When Will Larger Blocks Be "Okay"?

[titular]

I have been re-reading many of the old BIPs and I stumbled back upon BIPs 100-109.

For those of you who remember, BIPs 100-109 are a collection of attempted hard forks with the aim of increasing the max block size from 1mb. Clearly, these proposals were rejected.

I know this community is not very keen on increasing the block size, but it seems that everyone agrees it will need to be addressed in the future.

But when does this become "okay"?

When does it become okay to increase the block size, and by what increment? Or is it completely unnecessary because of L2's?

I understand the centralization issues a larger block brings, but how bad could it actually get from a small increase?

Educate me!

[a.a]

Loot at Bitcoin Cash and Bitcoin Satoshi Vision. Those are BlockChains with increased Block Size.

[Trojane]

When does it become okay to increase the block size, and by what increment?

Fine,I'd say it isn't okay to increase block size perhaps you have gotten a clue about the damages it causes and of course the demerit too. If you haven't this is it : bigger blocks may possibly slowdown or even end market transactions as the would not be competition and it might go as far as diminishing market turnover,reducing transactions fees because miners increase as blocks do,then finally reducing bitcoin to a minor price-level which is what every trader is avoiding.

Or is it completely unnecessary because of L2's?

To me I'd say 'Yes for now till bitcoin stands a better chance.

I understand the centralization issues a larger block bring, but how bad could it actually get from a small increase?

If bitcoin is centralised, even from the cause of mismanagement then it isn't way better than fiat money.
There's no small increase op miners had to opt these process of inducing block increase. According to these link below,they were trying also to avoid 'spam transactions'. It could do alot of harm even if it increases an inch for now
Read wide on these link below
https://bitcoin.stackexchange.com/questions/36085/what-are-the-arguments-for-and-against-the-increase-of-the-block-size-limit

[garlonicon]

But when does this become "okay"?

It could be "never" if some improvements will be done. For example, if transaction joining would be possible, then you could compress a huge chain of transactions into small and simple transaction, for example if you have Alice->Bob->Charlie->...->Zack, then it could be replaced with Alice->Zack transaction.

In Litecoin, Mimble Wimble Extension Blocks will be available soon, in Bitcoin, taproot will be available soon, such improvements could be used to compress some data. BTC is going to scale not by increasing block size, but by compressing that data. Lightning Network is in fact some kind of data compression, where you can compress N transactions into two transactions that open and close some channels.

Also note that it is still possible to use some older version with 1 MB blocks and without Segwit. If all future changes will be backward-compatible, so that each block size increase will be done in a soft-fork way, old nodes will see that block size will never be increased.

[pooya87]

But when does this become "okay"?

One problem I see (among others) that people don't talk about is the "fear of hard fork". Increasing block size requires everyone to upgrade their software since the fork would be backward incompatible, for for example if a miner remains on the old software the chain will split.
Some consider this a big negative.

When does it become okay to increase the block size, and by what increment?

Well we increased the block size back in 2017 but through a soft fork meaning the more people use SegWit the bigger the block size is going to get.

Or is it completely unnecessary because of L2's?

It is actually necessary because of layer 2 since people have to open and close channels. It is layer two not a stand alone network, it is built on top of "layer 1" and relies on it.

I understand the centralization issues a larger block brings, but how bad could it actually get from a small increase?

Whether it is a small increase or a big one, if it is done through hard fork it will have the same problem as I mentioned above.

Lightning Network is in fact some kind of data compression,

That is a wrong analogy. LN removes all those "extra" transactions from main-chain. You can open a channel with a single on-chain transaction and then make a million transactions on second layer.

[Wind_FURY]

But when does this become "okay"?

OP, ignoring all potential problems from/because of a hard fork, I believe no one truly knows, or no one has a definite reply. We plebs can give you many different replies, and all of them will not be wrong, but none of them will truly be the right one either. It might depend on how you value Bitcoin’s main value proposition.

[takuma sato]

As someone that saw all the iterations of hard fork attempts back then I would personally have it fixed to 1MB forever. Most people don't even want to transact, they hold it, and whenever you start with the next hard fork attempt crisis prices plummets, which is against the agenda of any hodler. Also, by looking at actual demand, the mempool is pretty empty:

https://mempool.space/

As of right now you can get a transaction going in 10 minutes for 9sats/byte. There is no demand for hard fork and time has proven that the blockchain was getting spammed by Roger Ver and Jihan Wu et al back then. And even if there were high fees it would still be debatable that this means a hard fork is urgent.

At this point it's probably not even possible to ever reach supermajority for a hard fork increase thus proving Bitcoin is in fact decentralized. Perhaps the only thing that would get everyone on board was a global ban on physical cash, then the only way out for financial privacy would be Bitcoin, at this point it would be an organic demand for a block space and maybe there would be supermajority reached.

