How a floating blocksize limit inevitably leads towards centralization

[markm]

So leaving the limit in place and forcing most peoples transactions to go through clearinghouses is going to suddenly make miners rich or give them an incentive to mine? Do people here really want to see transaction fees having to be 20 or more dollars each on the bitcoin network? And sure the fees might be enormous. The users will take the brunt of that blow. But will miners actually gain from such large fees? Isn't the amount miners are "incentivized" to mine from profit come not only from the size of the transaction fees but the quanity and wouldn't the quantity be reduced if you forced the majority of real world usage transactions to take place totally off the bitcoin blockchain?

The miners can mine a secondary chain pretty much as easily as they could mine double-sized blocks on the primary chain.

The primary good provided by the primary chain is the most difficult proof of work on the planet. Providing that with as few overhead costs such as bandwidth and storage as possible seems like a very good idea, the old do one thing and do it well approach. Accommodating day to day expenses, maybe even accommodating monthly salary paycheques, can easily be delegated to a secondary chain, and even smaller amounts to more secondary chains, allowing the primary chain to as efficiently as possible enable as many hashers as possible to contribute to the difficult work, fully verifying not just blindly rubber-stamping, while accommodating no more transactions, of no less value per transaction, than is absolutely necessary, taking into account the ability to apply that work to as many secondary chains as might turn out to be needed to accomodate any amount of increase in transaction volumes, even all the way down to children buying penny-candy on the latest and leastest secondary chain with its brownie-point coins that kindergartens and homeschoolers worldwide use to incentivise children to learn and teach them how the vast and variegated world of cryptocurrency works and thus to reap transaction fees on as many chains as they choose to scale themselves up to be able to merge all at once.

-MarkM-

[1715364033]

1715364033

[1715364033]

1715364033

[1715364033]

1715364033

[caveden]

Gavin and co, drop your silly suggestions of using an idea that can't be globally synchronized without a central authority. This suggestion does exactly what you are looking for in a decentralized manner and it's globally synchronized:
[snip]

An automatic adjustment is definitely better than keeping the limit as is, for sure. I used to support it.

Until I realized there's actually no better "automatic adjustment" than spontaneous order. Allow miners to set their own limits, together with "tolerance levels" (a way to say that if an "offending chain" is 2 or 3 or N block longer than mine already, I end up accepting it). Also allow them to define how large the block they build will be.
That's what you need.

I don't believe that pool operators would voluntarily decrease their revenues (by increasing the chance of losing their blocks) and increase their expanses (by paying for wider than necessary bandwidth) in the hope of kicking out lower-bandwidth pools. That's pretty much like price dumping, and price dumping doesn't work. (also, botnet operators have already kicked low-bandwidth pools out anyway... what's the percentage of bandwidth-expenses a pool must have just to be resilient against DDoS? That's likely to be more than what it takes to upload their blocks ).

* By the way, for totally different reasons, it's already not really possible to be a low-badwidth pool. The reason for this is not the Bitcoin protocol per se, but DDoS attacks from bot operators mining on different pools. Unfortunately most major pools were already DDoSed. Ultimately the only defense against such kind of attach is large bandwidth.

Or p2pool?

Which doesn't scale AFAICT.

[markm]

* By the way, for totally different reasons, it's already not really possible to be a low-badwidth pool. The reason for this is not the Bitcoin protocol per se, but DDoS attacks from bot operators mining on different pools. Unfortunately most major pools were already DDoSed. Ultimately the only defense against such kind of attach is large bandwidth.

Or p2pool?

Which doesn't scale AFAICT.

Huh? That is news to me, where did you discover that very important tidbit?

-MarkM-

[notig]

So leaving the limit in place and forcing most peoples transactions to go through clearinghouses is going to suddenly make miners rich or give them an incentive to mine? Do people here really want to see transaction fees having to be 20 or more dollars each on the bitcoin network? And sure the fees might be enormous. The users will take the brunt of that blow. But will miners actually gain from such large fees? Isn't the amount miners are "incentivized" to mine from profit come not only from the size of the transaction fees but the quanity and wouldn't the quantity be reduced if you forced the majority of real world usage transactions to take place totally off the bitcoin blockchain?

The miners can mine a secondary chain pretty much as easily as they could mine double-sized blocks on the primary chain.

