@adam I enjoyed that Sangye, i'd say if you want to go through an do a round of revision (just to tighten it up), i'd be interested in including your reading of it on an upcoming episode.
Sure, i'll do that, my writing process always needs a lot of revisions. I'll work on it and send it to you in case it is relevant to what you're doing. Some of what i say is regurgitating other ideas i have picked up, but i am doing rounds of contemplation on this to see if i can get into other perspectives.
@adam I enjoyed that Sangye, i'd say if you want to go through an do a round of revision (just to tighten it up), i'd be interested in including your reading of it on an upcoming episode.
Sure, i'll do that, my writing process always needs a lot of revisions. I'll work on it and send it to you in case it is relevant to what you're doing. Some of what i say is regurgitating other ideas i have picked up, but i am doing rounds of contemplation on this to see if i can get into other perspectives.
I don't think you need to change it that much.
It's only regurgitation if what comes up is worse than what went down.
@dlight sorry man I pretty much totally disagree with you here
I'm extremely surprised that LTB would be supporting this obvious attempt at a "coup d'etat" against the Bitcoin system. To put control over the system in the hands of one guy is nuts.
That's not really what is happening here. There are already several implementations of bitcoin which are run by different teams of developers (and perhaps some lone wolfs). These are the systems that run the back end of everything you can do in the bitcoin world. Such as BTCD, libbitcoin, Bits of Proof, BitcoinJ, picocoin, BitCore, Electrum Server, Ufasoft Coin, NBitcoin, caesure, Toshi, and of course Bitcoin Core, BitcoinXT and other variants of Bitcoin Core. Although IIRC only Bitcoin Core (the "reference client") and it's variants have 100% full mining functionality, but that is a separate issue already being worked on (via consensus libraries anybody can use)
Also IIRC Gavin actually has full administrative access to the Bitcoin Core repository and could just kick out all the other people with commit access if he wanted to, so really that project has been "controlled" by one guy the entire time. Correct me if I'm wrong, he might have handed the reigns over to another dev (forget the name at the moment)... but either way that's just how projects on GIthub work. One person is the owner of the project, and they can accept or reject changes from people or allow others to accept changes on their behalf (commit access). That doesn't mean though that they can magically force everyone to download and use any code they feel like. And because it's open source, any attempt at doing something fishy quickly becomes obvious.
@Cryptonaut's statement "For the record, at LTB we will be switching all our nodes over to Bitcoin XT to ensure we stay on the proper chain when the fork is triggered early next year."
is clearly stating that the LTB network is taking sides - the opposite of neutrality.
Yeah, I knew when I wrote that that I would get some shit for it. I reworded it a little to clarify a bit more. We actually haven't switched our nodes to XT yet because we need to get a patch called "addrindex" adapted to it, which is a requirement for using Counterparty (also I'v been travelling a fair amount this month). Also as @adam has mentioned, it's less about XT specifically and more about making sure we are always on the chain with majority hashrate and if a fork occurs we will receive little or no interruptions. Because we are not miners, our choice of backend client doesn't really matter to the network.
The whole point of a distributed system like Bitcoin is that it should be a set of rules that are extremely hard for anyone to alter.
I don't really see how this whole process so far has been "easy", and it's far from over. Don't forget that this debate has actually been going on for years, and look at how intense it has gotten in the past few months with still no final resolution. You can write entire books out of the arguments back and forth that have been made...
The idea that one guy (using the Gavin's name recognition) can fork the network is outrageous and pretty much defeats the whole purpose of crypto-currency.
Yes Gavin is well known and well respected (for most people anyways) in this space, and yes him and Hearn were the ones to pull the fork-the-project card.. but they still have to convince everyone else to implement the same protocol changes and/or use the BitcoinXT client instead. Nobody is in favour of XT or BIP101 because Gavin is Gavin or Mike is Mike... it's because the numerous arguments they have laid out just so happen to make sense. I couldn't care less if it was Gavin Andresen or an anonymous user named CheeseBurg3rs69. Plus, the Bitcoin Core project could just patch over the BIP101 implementation at any time and effectively stop the exodus to the BitcoinXT project in it's tracks.
While it probably doesn't hurt anything to increase the block-size to 8MB - the further increases up to 8GB are extremely questionable - opening the network to attack in the future.
8GB won't be for 20 years. In 1995, a 10GB harddrive cost something like $2000, 64MB of ram for $150 and internet access was available at a speedy 56Kb/s over dial-up. Today, you can get 1,000GB harddrives for like $30, 4,000MB sticks of ram for $25 and some of the top internet speeds available in the world clock up to 1,000Mb/s for both uploads and downloads. That means already with todays fastest connections, you would be able to download full 8GB blocks in around 8 seconds. Thats 1,000 times larger than what the initial 8MB limit will be over the next couple years. Computer technology may or may not improve at the same rate that it has been over the next 20 years, but it's hard to believe that 8GB will be anything to worry about by that time. By then, we will be measuring consumer disk space in petabytes or maybe even exabytes
Plus there are numerous other changes to the XT code which are not even being discussed. Why should all of Mike Hearn's pet projects suddenly become adopted with no discussion? Not to mention all of his future pet projects.
Well for one XT is just one project, so I don't see how people using XT all of a sudden means mass adoption for his other projects like Lighthouse and BitcoinJ. But sure, let's talk briefly about the additional changes in BitcoinXT (there aren't many)
You can see a list of patches over on the website here: https://bitcoinxt.software/patches.html
So as many have said - increasing the blocksize is just a stopgap measure until far more scalable solutions - like payment channels are introduced. Solutions which could handle the many orders of magnitude increases that will be required for Bitcoin to become widely adopted globally.
