review of proposals for adaptive maximum block size

[solex]

markm,

you are introducing a new debate which is about the maximum block size the network can handle. This needs to be kept separate from the max block size constant debate which is big enough already.

It might be that when the average block size is 700Kb the network begins seriously degrading because blocks are not being received/retransmiited effectively, it might be that this happens when they are an average of 12Mb each. If it is the lower size then we have been debating the wrong issue lately. If it is the upper size then, hopefully, the network will be wealthier (I am assuming that most owners of full nodes have BTC holdings) and some of that could be converted to fiat to buy better hardware.

This debate really needs to be in a new thread, and the results of models/testing made known there.

[1714952925]

1714952925

[1714952925]

1714952925

[1714952925]

1714952925

[1714952925]

1714952925

[1714952925]

1714952925

[Monster Tent]

You'll all have to excuse my stupidity, but what's wrong with unlimited (*) blocks? Let each miner set his own transaction fees and free market competition will ensure that transaction fees are kept low while still keeping the network secure. Surely putting some artificial limit on blockchain size in order to drive up fees is little different to central bankers imposing QE or inflation targets on a currency.

Of course, bitcoin mining will migrate to where there is cheap energy, but this might have a beneficial side effect - more effort will be put into building cheap energy sources (read: renewable (**)) so any given geographical region can contribute to mining and so benefit from incoming tx fees ("free money" for want of a better description).

(*) up to some reasonable limit that shouldn't ever be reached, which will depend on how widespread bitcoin adoption is - visa handles a peak of about 10000 tx per second, assume normal use is 3000 tx/sec, assume 250 bytes per transaction and you'd need on average 0.5GB sized blocks every 10 minutes. For the immediate future, I think there's no reason the blocksize should increase beyond 10MB at the outside.

(**) hopefully renewable but admittedly, not necessarily so.

Why should ordinary users be forced to subsidise visa or even SD spam ? This is akin to the current system where corporations offload as much cost and risk as possible onto the public via government subsidies derived from taxation.

[hazek]

You are still talking about the ideal size of blocks miners should make if they want mining to be a viable business.

The hard coded block size limit is not motivated by that.

I realize that but looking to get a certain amount of fees before a higher limit is allowed ensures that the networks gets a certain amount of funding to be hopefully able to buy the necessary hardware in order to handle the size limit increase. How high and how far this can go before it reaches the hardware limits regardless of how much money you throw at it I have no idea.. But that's a different problem entirely.


My goal was only to show a middle road between Bitcoin not scalling within it's peer to peer infrastructure and eventually requiring super expensive transactions fees due to a limited block size and between Bitcoin scaling as much as desired at the expense of many peers and at the expense of the incentive to pay enough fees to ensure miners get the needed funding despite the decreasing subsidy.

If there is such a middle road it has to be with somewhat higher transactions fees as a requirement for the ability to handle more tps.

[SimonL]

I also proposed something similar to Mr Big, but it would also take into consideration whether the difficulty has gone up or down between the current and last period. All of the proposals thus far consider the economic case, but not the propagation health of nodes in the network, and I reckon that difficulty can actually be used as a canary in the mineshaft warning that increasing the max block size, even given other conditions being met, will potentially cause network problems and put strain on nodes in the network because during that period were not able to safely propagate blocks fast enough and the max block size increase should be either avoided or even reduced in such cases, differences in difficulty and the sizes of blocks over the current and previous difficulty periods should reflect this.

I mention it briefly here:
https://bitcointalk.org/index.php?topic=145641.msg1555072#msg1555072

And fully flesh out my argument for incorporating difficulty here:

https://bitcointalk.org/index.php?topic=144895.msg1538386#msg1538386

Sorry if I sound like a broken record re-mentioning this again, but no-one has even commented or criticised why my approach is good or bad, makes me wonder if I'm deluding myself.

[markm]

Exchange rates determine whether a decreasing subsididy is a larger or smaller actual real income; double the exchange rate every four years and the subsidy is not shrinking at all, it is growing. Quadruple the exchange rate over the course of four years and the halved subsidy is actually double what the previous halving had adjusted it to.

So quite possibly mining will be completely covered by subsidies, maybe even ever-growing subsidies, until all the coins have been minted, in what is that, 136 years or so from now?

If the block size is to increase to fit more transactions into the money-supply, then unless we deliberately try to spam that size with trivial transactions wouldn't it be more reasonable to suspect that each increase in exchange rate could drive block size up, maybe not linearly but maybe somewhere between some value just under linear and some value down near the square root (some exponential function anyway) of the amount the exchange rate multiplied?

(Four times exchange rate = size multiplies by something between two and somewhat less than four.)

-MarkM-

[hazek]

So quite possibly mining will be completly covered by subsidies, maybe even ever-growing subsidies

False.