The primary good provided by the primary chain is the most difficult proof of work on the planet. Providing that with as few overhead costs such as bandwidth and storage as possible seems like a very good idea, the old do one thing and do it well approach. Accommodating day to day expenses, maybe even accommodating monthly salary paycheques, can easily be delegated to a secondary chain, and even smaller amounts to more secondary chains, allowing the primary chain to as efficiently as possible enable as many hashers as possible to contribute to the difficult work, fully verifying not just blindly rubber-stamping, while accommodating no more transactions, of no less value per transaction, than is absolutely necessary, taking into account the ability to apply that work to as many secondary chains as might turn out to be needed to accomodate any amount of increase in transaction volumes, even all the way down to children buying penny-candy on the latest and leastest secondary chain with its brownie-point coins that kindergartens and homeschoolers worldwide use to incentivise children to learn and teach them how the vast and variegated world of cryptocurrency works and thus to reap transaction fees on as many chains as they choose to scale themselves up to be able to merge all at once.

-MarkM-

I think we might be talking about different things... it sounds like you are talking about something like litecoin. Another chain. A currency for maybe smaller transactions. What I was mentioning was what someone mentioned earlier "bitcoin clearing houses" . Essentially mtgox is already sort of a clearing house from what I gather because when bitcoins exchange hands on the site they just do it all internally off of the blockchain. There is no "secondary chain" to mine.

[TierNolan]

One reasonable concern is that if there is no "block size pressure" transaction fees will not be high enough to pay for sufficient mining.

Here's an idea: Reject blocks larger than 1 megabyte that do not include a total reward (subsidy+fees) of at least 50 BTC per megabyte.

"But miners can just include a never broadcast, fee-only transactions to jack up the fees in the block!"

Yes... but if their block gets orphaned then they'll lose those "fake fees" to another miner. I would guess that the incentive to try to push low-bandwidth/CPU miners out of the network would be overwhelmed by the disincentive of losing lots of BTC if you got orphaned.

Since this locks in a minimum fee per transaction MB, what about scaling it with the square of the fees.

For example, every 2016 blocks, it is updated to

sqrt(MAX_BLOCK_SIZE in MB) = median(fees + minting) / 50

or 1MB if lower.

Also, a maximum change from the last value of +/- 25%.

So, MAX_BLOCK_SIZE would be

<=50 median gives 1MB (cost per transaction 1.0X)
100 median gives 4MB (cost per transaction 0.5X)
200 median gives 16MB (cost per transaction 0.25X)

This allows the fee size per transaction to decrease, but only as the total fees increase.

I also agree with the "public good" issue with a miner cartel.   Trying to push up the MAX_BLOCK_SIZE might benefit the cartel, but each individual member has an incentive to defect.

[solex]

*To increase max block size by n%, more than 50% of fee paying transactions(must meet a minimum fee threshold to be counted) during the last difficulty window were not included in the next X blocks. Likewise we reduce the max block size by n% whenever 100% of all paying transactions are included in the next X blocks.

This is a smart idea as it keeps the max block size to a minimum which still allows bitcoin to grow, so is TierNolan's solution. SimonL's solution seems a variation on this theme too.  Even a pathetic solution: the 1000000 constant is changed to a 1000000 config variable (which can then be changed quickly) is an improvement over what exists today.

All I care about here is helping to prevent bitcoin hitting a wall.

Thus far, all the scams, hacking of sites, lost/stolen coins etc are "user error". Despite all this bitcoin itself has performed perfectly, and this is regularly pointed out to critics. However, if transaction throughput chokes up due to the max block size limit then the "Great Bitcoin Failure of Late 2013" will be written in the history books. Critics will always be able to point out this event whenever bitcoin is promoted to new users. It will not be as good as "virtual gold" anymore, and that would be terribly sad.

[caveden]

Or p2pool?

Which doesn't scale AFAICT.

Huh? That is news to me, where did you discover that very important tidbit?

AFAICT P2Pool requires each participant to be a full validating node. Maybe that changed and I'm not aware of it. But if such requirement still exists, then it cannot scale as is.
Of course, P2Pool protocol per se could remain. But its participants would eventually be pool operators, and the actual miners (owners of processing power) would connect to these pool operators participating in P2Pool. It would fallback to the model of centralized pools, with the difference that they would be constrained by P2Pool protocol, what might convince some miners to support them. Or not. Can't tell that. What I can tell is that all individual miners doing full validation is not something that scales.
For instance, if BFL really ships their products, I might consider buying the USB-miner thingy. But that would have to run without a full node.

