currency).  It is NOT because suddenly, all people are rushing to Openbazaar and need bitcoins (and hence increase, through Fisher's formula, the amount of stuff bought with a currency, and hence its price).  It is because of more and more people are playing at "greater fool theory", rendering its usage as a currency totally dead, and hence, kill crypto in its infancy as a menace for fiat.    We thought that crypto was going to get killed by law, but it is killed by greater fool speculation.  Crypto will not succeed as a currency.  It will become a purely speculative asset, for which block chain tech isn't even needed any more.

I think this is a big mistake to think that option 2 is out of the question.  This sounds like Bill Gates thinking that "640 KB is more than enough memory for a PC", and hard-coded a limit in DOS, because when 4KB machines were selling, that sounded like a reasonable proposition.

At this moment, and with a projection of a huge adoption, it is true that "all financial traffic of the world on block chains" doesn't seem to be feasible due to network and storage limits.  But that financial traffic is a finite amount that doesn't grow exponentially like network and storage capacity is.   What seems to be "a big block chain" may very well fit on a cheap device a few decades from now, and what seems to be a serious network congestion may be like nothing special a few decades from now.   Bitcoin's block chain, of transmitting 1 MB per 10 minutes, was eating up a lot of capacity when people had 28kb modems.  Now, 1MB per 10 minutes is peanuts for most home amateurs and get this in a few seconds.  In the 1990ies I still had such a modem.  That's 20 years ago.  Scale up 20 years from now, and you'll find a few orders of magnitude more network and storage capacity that make these "huge block chain" discussions sound ridiculous.  However, good immutable crypto, especially if it has large world adoption, cannot change its protocol any more.  So we're doing a "Bill Gates 640K" if we hard-code limits in block chains because of technological capacity, which will cripple the system in a few decades at most.

monetary tokens can only be indistinguishable, if, well, we cannot distinguish one from another.  Now, monetary tokens that have a traceable history to them, will of course be distinguishable from tokens that have a different history to them.  Napoleon's handkerchief is different from another handkerchief, because once it belonged to Napoleon.  That sheer fact distinguishes this particular handkerchief from all other, similar, handkerchiefs.

Fungibility doesn't mean that *everyone* can see that coin 1 is different from coin 2.  Fungibility means that YOU cannot make the difference between someone offering you coin 1 or coin 2.

For instance, if ever some of Satoshi's bitcoins start to move, they may have "unique collector properties" which render them 20 times more expensive and valuable than your every-day Chinese mined coin.  The same way that a handkerchief used by Napoleon may be valued much, much more than your cousin's handkerchief.

He never said that, but then, Satoshi is not a god.  A "free money" is not compatible with an economically non-free society like ours, and as long as there are taxes in relationship to economic activity (production of value - income tax / consumption - VAT) economic activity is not free.  The concept itself of a freely usable means of economic exchange, which IS the goal of crypto, is orthogonal to our current tax-states.
Of course, we will not get rid of the tax-state institution that we're living in since about 5000 years overnight.  My hope is/was that crypto was going to be to the tax-state and the non-freedom of economic interaction, what the internet is to free speech: something that by its sheer adoption forces states to comply in the end.  That said, free speech is as of now also still a far-away goal, but less so than before the internet.  Mainly thanks to anonymization, decentralisation of communication.

Getting rid of inflation in itself is not really the important point (although I think that for Satoshi, it was).

That's the goal of crypto, isn't it ?  To forfeit "protection of the law" so that in the end, we can do away with law and state.
Anything else one does with crypto, begs the question.  Essentially, trading stuff and being able to live off it, without depending on a public extortion system called taxes, and being able to enjoy economic freedom.
You buy your protection then also with crypto (that is to say, with smart contracts tying armed groups to your bought rights).

That's because crypto is not made for trading.  Trading is just a parasitic phenomenon that overtook crypto, and has pushed true crypto adoption as a means of exchange for goods and services in the dustbin.

You don't need any block chain to trade on centralized exchanges, btw.  It just sounds good to pretend to have one attached to the web IOU that you're trading.

I think the debate is more difficult than that.

It is true that PoW, if specialized hardware is out, allows for such economies of scale, that centralization is unavoidable.  This is what happened to bitcoin, what happened to scrypt coins which lost their reason-d'être for it.   For the moment, there are still PoW schemes which allow spread-out mining, but the question is whether that's due to the fact that these coins have low enough stakes that it isn't worth *yet* to develop ASICs for it, or whether they found the magic solution.