A certain amount of mining will be covered which we cannot know whether or not it would be enough to not give the proper profit incentives to someone thinking about attempting an attack. With the growth of value of bitcoins it becomes more and more lucrative to perform an attack therefor mining should also bring in more and more profits encouraging more and more mining.

My proposal at max costs 12,5% and at minimum 6,25% of the entire monetary base anually (if the block size has been increased at least once). It's pretty hard to imagine someone would want to spend 25% or at min 12,5% of the value of the entire monetary base just to perform an attack. Now maybe this is overkill, but how can we know? How can we know at what % of the entire monetary base spent on mining is enough security?

[Sukrim]

The name of the game is keeping the block space as scarce as possible, encouraging fees that will eventually have to cover the costs of securing the network but not making it too scarce so that Bitcoin can scale batter.

If blocks are basically limitless (= have a limit that only serves for DoS protection of somebody mining a 10 TB block, i.e. a hard limit for the next decade of 1 GB/block), your argument would be that this means the total fees per block would be lower than if the limits are tight (= users actually not only have to convince miners with their fees to include their transaction but they also need to make their trasaction as small as possible or for example fitting in between existing transactions or combining a lot of these to make sure they are as efficient as possible)?

Block space currently is quite arbitrarily chosen at a quarter of the actual limit in bitcoin-qt and most miners even follow this ruleset. Still there are fees paid even though there is still quite some room to grow and block subsidy is high.

I'd also like to see a comment on a "too high to reach without malicious intent" block size that does not limit miners to include transactions based on scarce block size but only on their own metrics (e.g. "spammyness", "fees per kB"...). Something maybe along the lines of "10 (100?) times the median (not average, to remove potential outliers) block size of the past 4 years" adjusted together with the block subsidy every 4 years or even adjusted every 2 weeks (then it might be too easy for attackers to shift the median, but it might be easier to recover/grow as well) with the difficulty. Again the goal should be that the block size never limits miners at all on deciding which transactions to include.

[misterbigg]

How can we know at what % of the entire monetary base spent on mining is enough security?

We can't. Which is why I revoked my approval for my own proposal for adjusting the block size in order to drive the sum of the subsidy plus transaction fees per block to a constant amount. Note that my scheme is identical to the one you proposed (it is just more prescriptive with respect to actual implementation).

[misterbigg]

there are fees paid even though there is still quite some room to grow and block subsidy is high.

Did we forget that clients will only relay transactions through the overlay network if they have a minimum fee attached (computed based on the size of the transaction and the age of the coin)?

Tsk...tsk...

[theymos]

I like Gavin's proposal.  (I mean his actual proposal, not the "half-baked thought" quoted above.)

No hard limit, but nodes ignore or refuse to relay blocks that take too long to verify.  This discourages blocks that are too large, and "spam" blocks containing lots of transactions not seen on the network before.

This might create an incentive to mine empty blocks.  To discourage this, in the case of competing blocks, nodes should favor the block that contains transactions they recognize, and ignore (or delay relaying) the empty block.

That might work, though I'm worried that with such rules blocks would become gradually larger and over time the number of full nodes would shrink dramatically as weaker computers get separated from the network. For example, dial-up nodes would get separated right away. No one would care that dial-up users can no longer run full nodes, and they would themselves mostly just say, "Oh well, I guess my setup is too slow to run a full node. Time to switch to a lightweight node." This is probably reasonable for dial-up, but I think that it might over time spread to most people. As blocks become larger, people on average PCs would have to switch to lightweight nodes, then even hobbyists, and then even small businesses.

Maybe there should just be a planned hardfork every 4 years (or whatever) where the max block size is set to something reasonable and other fixes are made.

[mughat]

The miner that mines a block should be able to choose the block size.
Max size limited to: ( previous_block_size * 2 )

If most miners choose to double the size more transactions can be processed.
If most miners keep a small block size fees will have to go up to incentivise the miners to increase the block size.

The miners have an incentive to keep the block size to a realistic size to keep the network working and also earn as much as possible by processing as many transactions as possible.
This will let the market deside the best block size.

[Sukrim]

there are fees paid even though there is still quite some room to grow and block subsidy is high.

Did we forget that clients will only relay transactions through the overlay network if they have a minimum fee attached (computed based on the size of the transaction and the age of the coin)?

Tsk...tsk...

This is not a hard rule and you could directly connect to a "friendly" miner who also mines spammy transactions to get these into the block chain. Also I can be that miner myself. This is like the 250kB "limit" that currently seems to be something where people bump against.

I see your point that the network forces some people right now to pay fees, but the same effect could be done if miners start to claim that they won't mine these spammy transactions.

[markm]

The miner that mines a block should be able to choose the block size.
Max size limited to: ( previous_block_size * 2 )

If most miners choose to double the size more transactions can be processed.
If most miners keep a small block size fees will have to go up to incentivise the miners to increase the block size.