[Zangelbert Bingledack]

Gavin and co, drop your silly suggestions of using an idea that can't be globally synchronized without a central authority. This suggestion does exactly what you are looking for in a decentralized manner and it's globally synchronized:
[snip]

An automatic adjustment is definitely better than keeping the limit as is, for sure. I used to support it.

Until I realized there's actually no better "automatic adjustment" than spontaneous order. Allow miners to set their own limits, together with "tolerance levels" (a way to say that if an "offending chain" is 2 or 3 or N block longer than mine already, I end up accepting it). Also allow them to define how large the block they build will be.
That's what you need.

I don't believe that pool operators would voluntarily decrease their revenues (by increasing the chance of losing their blocks) and increase their expanses (by paying for wider than necessary bandwidth) in the hope of kicking out lower-bandwidth pools. That's pretty much like price dumping, and price dumping doesn't work. (also, botnet operators have already kicked low-bandwidth pools out anyway... what's the percentage of bandwidth-expenses a pool must have just to be resilient against DDoS? That's likely to be more than what it takes to upload their blocks ).

Yes, why not let the miners deal with it? They'd develop best practices. I'm automatically suspicious of any argument that sounds like "laissez faire = chaos." Many speculations about miner incentives are likely to fall prey to central planning fallacies, ignoring the unseen and long-term adaptive measures miners would take as the market situation unfolds dynamically. Rogue miners who tried to pad or whatever would likely be dealt with by the response from the rest of the miners. Solutions evolve spontaneously.

[prezbo]

Wouldn't something like just distributing block headers solve the "some miners have incentive to make big blocks"?

Obviously, with unlimited block size blockchain would become too large for the average user to store, but we don't need that many complete blockchains. If I'm correct a fully pruned blockchain would contain all the necessary information to keep the network safe and would still be small enough for an average user to store?

[Zangelbert Bingledack]

Also, I can't help but think that there is some fallacy in assuming that Bitcoin is reliable because it's "legislated." Even the 21-million BTC limit, which is the very backbone of its value as a currency, can be seen as not being maintained due to legislation, but due to consent. What really is the difference between tens of thousands of nodes consenting to a legislated protocol, and those nodes simply consenting to "a" protocol? The 21M BTC limit being set in stone isn't what makes Bitcoin a reliable store of value; it's simply that people choose to support that hard limit.

What users are relying on is not any kind of contractual obligation or unbreakable rules, but rather just the will of the preponderance of nodes to stick to those rules. It is that social phenomenon, that will, that is the source of Bitcoin's ironclad value guarantee.

When we tell people, "Only 21 million bitcoins will ever by created in the Bitcoin system," we are speaking not of a set of rules per se, but of something more like a social movement that is characterized by its dedication to those rules. That may not sound very reliable, but it has proven to be.

[Jouke]

I like what Maged is saying. It makes the block size future proof, while insuring that the network is less vulnerable to a single malicious miner.

[Akka]

I still fail to see how this max Block chain limit could work longterm. The 1MB Limit allows for ~500K transaktions a day.

Now Imagine 180 Mil. People using Bitcoin. That would mean on average every User could make only one Blockchain transaction a Year.

If Bitcoin could still work that way thanks to off Blockchain services. It could possibly even work without the Blockchain at all.
If there are enough off Blockchain services that the average User never needs to make a Blockchain transaction, thus has the possibility to verify his BTC actually exits, where is the difference to the system we already use now?

And if maybe a few Transactions on the Blockchain per month for our off Blockchain BTC services are enough, why should this create high fees?

[markm]

Yes, but, if we the people cannot ourselves examine the gold-plated bars of tungsten n Fort Knox are we fools to think it is actually gold, as some traditions or propaganda claim but none of our friends or family, or in general no-one we actually trust, is able to assay the stuff?

If we cannot validate blocks aren't we in the position of risking every ten minutes that our "gold" has been replaced, behind our backs and by means of deliberately bloating the network instead of using scalable solutions such as a plethora of merged-mined chains, any one of which we could be careful to ensure is within our means to verify, and thus can monitor it 24/7, checking all of its blocks?

Why this determination to prevent we the people from assaying those bars of unknown but possibly somewhat gold-like in appearance material?

