"That is a very good reason to oppose increasing the max block size" G.Andresen

[pawel7777]

For all those interested in proposed increase of the max block size, here's today's entry from Gavin's blog:

http://gavinandresen.ninja/utxo-uhoh

UTXO uh-oh…

I’m adding entries to my list of objections to raising the maximum block size as I run across them in discussions on reddit, in IRC, etc.

This is the technical objection that I’m most worried about:

More transactions means more memory for the UTXO database

I wasn’t worried about it yesterday, because I hadn’t looked at the trends. I am worried about it today.
...

--------
As a bonus:

M. Karpeless on the proposed changes:

https://twitter.com/MagicalTux/status/596622731711352832?s=09

[johnyj]

And it is highly likely that future PCs will stuck at 32-64GB memory for 10+ years (Main stream PCs have been mostly capped at 8GB for almost 5 years due to extra memory does not buy you any extra performance), while UTXO database should have grown to over 1TB if it doubles every year

Then 99% of the machine in the world (unless top class business cloud servers) are not able to run a full node if the UTXO database would be put into memory

In fact, besides UHD gaming and heavy graphics processing, any application that can utilize more than 8GB of RAM would require exponentially R&D resources and time to develop, it is unlikely future applications will take significant more memory than they are today

[Kprawn]

The trend has been, for users to adapt to the requirement of the applications running on their computers, not the other way around.

In my student days, I worked in a Computer shop... The sales people normally asked the same questions :

1. What applications will you be running? { Office or Gaming }

When they answer "Office use" they quoted less memory AND when the person was into "Gaming" they quoted more memory.

This should not be a determining factor, to base your decision on... If someone wants to host a node, they should increase the memory.   

[Amph]

And it is highly likely that future PCs will stuck at 32-64GB memory for 10+ years (Main stream PCs have been mostly capped at 8GB for almost 5 years due to extra memory does not buy you any extra performance), while UTXO database should have grown to over 1TB if it doubles every year

Then 99% of the machine in the world (unless top class business cloud servers) are not able to run a full node if the UTXO database would be put into memory

In fact, besides UHD gaming and heavy graphics processing, any application that can utilize more than 8GB of RAM would require exponentially R&D resources and time to develop, it is unlikely future applications will take significant more memory than they are today

can i ask how much memory it is needed to store UTXO if the block size is increased accordingly? right now everyone we are capped at 32 for the majority of the mother board, it would be a problem if we need 64 in a couple of years...

[electerium]

1- Bitcoin transactions are not experiencing exponential growth, even if you push the block size to 20mb.

2- Even for arguments sake, bitcoin will never see sustained (e.g. years) exponential growth- there is a limiting variable-- # of people on earth conducting transactions

3- UTXO is not growing exponentially. For arguments sake, if there is a sudden rise of bitcoin trasnactions within a very short time frame, it is very likely to taper off because of point #2.

4- at 70mib, we are nowhere even close, in relative terms, to the "30% full 1mb block" issue where confirmation times tangibly increase once we hit 40% full. if the average node runs with 2gb dram, we are probably about 1 order of magnitude away from it being on the same level of block size issue. So according to those napkin calculations we are about 3.5 years away from reaching 1gb UTXO.


UTXO is simply another issue that warrants being on the white board while more pressing issues like block size are resolved today.

[johnyj]

1- Bitcoin transactions are not experiencing exponential growth, even if you push the block size to 20mb.

2- Even for arguments sake, bitcoin will never see sustained (e.g. years) exponential growth- there is a limiting variable-- # of people on earth conducting transactions

3- UTXO is not growing exponentially. For arguments sake, if there is a sudden rise of bitcoin trasnactions within a very short time frame, it is very likely to taper off because of point #2.

4- at 70mib, we are nowhere even close, in relative terms, to the "30% full 1mb block" issue where confirmation times tangibly increase once we hit 40% full. if the average node runs with 2gb dram, we are probably about 1 order of magnitude away from it being on the same level of block size issue. So according to those napkin calculations we are about 3.5 years away from reaching 1gb UTXO.


UTXO is simply another issue that warrants being on the white board while more pressing issues like block size are resolved today.

Did you read Gavin's post? He states that today UTXO database is 650MB, and it almost doubled in a year. You don't need more users, just more business activities to make it grow even faster

[redsn0w]

Will this hypothetical improvement help bitcoin to go mainstream? If it will help it, I am sure it is needed but if it is only a thing to make more complicate everything that I do not think the majority of bitcoiners will update their wallet.... or am I wrong?


PS: This is the good thing in the decentralization, no one oblige you to follow an concept.... if you do not like it then simple don't update bitcoin core!

[johnyj]

The trend has been, for users to adapt to the requirement of the applications running on their computers, not the other way around.

In my student days, I worked in a Computer shop... The sales people normally asked the same questions :

1. What applications will you be running? { Office or Gaming }

When they answer "Office use" they quoted less memory AND when the person was into "Gaming" they quoted more memory.

This should not be a determining factor, to base your decision on... If someone wants to host a node, they should increase the memory.   

