He neglects that there is no reason to pay fees, if there is no limit on supply.


Since there is no marginal cost in including a transaction to the current block, a rational miner will always include a transaction with a non zero fee,
before it is included by any of its competitors.

Therefore a lower bound on fee will not work without a cartel or without a competition for space.

I prefer algorithms over cartels.

let me be more precise:
There is a marginal cost implied by block propagation speed being proportional to size and propagation being proportional to orphan rate. There is also a computation cost of updating the merkle tree and updating miner with it. These marginal costs are today however magnitudes below the lowest non-zero fees paid.

Because majority of miner nowdays use the bitoin core as published by the devs and that compiles blocks following a mix of commercial and alturistic rules.

Majority of miner do not care about this since inflation provides three-four magnitudes higher income than fees at the moment. They will. And this discussion is about the future and the limits.

Since I so often see the argument about, what is good for the miners:
Is there any statement from the large mining pools about this?

You know much less about Bitcoin that you think. I hope you have better things to do.

Miner do not make the rules. They have to obey the rules if in minority or can narrow existing rules if organizing a majority cartel.
The later would be needed to impose a non-zero fee in absence of a size limit in the protocol.

Your second and third paragraph contradict each other.

Transaction fees don't ONLY exist because there is a competition for block space. They ALSO exist to pay the miners to secure the network, as you clearly understood before you implicitly denied it. Fees are not an either/or thing. It absolutely ISN'T a case of lifting block limit->eliminates fees->eliminates mining.

We have instead a *feedback* process. LOWER fees (not ZERO fees) means LESS mining (not NO mining) which in turn means LONGER confirmation times (not COMPLETE COLLAPSE) which leads to MORE FEES which leads to mining power switching back on. It's what engineers call a negative feedback loop, designed to keep the hashing rate broadly stable, or at least oscillating within a fairly narrow range.

People really must stop thinking of all the causes and effects in the world as being ON/OFF switches. They aren't. They're analogue dials.

[If you're on board with the idea of bitcoin, you've probably had to deal with people saying a deflationary money supply can't work because NOBODY would ever spend ANY MONEY AT ALL. Same problem. "Less" is not the same as "none". Especially when "Less X" induces "Less Y" which induces "More X".]
 

you really dont get it do you?

even if we ignore the legion of premined scamcoins, we still have the following issue:

Suppose we use 4 different altcoins

Acoin has a 10 GB blockchain in this example
Bcoin has a 5 GB blockchain
Ccoin has a 20 GB blockchain
Dcoin has a 15 GB blockchain

now how exactly is that better than having 1 blockchain of 10+5+20+15=50 GB?

If you are going to store them all anyway, it doesnt matter if you have 1 blockchain of 50 GB or many smaller ones adding up to 50.

In fact, having multiple blockchains means you have a lot of extra redundant information and applications running, it's less convenient, less secure, etc. It only brings more problems and it doesnt solve anything.

Why am i even explaining this....

Oh another thing: re the concern about a huge blockchain size...

Increasing the block limit isn't the only thing Gavin and co have been giving a lot of thought to. They've also understood the (complementary) problem of blockchain bloat. I'm given to understand that several blockchain pruning schemes are out there. Note these are NOT the same idea as 'lightweight' or SPV clients. They're genuine info compression schemes. I haven't followed for a while, but one scheme I heard basically suggested that *on disk* the blockchain information up to the 'current block #X' should essentially be split into three parts:

1) A chain of block headers only up to block #X-N, where N might be a couple of thousand.
2) A database of UTXOs for all the blocks up to #X-N
3) A complete sequence of the last N blocks

The header-only chain provides the proof of the next block's valid membership of the chain. It grows linearly for ever more, at a rate of ~(640*6*24*365)/(1024^2) = 32 MB a year. Which is peanuts.
The UTXO database grows more unpredictably, but I guess broadly proportionally to global bitcoin usage. It's there so the node knows who's got what and can verify new transactions are valid.
The uncompressed sequence of the last N blocks, where N might vary by individual preference, remains fixed length for a fixed blocksize. The node keeps the last N rolling blocks uncompressed in case there's a fork.
Even if the blocksize grows, so that this tail of blocks grows, it's unlikely to gum up disk space. If it does, the node could choose to reduce N (I pick N=2000 as rather overly cautious).

Anyway TLDR: disk space is NOT a problem. Can't comment much on network propagation issues, except I'm led to believe these can be addressed by an approach such as 'send the header first, then the block info separately'.