[miner2251]

At this point it's probably not even possible to ever reach supermajority for a hard fork increase thus proving Bitcoin is in fact decentralized. Perhaps the only thing that would get everyone on board was a global ban on physical cash, then the only way out for financial privacy would be Bitcoin, at this point it would be an organic demand for a block space and maybe there would be supermajority reached.

Even in that case, creating an "evil" soft fork could still be better than hard fork. But I guess that something like an extension block could be a better solution, because then you have to/from transactions on-chain and everything else can happen inside this additional block, that can be visible only by upgraded clients and can be for example 1 GB in size if needed.

[pooya87]

As someone that saw all the iterations of hard fork attempts back then I would personally have it fixed to 1MB forever.

Bad news, the block size hasn't been capped at 1MB ever since mid 2017

Most people don't even want to transact, they hold it, and whenever you start with the next hard fork attempt crisis prices plummets, which is against the agenda of any hodler.

A holder doesn't care about short term fluctuations. You are also making an assumption based on no proof that "most people" don't want to transact.

Also, by looking at actual demand, the mempool is pretty empty:
As of right now you can get a transaction going in 10 minutes for 9sats/byte.

You just contradicted yourself. Fees in a "pretty empty" mempool would be 1 sat/vbyte not 9.

There is no demand for hard fork and time has proven that the blockchain was getting spammed

This I can agree with. There is no urgency to implement a hard fork to increase the block size, specially since we are getting bigger blocks and use the block space more efficiently with soft forks but that doesn't mean there is no demand for it. In fact if you look at the blocks ever since the 2017 spam attack ended, majority of them are full which proves that we are at capacity already.

[HeRetiK]

But when does this become "okay"?

Looking at current community stance, i guess either when
1. Most scaling solution already deployed and widely adopted.
2. There's huge spike on Bitcoin adoption / usage.

I believe it will require a little bit of both.

The first, because without economic incentive there would be no reason for people to upgrade (similar to how merely adding new highway lanes doesn't help congestion as car traffic increases to meet up with supply)

The second, because blocks need to remain somewhat filled as otherwise there'd be no money for miners as block subsidies subside.

Either way it's probably going to be a few more years, if not a decade or so, until a potential blocksize increase will be revisited.

[kano]

...
The first, because without economic incentive there would be no reason for people to upgrade (similar to how merely adding new highway lanes doesn't help congestion as car traffic increases to meet up with supply)

The second, because blocks need to remain somewhat filled as otherwise there'd be no money for miners as block subsidies subside.

Either way it's probably going to be a few more years, if not a decade or so, until a potential blocksize increase will be revisited.

Alas, clearly many people haven't been around bitcoin for more than a few weeks ...

There have been multiple times even this year, that the mempool has been too full and people have turned away from using Bitcoin to go to scamcoins or back to fiat due to the transaction fees being way too high.

When that happens and then it dies down, that simply means it failed and people went elsewhere.

The core developers like to think they've done all that is necessary and pretend it doesn't happen.
Everyone else seems to ignore the fact that it does happen and just stare and drool at the price ... and think 'money money money'

[stompix]

The first, because without economic incentive there would be no reason for people to upgrade (similar to how merely adding new highway lanes doesn't help congestion as car traffic increases to meet up with supply)

Oh god!!!, again with this argument for the public transport against the ones not wanting to depend on sticking busses and metro.

So, if we increase the number of lanes on all highways in the US by ten, the number of cars will also increase from 600 per thousand people to 6000/1000?
And every single citizen will drive around in six cars at the same time to increase congestion? 

Then probably the only solution to get rid of congestion would be to get id oh highways, so how about we lower the blocks to 100kb so more poeple would migrate to LN? Of course, at this point, we have to ignore the fact that in current conditions without using a centralized solution there is no way in hell for even 10% of the world population to migrate to the LN in the next two decades.

[HeRetiK]

Looks like the current community stance is already closer to rethinking future blocksize increases than expected?

There have been multiple times even this year, that the mempool has been too full and people have turned away from using Bitcoin to go to scamcoins or back to fiat due to the transaction fees being way too high.

When that happens and then it dies down, that simply means it failed and people went elsewhere.

The mempool mostly filled up due to heavy price fluctuations leading to people moving coins on and off exchanges rather than "regular" usage though. Not even mentioning the sudden hashrate drop when Chinese miners went offline due to the power outage in April.

I'm not saying that full blocks don't exist, I'm just saying that it's not the every day state of Bitcoin. Despite everything average Bitcoin fees (in terms of BTC) have gone down compared to the last time we saw these type of market fluctuations in 2017.

