What are the downsides to 8MB blocks?

[johnyj]

In fact, the size of the block is not limited by either core devs or miners, but ISPs around the world. I believe without major upgrade of internet infrastructure, the bitcoin network can not handle 2MB blocks well, not even mention 8MB

When there was a talk about 20MB blocks, people calculated that they would barely fit into the "total bandwidth per month cap" imposed by ISPs.
So, I assumed that 8MB should fit that situation better, provided that it is only a cap and not the actual block size for the next few years.

Yes, people also calculated that they can do space travel to another galaxy through a worm hole  

Until you have done a stress test under the real traffic, you would never know what kind of problem can arise. A small increase in the amount of the transaction might cause a total change in system's behavior. You can test the technical aspect of the network in a test lab, but you can never anticipate what real people would do when they detected there is a chance to clog and crash the network and greatly profit from it

A possible failure scenario: After raising the block size to 8 MB, there is nothing wrong, people start to develop new solutions and new business around the new blockchain, but after another 5 years, there is a heavy increase of the network traffic but the internet infrastructure upgrading has totally stopped due to a new financial crisis and recession, ISPs start to cut off bandwidth and charge more on it, then at that stage, you can not reduce the block size to cope with the network change, the whole network are under traffic jam for another 5 years

A service oriented mindset might not suitable for a monetary system, you should let people adopt to bitcoin, not let bitcoin adopt to people. Banks never adopt to people, but people worship them

[coalitionfor8mb]

In fact, the size of the block is not limited by either core devs or miners, but ISPs around the world. I believe without major upgrade of internet infrastructure, the bitcoin network can not handle 2MB blocks well, not even mention 8MB

When there was a talk about 20MB blocks, people calculated that they would barely fit into the "total bandwidth per month cap" imposed by ISPs.
So, I assumed that 8MB should fit that situation better, provided that it is only a cap and not the actual block size for the next few years.

Yes, people also calculated that they can do space travel to another galaxy through a worm hole  

Until you have done a stress test under the real traffic, you would never know what kind of problem can arise. A small increase in the amount of the transaction might cause a total change in system's behavior. You can test the technical aspect of the network in a test lab, but you can never anticipate what real people would do when they detected there is a chance to clog and crash the network and greatly profit from it

A possible failure scenario: After raising the block size to 8 MB, there is nothing wrong, people start to develop new solutions and new business around the new blockchain, but after another 5 years, there is a heavy increase of the network traffic but the internet infrastructure upgrading has totally stopped due to a new financial crisis and recession, ISPs start to cut off bandwidth and charge more on it, then at that stage, you can not reduce the block size to cope with the network change, the whole network are under traffic jam for another 5 years

A service oriented mindset might not suitable for a monetary system, you should let people adopt to bitcoin, not let bitcoin adopt to people. Banks never adopt to people, but people worship them

Your space travel analogy is not that far fetched from reality.
But, I will be blunt and twist this way.

1) It's Bitcoin's first time. That's why there is so much confusion and hesitation.
Should we do it this way or that way? Maybe we should wait a little longer and see if the urge (stress-test) goes away?

The alternatives that we are presented with currently aren't very bright:
Turn Bitcoin into a hooker (XT-exponent), or let it become an old maid (Core-1MB).
I apologize if that doesn't fit the style of the conversation here, but the truth is the truth.

2) You might like another analogy a bit better. It's a game of Chess, and now is Bitcoin's turn to make a move.
The are multiple ways we can approach this issue, as there are multiple figures on the chessboard we can move.

The move to 8MB in question is intended to save one important figure on Bitcoin's side of the game.
We are talking about the first mover advantage here. If we allow the competition to gain enough momentum
as a transactional medium, we may never re-gain that due to the open source nature of the systems in question.

All in all, there is always some risk.
But that's why we love the game!

[RoadTrain]

Why exactly will they get orphaned? I consider this statement unfounded, because nowadays miners are able to propagate blocks between themselves in constant time (generally) with help of Matt Corallo's relay network. In turn, the nodes don't have this privilege, so we better have some testing, I guess.

Ok, there are finally some good points here!
We definitely need some testing.