If the most high end gaming never need more than 32 GB of memory, I'm afraid the motherboard will be designed so that they only support 32GB memory maximum. It is impossible for Intel to design a motherboard specifically suitable to run bitcoin nodes, and even if they does, the price will be as high as some enterprise level cloud servers

[QuestionAuthority]

The trend has been, for users to adapt to the requirement of the applications running on their computers, not the other way around.

In my student days, I worked in a Computer shop... The sales people normally asked the same questions :

1. What applications will you be running? { Office or Gaming }

When they answer "Office use" they quoted less memory AND when the person was into "Gaming" they quoted more memory.

This should not be a determining factor, to base your decision on... If someone wants to host a node, they should increase the memory.   

If the most high end gaming never need more than 32 GB of memory, I'm afraid the motherboard will be designed so that they only support 32GB memory maximum. It is impossible for Intel to design a motherboard specifically suitable to run bitcoin nodes, and even if they does, the price will be as high as some enterprise level cloud servers

I thought operating full nodes and mining was pretty much expected to be taken over by big business? It's in the best interest of big business to make sure Bitcoin keeps chugging along. Why do we need every teenager with a gaming computer to mine Bitcoin in his basement or run a full node? All the users need are spv clients. We don't need every user to have control of the network. We should simply place our trust in big business like we always have. lol

[johnyj]

The trend has been, for users to adapt to the requirement of the applications running on their computers, not the other way around.

In my student days, I worked in a Computer shop... The sales people normally asked the same questions :

1. What applications will you be running? { Office or Gaming }

When they answer "Office use" they quoted less memory AND when the person was into "Gaming" they quoted more memory.

This should not be a determining factor, to base your decision on... If someone wants to host a node, they should increase the memory.   

If the most high end gaming never need more than 32 GB of memory, I'm afraid the motherboard will be designed so that they only support 32GB memory maximum. It is impossible for Intel to design a motherboard specifically suitable to run bitcoin nodes, and even if they does, the price will be as high as some enterprise level cloud servers

I thought operating full nodes and mining was pretty much expected to be taken over by big business? It's in the best interest of big business to make sure Bitcoin keeps chugging along. Why do we need every teenager with a gaming computer to mine Bitcoin in his basement or run a full node? All the users need are spv clients. We don't need every user to have control of the network. We should simply place our trust in big business like we always have. lol

That will require the amount of bitcoin business rise over 1000 at least, but now banks are closing their accounts

[QuestionAuthority]

The trend has been, for users to adapt to the requirement of the applications running on their computers, not the other way around.

In my student days, I worked in a Computer shop... The sales people normally asked the same questions :

1. What applications will you be running? { Office or Gaming }

When they answer "Office use" they quoted less memory AND when the person was into "Gaming" they quoted more memory.

This should not be a determining factor, to base your decision on... If someone wants to host a node, they should increase the memory.   

If the most high end gaming never need more than 32 GB of memory, I'm afraid the motherboard will be designed so that they only support 32GB memory maximum. It is impossible for Intel to design a motherboard specifically suitable to run bitcoin nodes, and even if they does, the price will be as high as some enterprise level cloud servers

I thought operating full nodes and mining was pretty much expected to be taken over by big business? It's in the best interest of big business to make sure Bitcoin keeps chugging along. Why do we need every teenager with a gaming computer to mine Bitcoin in his basement or run a full node? All the users need are spv clients. We don't need every user to have control of the network. We should simply place our trust in big business like we always have. lol

That will require the amount of bitcoin business rise over 1000 at least, but now banks are closing their accounts

I guess we should leave the block size alone for now then.

[goosoodude]

I would be interested to know just how much of the growth in UTXO is from multiple outputs into the same addresses verses new unspent outputs into new addresses. If a large amount of the growth is from two (or more) unspent inputs contained in the same address then Bitcoin could be forked somehow so that all unspent inputs a certain address has are combined into one.

I would also be curious to know how much of the UTXO growth was from the likes of what most would consider to be blockchain spam related transactions. If a large part of it was from this then some initiative could be done to either address these spam transactions or to somehow provide incentives to spend/combine these spam sized inputs

[tl121]

At best the $4000 RAM cost projection is BS.  Most likely it's negative people spreading FUD.  There is no requirement for the UTXO to fit entirely in RAM.  So long as it can fit on SSD there will be no problem.  I note that Amazon is currently selling 1 TB SSD drives for $400.

These is no reason for  data bases to fit entirely in main memory. Indeed, this has rarely been possible throughout the history of computing.  Present day software is obscenely inefficient in resource consumption compared to software used three decades ago, something that makes sense considering present day economics. These tradeoffs can be re-evaluated should it become necesssary.

If bitcoin transaction bandwidth continues to increase there may be software performance problems and software may require retuning and bloated software may have to be replaced.   However, this has nothing to do with the block size question.  If block sizes are not increased then bitcoin will soon cease to be a useful system and all of it's "problems" will be moot.

[BTC_Superman]

We should think twice before take next step or leave block size for now.

[tokeweed]

For all those interested in proposed increase of the max block size, here's today's entry from Gavin's blog:

http://gavinandresen.ninja/utxo-uhoh

UTXO uh-oh…

I’m adding entries to my list of objections to raising the maximum block size as I run across them in discussions on reddit, in IRC, etc.