The first, because without economic incentive there would be no reason for people to upgrade (similar to how merely adding new highway lanes doesn't help congestion as car traffic increases to meet up with supply)

Oh god!!!, again with this argument for the public transport against the ones not wanting to depend on sticking busses and metro.

So, if we increase the number of lanes on all highways in the US by ten, the number of cars will also increase from 600 per thousand people to 6000/1000?
And every single citizen will drive around in six cars at the same time to increase congestion?  

Then probably the only solution to get rid of congestion would be to get id oh highways, so how about we lower the blocks to 100kb so more poeple would migrate to LN? Of course, at this point, we have to ignore the fact that in current conditions without using a centralized solution there is no way in hell for even 10% of the world population to migrate to the LN in the next two decades.

That's a strawman you're hitting, not the actual argument.

Point being, if you increase the blocksize limit to a value that makes on-chain transactions just as cheap as LN transactions then no one's going to use LN. Why should they? There's nothing to gain except a warm fuzzy feeling that you kept the blockchain just a bit smaller. The tricky part is finding the sweet spot in terms of blocksize with the added challenge that increasing blocksize is pretty much a one-way street.

[NotATether]

Point being, if you increase the blocksize limit to a value that makes on-chain transactions just as cheap as LN transactions then no one's going to use LN. Why should they? There's nothing to gain except a warm fuzzy feeling that you kept the blockchain just a bit smaller. The tricky part is finding the sweet spot in terms of blocksize with the added challenge that increasing blocksize is pretty much a one-way street.

Bitcoin Cash already did that, however because their userbase is nowhere near as large as ours, we can't properly compare our average transaction fee to theirs. Theirs is about $0.01 USD or a penny - but we can't compare this to our $5 $6 $7 because the BCH network doesn't have a comparable transaction volume.

[stompix]

That's a strawman you're hitting, not the actual argument.

I've taken your example and I simply projected the results of an increase.
Where is the strawman argument?
If you think that once we built twice as many highways and roads we will have twice the traffic you can show us from where we will get those numbers.
If we make three new highways between Ney York and Philadelphia will they all 3 be congested the same as the current roads while the rest keeping the same?

Sorry, but yours is one, not mine!
The exaggeration is exactly in your argument, that somehow twice as many poeple will be on the roads just because there are more roads.

The tricky part is finding the sweet spot in terms of blocksize with the added challenge that increasing blocksize is pretty much a one-way street.

Since now we have around 10-20% maximum spare capacity before fees spiking and another maybe 10-20% till making even opening an LN channel too pricy, for the fun of it saying that one user opens and closes only a channel a year, can you let me know in what century will the entire population of the earth manage to get in? Oh, let's drop just to that of Europe because obviously there is a catch in that!   I have a feeling that sweet spot is nowhere near 1 or 4 Mb.

[HeRetiK]

I've taken your example and I simply projected the results of an increase.
Where is the strawman argument?

The strawman is in that you focused on the example (traffic) rather than the core of the argument (blocksize and L2).

If you think that once we built twice as many highways and roads we will have twice the traffic you can show us from where we will get those numbers.
If we make three new highways between Ney York and Philadelphia will they all 3 be congested the same as the current roads while the rest keeping the same?

Sorry, but yours is one, not mine!
The exaggeration is exactly in your argument, that somehow twice as many poeple will be on the roads just because there are more roads.

Absurdities aside though, induced demand is a well known and well studied real world phenomenon:
https://en.wikipedia.org/wiki/Induced_demand

There may be short term relief, but in the long term you're generally not much better off than before, congestion wise.

The tricky part is finding the sweet spot in terms of blocksize with the added challenge that increasing blocksize is pretty much a one-way street.

Since now we have around 10-20% maximum spare capacity before fees spiking and another maybe 10-20% till making even opening an LN channel too pricy, for the fun of it saying that one user opens and closes only a channel a year, can you let me know in what century will the entire population of the earth manage to get in? Oh, let's drop just to that of Europe because obviously there is a catch in that!  I have a feeling that sweet spot is nowhere near 1 or 4 Mb.

In the long term Channel Factories should significantly alleviate this part of the problem:
https://wiki.ion.radar.tech/tech/research/channel-factory

I'm not arguing that increasing the blocksize in the future should be completely dismissed, I'm just saying that linear scaling alone won't be enough.

Point being, if you increase the blocksize limit to a value that makes on-chain transactions just as cheap as LN transactions then no one's going to use LN. Why should they? There's nothing to gain except a warm fuzzy feeling that you kept the blockchain just a bit smaller. The tricky part is finding the sweet spot in terms of blocksize with the added challenge that increasing blocksize is pretty much a one-way street.

Bitcoin Cash already did that, however because their userbase is nowhere near as large as ours, we can't properly compare our average transaction fee to theirs. Theirs is about $0.01 USD or a penny - but we can't compare this to our $5 $6 $7 because the BCH network doesn't have a comparable transaction volume.