Multiple chains is very scalable. Trying to create one monolithic chain too vast for an ever growing fraction of the people to verify is, it seems, almost deliberately trying to keep the people in the dark, maybe motivated by the lure of one massive ("too big to fail"?) chain.

What is next? Maybe "sorry, you cannot examine the blockchain because that would violate the privacy of our users, only we the lords and masters are permitted to do that" ?

-MarkM-

[Nagato]

I still fail to see how this max Block chain limit could work longterm. The 1MB Limit allows for ~500K transaktions a day.

Now Imagine 180 Mil. People using Bitcoin. That would mean on average every User could make only one Blockchain transaction a Year.

If Bitcoin could still work that way thanks to off Blockchain services. It could possibly even work without the Blockchain at all.
If there are enough off Blockchain services that the average User never needs to make a Blockchain transaction, thus has the possibility to verify his BTC actually exits, where is the difference to the system we already use now?

And if maybe a few Transactions on the Blockchain per month for our off Blockchain BTC services are enough, why should this create high fees?

It wont be like our current system, but more like the classical gold standard where your gold(Bitcoins) are redeemable on demand. But since Bitcoins are slower/more expensive to transact with, people would over time stop using it at all and all future services may require interfacing with the service layer built atop of Bitcoin eventually leading to the demonitisation of Bitcoin just as it happened with Gold.

Although there are a few key differences between Gold and Bitcoin which may not lead to such a scenario
-Bitcoins are instantly assayable by anyone with an internet connection or an upto date blockchain.
-With declining usage, cost of Bitcoins TX will fall to compensate, possibly resulting a balanced equilibrium where only low cost TX are sent via the service layer.

But i think you do have a valid concern.

Ultimately it comes down to 2 forms of centralisation,
1) Centralised mining pools which have the hardware and infrastructure to run a full node process all payments at a low fee  (Large Block Size)
2) Centralised Payment processors(or possibly alt chains) built atop of Bitcoin to offer cheap/free transactions for micro-payment(as a complimentary service to their existing e-wallet/exchange service) in order to keep high levels of decentralisation (Small Block Size)

[markm]

Can we go with specialisation rather than centralisation?

By keeping the primary chain as small and efficient as possible, focussing on its being 24/7 live-verifiable by as many people as care to verify it as it happens, maybe we can keep mining accessible.

Meanwhile by creating many many many secondary chains, maybe we can also keep the verifying of any one secondary chain, maybe even the primary chain plus one secondary chain, accessible?

Large mining outfits could merge absolutely every regional, niche, and special-purpose chain, smaller outfits could pick and choose which chains to merge, depending on what economic sectors, industries, scales, neighbourhoods or whatever they happen to be interested in verifying at any particular moment.

But hey, we already have lots of blockchains, so if one of them wants to go ahead and use terabyte blocks why not? We all should have plenty of advance notice and plenty of time to trade our holdings on that chain for holdings on other chains, right?

We already see though that folk who favour the "primary" chain which all the merged chains are designed to depend on to provide the world's most difficult proof of work, have quite some tendency to refuse to even consider merged-mining.

Thus maybe we need to abandon merged mining as too impractical in the face of the prejudice of the entrenched / vested-interest "establishment" and urge them to go ahead and make blocks large enough that we can migrate all our altcoins to coloured-coin format, putting them directly on the world's largest blocks blockchain due to the "establishment" wanting to monopolise not only mining difficulty but also blockchain-size?

Or maybe terracoin and litecoin and bbqcoin have it right, that the "establishment" is too prejudiced to merge with; maybe terracoin and litecoin can be the new "primary" chains, one for SHA256 the other for Scrypt, and grassroots power to the people mining afficonados should eschew the "establishment coin" and seek to build merged mining setups that can beat its hashing power? Afterall, ensuring "the people" have more hashing power than "big business", "the military-industrial complex", "the world banking cartels" - in short, "the establishment" is a well established tradition in the cryptocoin world already...

Maybe we need to know what exactly all the various chains have in store in the way of immutable or arbitrarily vast / uncapped block sizes so we can start planning how much of our portfolios to place with how "eatablishment" and how "grassroots" a collection of blackchain projects?