In other words, PoW avoids too much centralisation, as long as general purpose computing can compete in the mining race.  From the moment that you cannot do PoW on your PC any more, it is essentially over.  The question is whether there are PoW algorithms that will remain complex enough, even in the face of enormous potential gains, to resist specific hardware development.  I might think that an algorithm that is *regularly forked* might be a solution.  You are not going to invest into billions of hardware if 6 months later, the algorithm is forked into something that renders your hardware useless.   Also, if general purpose computing is the best solution, that might at best stimulate the development of better general purpose computing hardware, instead of totally useless hardware EXCEPT for mining a specific coin.

So maybe a morphing, forking, complicated PoW algorithm is the best protection against PoW centralisation. 

But PoS has also its integrated centralisation, because the compound interest formula diverges, which means that in a toy world where N parties have different stakes, and you let evolve a PoS system (with "interest") in such a world, then the initial top stake holder will ultimately obtain all coins.  One can even estimate the time frame over which this divergence takes place.  Once the top stake holder has 50%, or 66%, or 95% of all stakes, he is total master of the network (which, at that point, becomes most probably worthless).

The PoS system has the inherent instability of any "capitalist" caricature, where the rich get richer, because your possessions give you a a proportional fraction of the gains, which increase your possessions, which entitle you to an even larger fraction of the gains, until you reach 100% of all gains.

There is essentially no solution to the centralisation problem, apart from 'changing the rules' regularly and in an unforeseeable way, which is exactly what "immutability" tries to avoid.  In the real world, that's also what happens, and avoids all richness to concentrate in the hands of an oligarchy: revolutions, political overhaul, unforeseen crises, war, ... 
Without those perturbations, there would just be a few rich families possessing the earth, and all the rest of us would be their slaves.

Kraken announced XMR trading.

Look at the answer in my previous post.   I fully agree that *if there is no real world knowledge* then it will not be propagated (obviously) on the block chain.  This is probably the kind of anonymity Satoshi thought he implemented.  But using money means of course giving out real world knowledge about a specific transaction.  Most of the time, the buyer and the seller know each other (it is even legally required, with KYC).  That a buyer and a seller know each-other, and know the specific transaction between them, is normal.    However, that this knowledge (to the buyer - emitter of coins - for future information propagation, to the seller - receiver of coins - for information propagation in the past) LINKS to other transactions is the privacy nightmare of a public ledger.

In my example of me, the plumber, Joe and Mary, Joe received future knowledge as a buyer about MY transactions ; Joe further received coins from Mary, which propagated information backward about HER transactions.  When he combined both, he found out something new, namely that I (he knew me) transacted to Mary (he knew her).

The fundamental error in thinking that bitcoin's pseudonymity implies transaction anonymity is by looking at addresses and transactions in an abstract way.  Of course, there's no way to find out, JUST from looking at the ledger, what real world identity is connected to what bitcoin addresses or transactions.  But that's not the point.  ONCE you know something, and with use of money, that's very normal, as money is about agreements and interactions between people, that knowledge
PROPAGATES.

"mixing bitcoin" is a somewhat clumsy way to obtain a kind of anonymity in bitcoin, but you need a third, trusted party for that, and (shady ?) partners that want to mix too.  With sufficient effort, such "mixing constructions" can be obtained with fiat too if you are allowed to trust third parties.  The bitcoin protocol IN ITSELF doesn't allow any form of automatic anonymity.   
The point is that mixing is clumsy in bitcoin, that most coins aren't mixed, that the mixing itself "colours" the coins, and that non-mixed coins, with the network of change addresses and combinations to arrive at the right amounts of payment, are propagating real-world knowledge to *anyone*, not just to banks and law enforcement if they obtain a warrant for it.

If I'm a plumber, and I get paid in bitcoin to repair the toilets in Joe's bar, and then I give some of those coins to Mary, who is my friend, and Mary goes and has a coffee at Joe's bar, then Joe can EASILY find out that the money he gave me to pay for his toilet repair, ended up in Mary's hands, so that there is a link between me and Mary.  If the same is done with fiat, Joe doesn't know such a thing.  Bitcoin's ledger is propagating real world knowledge like a bush fire.
Yes, I can mix my coins.  Yes, I can do 20 transactions between my own accounts and pay a lot of fees to "hide" this.  Joe will nevertheless see that Mary got the bitcoins he gave me at a certain point, even though now Mary COULD also have obtained them from a shady drugs dealer on the other side of the world I happened to mix with. But what should have remained (innocent) private information, is shouted out to the world with bitcoin's transparent ledger, and to even try to hide it takes me a lot of effort, and the necessity to trust a third (mixing) party ; while with fiat, that kind of innocent privacy comes automatically.  If Joe has paid me with a wire transfer, and I gave money to Mary with a wire transfer, and she paid with her VISA card at the bar, then Joe cannot know in the most remote way that the money he got from her was the money he paid me.  Actually, in fiat, that doesn't even make sense, because bank money is fungible while bitcoin isn't (by definition, every coin has a whole history to it which distinguishes it from another one - apart from a mixer operation).