This is the technical objection that I’m most worried about:

More transactions means more memory for the UTXO database

I wasn’t worried about it yesterday, because I hadn’t looked at the trends. I am worried about it today.
...

--------
As a bonus:

M. Karpeless on the proposed changes:

https://twitter.com/MagicalTux/status/596622731711352832?s=09

So you think hard fork will not happen next year?

[tokeweed]

Let me change my question.  Is Gavin Andresen saying the hard fork might not happen?

I was really excited to see what would happen if it was on.  He's probably having cold feet?

[BillyBobZorton]

For all those interested in proposed increase of the max block size, here's today's entry from Gavin's blog:

http://gavinandresen.ninja/utxo-uhoh

UTXO uh-oh…

I’m adding entries to my list of objections to raising the maximum block size as I run across them in discussions on reddit, in IRC, etc.

This is the technical objection that I’m most worried about:

More transactions means more memory for the UTXO database

I wasn’t worried about it yesterday, because I hadn’t looked at the trends. I am worried about it today.
...

--------
As a bonus:

M. Karpeless on the proposed changes:

https://twitter.com/MagicalTux/status/596622731711352832?s=09

Holy shit, is Mark Karpeless still a thing? How's the guy tweeting like it's nothing? That guy scammed millions of BTC of poor people's wallets for god's sake, I would be hiding under a rock.

[tvbcof]

Transplant relevant to this thread:

...
but even way before then it was entirely clear that there was a schism developing btwn Gavin, who has been pushing for an even more radical progressive block size increase, and the rest of the Blockstream devs looking to continue capping at 1MB while at the same time pushing their version of SC's.  it was clear, to me at least, that he would have to pull rank at some point.  and that is not a bad thing; that is what leaders do.  he is the lead dev after all.
...

Couldn't happen to soon.  Hopefully Gavin (who I've suspected for some time as being not much more than Hearn's mouthpiece to the world if not worse) will siphon off a huge count of the lumpenbitcoiners who constitute the main problem in the ecosystem in my humble opinion.

I've sat tight on my stash for well over a year for economic reasons and would be happy to sit tight for another few years until it is widely understood what a smoking crater is at the end of Gavin's plans for Bitcoin.  In the interim I intend to capitalize handsomely.  Or try to.

In looking for Maxwell's stuff on reddit I ran across something supposedly by Gavin (though I have my doubts.)  For one thing, nobody who has any bitcoin in significant quantity should have only just become aware of UTXO issues, and it's beyond absurd to suggest that the 'principle scientist' would.  For two, UTXO has nearly zero to do with block size anyway so the lesson he supposedly drew from it ("That is a very good reason to oppose increasing the maximum block size.") is either hopelessly ignorant or a weasel move to get his bloatcoin system in place when a 'solution' to the 'problem' pops out of the ether.

To expand on my point about UTXO issues, a grand total of one user worldwide could create a problem with UTXO size if they put (a fairly technically trivial) amount of effort into doing so.  Conversely, the UTXO problem could be managed with much or perhaps most of the world's population as users if it were addressed at the system design level.  Satoshi seems to have made some feeble stabs at doing this (which is why there is even such a thing as UTXO) but a proper solution to the problem would pretty much require a dynamic global circulation involving periodic re-issues.  That is, nobody socking their stash into deep storage or if they do wish to do so they pay a tax of some sort upon retrieval.  That would make for a fine system for SOME use-cases actually.  I (and I presume Maxwell and company) would be happy to see and happy to use such a solution but I would like to see it as a sidechain.

Edit quote above:  call attention to a critical statement with formatting.

Edit:  also:

Latest news from gavin experimenting on max block size increases:

https://mobile.twitter.com/gavinandresen/status/557620205443350529

He's being doing a pretty good job imho, particulary if you take into account also the previous post where he tested a version of bitcoin core that could handle a 20MB max block size:

http://gavintech.blogspot.it/2015/01/looking-before-scaling-up-leap.html

A more appropriate title would in my opinion be 'Leap before anyone looks.'

I've always considered Gavin a fairly honorable dude, especially by the ecosystem standards, but one still has to watch the pea when reading what he writes.  Especially when he talks UTXO.

[goosoodude]

Let me change my question.  Is Gavin Andresen saying the hard fork might not happen?

I was really excited to see what would happen if it was on.  He's probably having cold feet?

No one person has control over as to if the network will actually be forked or not. In order to fork Bitcoin, someone first needs to propose how the protocol will be changed, then needs to sell it to the various stakeholders who makeup the network, push the change through the network and the network needs to accept the changes and the miners need to mine blocks on the blockchain that is utilizing the changed protocol, otherwise the changes will not be implemented

[ArticMine]

There is very little evidence to justify extrapolating an annual drop of 20% in memory prices over a period of ten years or longer. Here is a history of memory prices since the 1960's. http://www.jcmit.com/mem2014.htm. If one follows the long term trend this threat from the UTXO is simply a non issue.

Memory prices have followed periods of sharp drops alternated by periods of relatively low drops. This makes sense as new technologies reach the market.