But if the "constant propagation time" for miners is true, why Chinese are then worried about the bandwidth with their 8MB figure?
What would prevent implementing the relay network for other full nodes and what was the rationale for not doing so in the first place?

The thing about the relay network is that it depends on miners behaving cooperatively. On the other hand, some game-theoretical research (e.g. selfish mining paper) suggests that this is not necessarily the most rational approach.
So, my point is that we must be very careful about changing these constants. I'd rather go with 2MB first, not 8MB, and then re-evaluate.

I also thought about more conservative approach first, but then quickly realized that being over-protective would necessitate having this discussion (about the limit increase) more often than we want it to be.

In the extreme case, we will be constantly debating the issue and not let the actual limit to solidify in order for it to have any effect, which at the same time would distract us from doing any other positive and creative things in the ecosystem.

If 8MB is considered unsafe, then we should look at how competition is currently surviving with higher limits (not blocks though). Of course, the counter argument in this case will be that their market cap and importance are less than that of Bitcoin and therefore there are no incentives to disturb the network by storming it with bigger blocks.

It's still somewhat uncertain, but if we demonstrate that we can agree on a single number and move forward, then the positive momentum in the ecosystem might encourage other players (like bandwidth-limited Chinese miners) to protect the network from bloating too fast by building on top of blocks that are not bigger than 2MB, for example.

That's the scenario I keep in mind when talking about 8MB limit.

I think if we succeed with 2MB, and it goes smoothly, the next time will be much easier, because it won't be uncharted territory.
If that's deemed too conservative, we could do 2-4-8 programmed instead with doubling every 2 years. After all, we can alwasy soft-fork to lower limit, though I think it's inelegant.

[brg444]

Why exactly will they get orphaned? I consider this statement unfounded, because nowadays miners are able to propagate blocks between themselves in constant time (generally) with help of Matt Corallo's relay network. In turn, the nodes don't have this privilege, so we better have some testing, I guess.

Ok, there are finally some good points here!
We definitely need some testing.

But if the "constant propagation time" for miners is true, why Chinese are then worried about the bandwidth with their 8MB figure?
What would prevent implementing the relay network for other full nodes and what was the rationale for not doing so in the first place?

The thing about the relay network is that it depends on miners behaving cooperatively. On the other hand, some game-theoretical research (e.g. selfish mining paper) suggests that this is not necessarily the most rational approach.
So, my point is that we must be very careful about changing these constants. I'd rather go with 2MB first, not 8MB, and then re-evaluate.

I also thought about more conservative approach first, but then quickly realized that being over-protective would necessitate having this discussion (about the limit increase) more often than we want it to be.

In the extreme case, we will be constantly debating the issue and not let the actual limit to solidify in order for it to have any effect, which at the same time would distract us from doing any other positive and creative things in the ecosystem.

If 8MB is considered unsafe, then we should look at how competition is currently surviving with higher limits (not blocks though). Of course, the counter argument in this case will be that their market cap and importance are less than that of Bitcoin and therefore there are no incentives to disturb the network by storming it with bigger blocks.

It's still somewhat uncertain, but if we demonstrate that we can agree on a single number and move forward, then the positive momentum in the ecosystem might encourage other players (like bandwidth-limited Chinese miners) to protect the network from bloating too fast by building on top of blocks that are not bigger than 2MB, for example.

That's the scenario I keep in mind when talking about 8MB limit.

I think if we succeed with 2MB, and it goes smoothly, the next time will be much easier, because it won't be uncharted territory.
If that's deemed too conservative, we could do 2-4-8 programmed instead with doubling every 2 years. After all, we can alwasy soft-fork to lower limit, though I think it's inelegant.

We need no more than 2 years to solve the issue forever. By then hopefully payment channels should be integrated in most wallets and easily accessible. Their use will be a no-brainer for the user as they will be considerably faster, cheaper and generally more efficient.

All the noobs now are scared because there is no immediate and obvious alternative for them to use to buy candies.

When we feed these shiny new toys to the reddit crowd I expect they stop pouting and shut up.