That must also be the reason why no other car exists but Daimler-Benz.  After all, they invented the internal combustion engine with 4 strokes, and all other cars are just copycats.  So they never made it to the market, and Toyota never got beyond a measly market cap compared to Daimler-Benz.  Or did it ?

There's still another difficulty with bitcoin: "protocol-stiffness".  It is increasingly difficult to modify profoundly the technology of an existing crypto currency, because its value proposition resides exactly in the immutability of its basic principles.  The more mature a crypto currency gets, the harder it becomes to change anything fundamentally.  This is not true with other technologies, where the first mover often also has most experience in *improving* technology.  Bitcoin is stuck with old technology, and it is much harder to get new features into bitcoin, than it is to create a new coin, or implement new features in a young coin that is still malleable enough to accept modifications.

The most striking problems of bitcoin are of course the backed-in limited amount of transactions per second, which seem hugely difficult to overcome.  And the transparency of the block chain (in other words, the failure of pseudonymity, because of the networks of change payments, and combinations, which propagate unwillingly a lot of real world knowledge of identities), together with the too simple PoW scheme which has led to ASIC mining and hence, re-centralisation.

That said, because of most of crypto today is speculation on exchange IOU tokens, the actual block chain tech doesn't really matter for the moment.  The speculative game that supports most of the market cap of bitcoin and crypto in general doesn't even need any block chain, so the technology doesn't really matter (apart for boasting and hyping).  The little use that is really made of crypto can do with just any block chain tech. 

Indeed.  Economic freedom is something states are afraid of.  Then they cannot rip off their productive people any more.  Control their money, and you control them.  Bitcoin was a way out.  They've seen it.

Yes, you can home mine monero.  So yes, if you home-mine monero, you really contribute to the non-centralization of the network, something that is gone since long in bitcoin.  However, I don't know how long this will still be possible with the rising market cap of monero.

Well, storage capacity grows over time.  But frozen-in protocols don't.  The non-mining nodes in bitcoin are worthless what concerns the decentralization of the network.  They help somewhat in having a robust transmission network of transactions from users to miners, but any non-mining node that doesn't agree, can only do one thing: drop out of the system.  A miner that doesn't agree, can mine according to his preferences, and fork off.  But a non-mining node has nothing to say.  If he doesn't propagate a transaction he doesn't like, another non-mining node will propagate it, and even in the end, the user can transmit his transaction directly to a mining node which is the only communication that matters. 

So the number of non-mining nodes for bitcoin doesn't mean much: the centralized mining is much more of a concern.

It is nevertheless interesting for an individual to keep a non-mining node running, because it can help obfuscate his own transactions, and it can help him against fake information about the bitcoin block chain.  But that's about it.

As such, even if there were 10 times less full but non-mining nodes on the bitcoin network, that wouldn't matter much.

However, what matters is the fact that there's a hard limit on the growth in the number of transactions that the block chain can handle.  We're not even a factor 10 from that ultimate limit.  Bitcoin has no order of magnitude growth in transactions left.
As most transactions are, for the moment "greater fool speculation" transactions, and not transactions used for an actual currency usage, there's still a lot of potential currency usage adoption possible in bitcoin, if the trading would go down.  But there's a hard limit built in it.

As to the size of the block chain, even a chain that would grow with 1 TB a year only means a very modest investment for a node owner at this moment.  You get a 1TB storage for the order of $100,-.  So that's peanuts.  10 years ago, 1 TB of storage was much more expensive.  We can presume that 10 years from now, 1 TB of storage will cost much less.  So if 10 years from now, a block chain grows with a few 10 TB per year, wouldn't be a big issue for an individual wanting to have a full node either.
However, if the protocol has a frozen-in limit, then these technological improvements in storage don't mean anything.

A finite block size limit sounds like Bill Gates telling that 640 KB of RAM is more than enough for any personal computer (remember ?).

Unless one implements a form of anonymity in bitcoin, bitcoin is a privacy hell as compared to electronic fiat cash.  In fact, I guess it would be a dream for all agencies of whatever nature that want to audit people for their behaviour, that people use bitcoin.  Bitcoin is a privacy nightmare with its transparent ledger as it stands today.  Electronic cash (fiat) can only be audited with a special request, but the block chain is open to read and analyse for anyone.  It propagates real world knowledge of transactions like a bush fire.