Not the coin that's the biggest bully and has the most vocal support wins but the coin that's most efficient. (not even network effect and merchant adoption are as important as efficency in the long run)

So fork it all up, fools. Do it right now!

That's probably the best argument I've seen.  I think of how today I have many paper wallets, mobile wallets, and hosted (shared) E-Wallet services (i.e., custodial services) where I had a balance of a few millibits or less in them and won't (or can't) withdraw because the Bitcoin network transaction fee is prohibitive for that small amount of bitcoin.    Now as fees rise due to scarcity of empty space in blocks then I'll essentially have permanently abandoned those wallets with the small values (i.e., those funds become economically unspendable, ... worthless!).

I suppose for the paper wallets and mobile wallets where I have the private key I could combine them into a single transaction without incurring a higher fee, but that's possible only today because there's still some low-cost space left and the fee required doesn't rise linearly when the size of the transaction rises.   Without a fork then eventually we are essentially paying a fee for each additionally byte and the resul will likely be that millions of UTXOs with small amounts (e.g., sub-millibit -- less than 0.001 bitcoins) will become worthless.   Hey now,  that's some real money we're talking about discarding!  [Anyone care to crunch the numbers -- what is the sum total of bitcoins that exists in UTXOs under one millibit?)  

It makes me think back ... First they came for the dust, and I didn't speak out because I didn't play SatoshiDICE.    When this day comes (when the 1MB cap is reached, and no-hard fork), maybe a millibit becomes the new dust!   And then let's say the fee required rises even more, maybe UTXOs under 0.01 even become economically unspendable.    That becomes a real problem for most of us!

So what to do?
I'ld like a Bitcoin that can scale and grow with transaction demand.  I'ld also like my car to get 150 MPG.  The latter can't happen simply because a gallon of gasoline doesn't have the energy needed for an internal combustion engine to propel today's car for 150 miles.  Now I'm not entirely convinced the former can happen either.  That's because a hard fork that doesn't have the consent of the Economic Majority ( http://en.bitcoin.it/wiki/Economic_majority ) will fail.

Let's go through the first hours of the hard-fork.  Let's say it happens at block 400,000 (a lttle over a year from now).  Everything was in place -- miners with the right nVersion indicating consent was well above the threshold (e.g., 80% of last 1,000 blocks had nVersion=4).     But ..., I'm not willing to believe that exchanges, merchants, merchant processors, etc. are going to themselves take on the full risk of double spending that would occur if the hard fork eventually fails (maybe a day or three later even).   The only way to prevent the double spending that would result would be to require that a transaction confirms on the block chains on both sides of the fork.  So these entities are going to watch both sides. Well, a transaction that has any taint from a coin generated on the side with the larger block size rule change (I hate calling them "gavincoins", but for the purpose of this argument that name is short and everyone here knows what it means) will not confirm on the other side of the fork where the 1MB limit is still followed.

So the market instantly realizes this difference between a Bitcoin and a GavinCoin.  So the value of a GavinCoin will drop relative to a Bitcoin.   Miners can't convert or spend these newly mined coins nearly anywhere, so all you have is buying from speculators.   Now those mining on the side which still recognizes the 1MB max limit are still mining blocks (albeit at a much slower rate because of the dramatic loss of hashing capacity) and some market for those newly mined coins exists -- again, thanks to speculators.    Things can flip quick.  Maybe a day goes by and all of a sudden a large amount of hashing capacity switches back to the 1MB max side due to the dropping exchange rate of GavinCoin, and it becomes quite possible (if not probable) that the 1MB limit will be with us for some time longer.

The exchanges and merchants that played both sides (i.e., required confirmations on both sides of the fork) lost nothing as either way they have confirmed transactions on what is eventually the sole winner.   If the hard fork fails then the losers are those who had E-Wallets (custodial accounts) and found their pre-fork bitcoins were spent and they only end up with tainted GavinCoins that can now never be spent (at least not anywhere near parity with a bitcoin).   Likely most every custodial service (e.g., exchanges, hosted/shared eWallets, etc.) that didn't require confirmations on both sides of the fork ends up bankrupt as a result of getting dumped on with GavinCoins while allowing withdrawals of untainted bitcoins.

I just don't see how a hard-fork succeeds.   There is risk of accepting GavinCoins.  There is no risk (excluding exchange rate risk) of putting your own pre-fork coins into storage for a (long) while and not letting them become tainted GavinCoins (as they can still be spent a year, two or ten later).

[Edited: A couple small readability changes.]