The miners have an incentive to keep the block size to a realistic size to keep the network working and also earn as much as possible by processing as many transactions as possible.
This will let the market deside the best block size.

No. First, as a matter of national security, we need a hard limit no terrorist miner coalition or axis of allied enemy nations can exceed to disable the nation's financial infrastructure.

Within that, miners can go ahead and double and halve and bicker and moan all they like.

-MarkM-

[mughat]

The miner that mines a block should be able to choose the block size.
Max size limited to: ( previous_block_size * 2 )

If most miners choose to double the size more transactions can be processed.
If most miners keep a small block size fees will have to go up to incentivise the miners to increase the block size.

The miners have an incentive to keep the block size to a realistic size to keep the network working and also earn as much as possible by processing as many transactions as possible.
This will let the market deside the best block size.

No. First, as a matter of national security, we need a hard limit no terrorist miner coalition or axis of allied enemy nations can exceed to disable the nation's financial infrastructure.

Within that, miners can go ahead and double and halve and bicker and moan all they like.

-MarkM-

Well. The scale up factor could be set lower. Example: Max size limited to: ( previous_block_size * 1.1 )

No attacker would be able to mine every block in succesion and increase the size to unrealistic size.
Once in a while a frendly miner would keep the size to a realistic size and prevent a size increase attack.

Problem solved.

[markm]

Compound interest is a bitch.

The progrssive doubling proposals are suspicious for the same reasons, percentages instead doesn't really change the fundamental unsustainability of compound interest.

Better to be a backward nation with an unassailable infrastructure than a wet paper bag any superpower or axis of allied determined enemy nations will blow right through simply as a side-effect of catering to the endless greed/gluttony of their ever more obese moar moar moar proletariat.

-MarkM-

[mughat]

Compound interest is a bitch.

The progrssive doubling proposals are suspicious for the same reasons, percentages instead doesn't really change the fundamental unsustainability of compound interest.

Better to be a backward nation with an unassailable infrastructure than a wet paper bag any superpower or axis of allied determined enemy nations will blow right through simply as a side-effect of catering to the endless greed/gluttony of their ever more obese moar moar moar proletariat.

-MarkM-

You give no logic reasons.
The miners have incentives to keep the size to a realistic size and also include as many transactions as possible.

The market will clearly find the right size based on what is possible and what makes the most money. Evil miner Attackers that can create 10 blocks in a row and increase size to problematic size is not realistic.

[Ari]

That might work, though I'm worried that with such rules blocks would become gradually larger and over time the number of full nodes would shrink dramatically as weaker computers get separated from the network. For example, dial-up nodes would get separated right away. No one would care that dial-up users can no longer run full nodes, and they would themselves mostly just say, "Oh well, I guess my setup is too slow to run a full node. Time to switch to a lightweight node." This is probably reasonable for dial-up, but I think that it might over time spread to most people. As blocks become larger, people on average PCs would have to switch to lightweight nodes, then even hobbyists, and then even small businesses.

We're a long way from that.  Even something at the scale of Paypal (~85 tps) could run on an average desktop PC with 1mb/sec internet connection.

The post from Gavin that I was referring to is here:

https://bitcointalk.org/index.php?topic=140233.msg1503099#msg1503099

[markm]

51% attackers were not realistic a while ago, then came ASICs. Once we get tons of ASICs in place, 51% attacks will again start to look unrealistic. What is realistic is the concept of detemined, fanatical foes, even ones willing to sacrifice not just other people's lives but their own. Is suicide bombing realistic? Yet it happens.

As soon as we have some manipulable (and to adapt really just means to be manipulated, whatever is being adapted to is the manipulating variable) increase system, that is a whole new vector for realistic and unrealistic attacks.

-MarkM-

[mughat]

51% attackers were not realistic a while ago, then came ASICs. Once we get tons of ASICs in place, 51% attacks will again start to look unrealistic. What is realistic is the concept of detemined, fanatical foes, even ones willing to sacrifice not just other people's lives but their own. Is suicide bombing realistic? Yet it happens.

As soon as we have some manipulable (and to adapt really just means to be manipulated, whatever is being adapted to is the manipulating variable) increase system, that is a whole new vector for realistic and unrealistic attacks.

-MarkM-

The best way to protect agains an attack is to let the system grow as fast as possible. Limiting growth in any way increases the chances of an attacker to catch up to the system.

A free market without price controls and arbitrary limits is the fastet growing. Limiting size will slow growth.

That is why we need the miners to be able to bump up size when needed and not wait months for the community to agree.

[markm]

Then maybe it really is best to just let whoever can manage to build petabyte blocks do so as soon as possible to raise the bar as high as possible as fast as possible?

Its not like there aren't alt-chains nations of various sizes, technology-levels and industrial base can use if the G7's ideal uber-network is a bit too much for them to install/use domestically.

-MarkM-