It's also hard to compare due to the difference in exchange rate. As of right now 1 sat/b is viable with Bitcoin as well, albeit you might have to wait a few blocks. Around 10 sat/b should guarantuee placement in the next block which is only relatively expensive because Bitcoin itself got so expensive.

That's the thing with "people going elsewhere" though, I'm not quite sure where they supposedly went. Because they sure didn't go for Bitcoin Cash and its hard forks. Ethereum, maybe, but in terms of fees they aren't looking so hot right now either. Admittedly there's alts that I personally prefer for moving assets between exchanges, however at least to me they are completely interchangeable with any other alt and nothing I'd hold over the long term.

[stompix]

The strawman is in that you focused on the example (traffic) rather than the core of the argument (blocksize and L2).

You've brought this into the discussion as an example, you deal with the wrongs in it! 

Absurdities aside though, induced demand is a well known and well studied real world phenomenon:
https://en.wikipedia.org/wiki/Induced_demand
There may be short term relief, but in the long term you're generally not much better off than before, congestion wise.

And 90% of those forget to focus on the simple linear or exponential growth that would have happened no matter what. Just like people are wondering why we have so much traffic despite building another road obviously ignoring the fact that we have 3 times more population owning a car.
Exactly how here just because bitcoin transactions would be as cheap as possible it doesn't mean miraculously we will have 10 billion users, how we won't have 300 million cars in NY because we create 400 million parking spaces and how we won't have the world using 30 billions of condoms a day even if they were free for the same obvious reasons.

And you can see this right now, as we're still getting confirmation on 1sat/b transaction.
So for your theory leading to 1 GB blocks that will be filled just as easily as 1MB blocks, don't you think we would first need to fill these?
This is not happening, so where is the induced demand right now?

In the long term Channel Factories should significantly alleviate this part of the problem:
https://wiki.ion.radar.tech/tech/research/channel-factory
I'm not arguing that increasing the blocksize in the future should be completely dismissed, I'm just saying that linear scaling alone won't be enough.

It was segwit and LN that was the solution, then we have to deal with factories, those might now be enough and we might need to increase the block size too..., Don't you think it stating too look like there won't be a final solution and just you keep tunning and tunning for fewer gains till you hit a dead end and you need to start with a better engine?

The whole focus is how we could improve some 10-20% in space used rather than acknowledging the fact that 1TB SSD has dropped to 100$ and 90% of the population would have trouble understanding the basics of LN with 9% probably slapping you if you show them something like this:

[Syke]

Point being, if you increase the blocksize limit to a value that makes on-chain transactions just as cheap as LN transactions then no one's going to use LN. Why should they?

Because LN transactions are faster and private, I will tend to use it first.

[PrimeNumber7]

But when does this become "okay"?

It could be "never" if some improvements will be done. For example, if transaction joining would be possible, then you could compress a huge chain of transactions into small and simple transaction, for example if you have Alice->Bob->Charlie->...->Zack, then it could be replaced with Alice->Zack transaction.

I don’t think there are many chained transactions that involve actual economic transactions that are included in a single block. For example, if I receive a withdraw transaction from Coinbase, I will wait for at least one confirmation before I pay Alice for widgets, and Alice will wait for some number of confirmations before using some of that coin to pay his supplier, Bob, and so on.

As you note, LN somewhat improves upon the scaling problem by consolidating many transactions between Alice and Bob into two transactions. LN has its problems, and drawbacks, but I believe improved scaling solutions will be derived from LN.

I think the next initial step in increasing the max block size is to decrease the weight that signature data counts towards the block size.

From a pure technology standpoint, there isn’t any reason why we couldn’t increase the max block size today, or even years ago. The reason why the max block size should not be raised today, is because doing so would require a hard fork, that hard forks are very costly and there are other alternatives. When there is a hard fork, everyone using bitcoin must upgrade before the hard fork is implemented, or else those that do not upgrade will be unable to use bitcoin. It is very difficult to tell if users will actually accept a hard fork proposal, so there is the potential for a split chain with users and businesses relying on various chains.

If there is some emergency, for example if EDSCA is broken, a hard fork would need to be quickly implemented, and if this happens, we might as well increase the max block size.

[tromp]

Lightning Network is in fact some kind of data compression,

That is a wrong analogy. LN removes all those "extra" transactions from main-chain. You can open a channel with a single on-chain transaction and then make a million transactions on second layer.

Compression is an apt description. When the two parties in a payment channel transact, they don't create a transaction for just that payment, but they update the previously latest settlement transaction into a new latest settlement transaction. These settlement transactions are thus each a cumulative form (i.e. compression) of all the payments made in the channel (which never existed as separate transactions).