[coalitionfor8mb]

I think if we succeed with 2MB, and it goes smoothly, the next time will be much easier, because it won't be uncharted territory.
If that's deemed too conservative, we could do 2-4-8 programmed instead with doubling every 2 years. After all, we can alwasy soft-fork to lower limit, though I think it's inelegant.

Ok, I see.

I wasn't aware of 2-4-8 programmed, so that might be another good candidate.
For as long as we don't keep the 1MB forever and instead provide competitive enough capacity in the category of PoW-secured single ledgers, we will stay in the game, the dynamics of which will be interesting to watch unfold.

We need no more than 2 years to solve the issue forever. By then hopefully payment channels should be integrated in most wallets and easily accessible. Their use will be a no-brainer for the user as they will be considerably faster, cheaper and generally more efficient.

All the noobs now are scared because there is no immediate and obvious alternative for them to use to buy candies.

When we feed these shiny new toys to the reddit crowd I expect they stop pouting and shut up.

Haha!

The only way to "solve the issue forever" is to get yourself out of the game and even that won't last very long.
There is no such thing as a permanent solution. The existential paradox keeps constantly resolving itself.

I will be blunt again and say that hard forks are healthy if done regularly but not too often.
Without them we will be constantly (mass-)debating about the issue.  

[johnyj]

The biggest problem of a drastic change in block size is its lacking of consistency. People trust banks because they are very consistent and stable over decades, they never make sudden moves. Money is all about trust, without time tested trust, the value of bitcoin will stay at game level

When you plan an IT infrastructure's capacity, you never use more than 30% of its planned capacity, because you must leave lots of room for emergencies. So if average home bandwidth limit allow you to run 8MB blocks, then 2MB blocks will leave you enough room to deal with emergency situation where bandwidth suddenly dropped due to unforeseeable events (I still remember that during 2007 under sea fiber optic cables incident, there was almost no way that you can connect to Taiwan and Chinese mainland network in any speed higher than 56k dial up)

[coalitionfor8mb]

The biggest problem of a drastic change in block size is its lacking of consistency. People trust banks because they are very consistent and stable over decades, they never make sudden moves. Money is all about trust, without time tested trust, the value of bitcoin will stay at game level

When you plan an IT infrastructure's capacity, you never use more than 30% of its planned capacity, because you must leave lots of room for emergencies. So if average home bandwidth limit allow you to run 8MB blocks, then 2MB blocks will leave you enough room to deal with emergency situation where bandwidth suddenly dropped due to unforeseeable events (I still remember that during 2007 under sea fiber optic cables incident, there was almost no way that you can connect to Taiwan and Chinese mainland network in any speed higher than 56k dial up)

I see your point.

I would like to define Bitcoin first as precisely as possible. Attention to details here is crucial.
Bitcoin was created as a "single PoW-secured transparent unified ledger runnable by users at home".
That's the niche it came here to fill and that's what it's known and valued for. This has proven itself to work.

Now, with technology constantly improving itself that niche continues to grow, while Bitcoin has approached its temporary limit.
We need to find a right balance between the two opposites here.

If we over-protect ourselves and let the gap between Bitcoin's current capacity and that of the home-based internet infrastructure grow too big,
the competition will take the chance and begin filling that niche not leaving any empty space for Bitcoin to grow further. That's where Bitcoin's game turns bleak.

If we, on the other hand, relax the idea of the limit by removing it altogether (or hooking it to the exponent), we might shift the incentives structure to the point, that the network will begin consolidating around high-bandwidth lanes, leaving the home-based demographic out of the picture. That's where Bitcoin stops being itself.

If we show ourselves undecided either way, chances are, that the network will split itself into two extremes outlined above and Bitcoin will have to accept a technical defeat. That's why Chess has a clock. Stall for too long and the game is over.

So, if it's shown that 8MB is a reasonable limit to satisfy the conditions above for the next 4-6 years, I argue, that we shouldn't target anything less.

The idea that global internet infrastructure might suddenly shrink is theoretically possible, but since we are targeting home internet connections here, I'm almost certain that 8MB should be workable across the ocean at any given moment. If it isn't (workable), then smaller blocks propagating faster will simply build a longer chain.