-MarkM-

[johnyj]

Another possible scenario:

After block size limit reached, transaction slowed and there is a stronger motivation for a hard fork, so it happened. After the hard fork, the hashing power will divide, some miner will stay on the original chain, some others will hash on the new chain

And then transaction volume is divided, the original chain will suddenly get much less transaction, so it becomes good again, then people will slowly move back to the original chain, since it has much longer history and credibility (And most important - higher exchange rate), and this will slowly bring up the problem on that chain again, then some people could not tolerate the slow transaction will move to new chain

So it basically will create a self-balanced original chain, there is just enough transactions happened on the original chain to keep it working well and all the long term investors will stay there. All the people want fast transaction will move to new chain and tolerate the lower price per coin

Unless the block generation on the original chain can be stopped totally, the inertia of the original chain is so big that no matter what benefit new fork bring, it won't stop it from becoming more popular. Even mtgox move to the new chain, there will be other exchange for original chain and the price of original bitcoin stay strong

In the normal development of a fast transaction system, the transaction system itself is seperated from the value it transfer, but in bitcoin's case, it is much more complicated, the system itself will affect the value of bitcoin and it's successfulness will depend on other factors than pure technology

[markm]

Wow nice idea johnyj, I like it! Well done!

Of course whether it will play out that way remains to be seen, but really nice story, especially if it does play out.

-MarkM-

[misterbigg]

There's one possibility that I'm surprised no one has mentioned. That the rise and performance of Bitcoin as a store of value (forget about fast payments) creates competitive pressure on fiat currencies to stop inflating and remove capital controls. Wishful thinking? Or could fiat currencies become the "alt chains" needed to fill in the gaps?

[MoonShadow]

There's one possibility that I'm surprised no one has mentioned. That the rise and performance of Bitcoin as a store of value (forget about fast payments) creates competitive pressure on fiat currencies to stop inflating and remove capital controls. Wishful thinking? Or could fiat currencies become the "alt chains" needed to fill in the gaps?

They already are.  Bitcoin is not intended to, nor is it likely to, replace national currencies.  It's going to be a long time coming before Bitcoins displace paper cash in small daily transactions.  That said, it's not unreasonable to assume, should the central banking model not be able to adjust to the reality of Bitcoin (and presumedly not be able to destroy it), that new paper cash currencies that claim a reserve backing of bitcoins that displace national fiat currencies.  So yes, in many cases simple paper cash would function as an "alt-chain" just fine; although I'd predict that should a bitcoin backed banknote ever come into existance, it's more likely to be a plastic token with a digtial artifact imbedded into it, probably using many of the very advanced security features that casino tokens use today.  And just like the gold in a vault never actually moves under a (auditable) gold standard, the bitcoins that presume to back those banknotes would rarely, if ever, need to move across the blockchain.  The important part to all this is that, in the end, any peer forever has the ability to move transactions across the blockchain.  That is the part that makes it 'decentralized', not the number of network nodes (although that is an important minimum) or the common size of a mining operation.  The important part is that there are no 'super-nodes' with special abilities to ignore or exclude other nodes from the network.  If the block-max-size issue ever started to trend toward nodes that were so much larger than the entry level node as to functionally have the ability/incentive to ignore the small nodes, and have that actually work, then I'd be concerned.  I do not claim to know what is the best method for increasing the blocksize limit, but I do not believe that increasing it to 10mb, 20mb or even 50mb as a hard limit would be a dramatic burden on most full nodes.  Perhaps a combo solution; a higher than currently reasonable hard limit combined with an adjustment algo for a soft limitation.  For example, the hard limit could be set to something like 250Mb, and an algo that can adjust the soft limitations within that hard limit withn a range based upon, perhaps, the average number of blocks delayed that a minimum fee transaction experiences during the prior two week period. 

Although I havn't a clue how that could be actually implimented.

[Gavin Andresen]

RE: particular ideas if "we" decide the blocksize has to be increased:

I think the first step is to come to rough consensus that, at some point, we WILL need a hardfork to increase the block size limit.

If we can come to rough consensus on that, then we can figure out the safest way to accomplish that.

I don't think we'll get consensus; retep and others will argue "we need to run into the hard limit to FORCE alternatives to be created first."

I keep saying we should see what happens as we run into the soft blocksize limits.  To people on both sides of this debate:  what do you predict will happen?

If what you predict will happen doesn't actually happen, will that make you re-evaluate your position?

(I haven't spent enough time thinking about this problem to answer those questions, but that is how I'm going to think about it).