Yes and no. It's a step along the way and definitely not the final solution, but we also can't sit around and wait for someone to find the "perfect answer" to scalability, when we already know how to easily bump up the current network a few notches. Payment channels also have their own problems, such as the fact that they require funds to be locked up for periods of time (if you can only open and close one channel per year.. well I sure hope you have a whole years worth of income handy), are semi-centralized, take fees away from miners (the ones securing the network), and of course they don't even really exist yet anyways, it's all just on paper.
We need to remember that these increases tend to happen all at once - not gradually. The network needs to be able to very rapidly explode in size without choking. Just altering the blocksize doesn't achieve this.
Sure. BIP101 with 8MB doubling every 2 years gives us some solid predictability of the network limits so that we can plan accordingly.
But allowing a coup just to make this temporary fix - is crazy. As the person claiming to be Satoshi said - it will give Bitcoin the appearance of a failed experiment - since it will give dictatorial control to a single person - one who has vowed to do whatever he wants regardless of what the community thinks.
As explained above, no single person actually has control here.. there are no dictators in Bitcoin. Multiple competing implementations are healthy for the network and allow progress to be made instead of everyone getting caught up in a deadlock because no one wants to make a decision or someone is being too stubborn. Here is an interesting article that came out today from Meni Rosenfeld about forks.
In related news, these two things happened today:
http://blog.blockchain.com/wp-content/uploads/2015/08/Industry-Block-Size-letter-All-Signed.pdf
https://www.reddit.com/r/Bitcoin/comments/3i9qdj/25_of_hashing_power_is_now_publicly_backing_bip100/
to quickly clarify on that second link - there is no actual code made for BIP100 (Jeff Garzik's proposal), so those miners are basically just saying "Hey we like this idea better" - but via adding some text to the blocks they find instead of making a public statement on their blog or reddit or twitter or w/e
8GB won't be for 20 years. In 1995, a 10GB harddrive cost something like $2000, 64MB of ram for $150 and internet access was available at a speedy 56Kb/s over dial-up. Today, you can get 1,000GB harddrives for like $30, 4,000MB sticks of ram for $25 and some of the top internet speeds available in the world clock up to 1,000Mb/s for both uploads and downloads. That means already with todays fastest connections, you would be able to download full 8GB blocks in around 8 seconds. Thats 1,000 times larger than what the initial 8MB limit will be over the next couple years. Computer technology may or may not improve at the same rate that it has been over the next 20 years, but it's hard to believe that 8GB will be anything to worry about by that time. By then, we will be measuring consumer disk space in petabytes or maybe even exabytes
I disagree with this optimistic projection.
I once worked with a guy who was into over-clocking 286 CPU (back then big fans weren't attached to CPUs) and i would fight with him and say ... hey, sure you can cool it and improve performance but eventually the laws of physics limit its speed. He didn't agree, but the laws of physics did limit the speed and it just wouldn't go faster beyond a certain point not determined by cooling. I also remember using a 20MB hard disk that took an eternity to copy files around - which, yes, does happen in a relatively small time now. So a whole hard disk to move fast - but i am still not going to agree with your point. That was back in 1988 and its 27 or so years since then to now. The curve doesn't just go up exponentially forever, you do hit the laws of physics. Computer power has hit a wall - and moved into parallelism. Network speed is hitting a similar wall, but it hasn't gone into parallelism. Part of what makes a lot of this work is that miniaturization had to continue on and on for a while but they are getting to the atomic structure limitations and frequency limitations of "single core" processing in network speed.
Lightweight software is a marvel that is still appreciated today - p2p software like uTorrent - its tiny, its efficient. Sure it can deal with absolutely massive torrents, but the torrents don't just keep getting bigger - they haven't really been getting much bigger at all. That is network transfer - its more adopted, widespread and with better algorithms of video compression. So files that are smaller for the same video are actually working - h.265 - and that is because we don't want to unnecessarily burden. What about the equivalent to an h.265 of blockchain compression! I want people who will make a consortium there because transaction storage will be like video storage - heavy, annoying and clumsy. To keep it running well and tight ... you do whatever you can, or it does break.
An algorithmic advance will need to happen or Bitcoin will just become a lesser coin when the "portable" blockchains, Where a node can get online relatively quickly and sync, beat out bloated blockchains. The projection of growth is not a physical law. Moore revised his doubling from every 1 year to every 2 years and only said it would last for this decade at one speed and that decade at another. So ... take it from the one everyone bases these assumptions on https://en.wikipedia.org/wiki/Moore%27s_law There are also economic limitations - and the worlds bubbles in currency, markets is quite related to interest in bitcoin. Bitcoin might take off when the next bust cycle hits. In that type of economic depression - the growth of these capacity projections (doubling of storage, data transfer) will also be retarded. Its interesting how people take to learning about Bitcoin.
So I think BIP101 is somewhat infantile because what makes it work is not just storage and network speed but data handling THAT is a big part of the other two. So i don't think we could possibly be able to deal with 8gb blocks in the next 20 years at all. Take it from someone who was there 26 years ago. 20mb blocks today are still considered very large, such that they dropped back to 8mb for the next proposed max block size limit in BitcoinXT. The doubling is more than i would buy into - i want to see a moderate increase and i want to support better algorithms.
BIP101 is more based on a lazy "i don't want to keep having this fight" approach to blocksize growth. Assuming that a simplistic projection will hold true and doubling is fair. I am not against your freedom to run BitcoinXT - but i am against your refutation of the arguments presented. So i offer this rebuttle - and i am swinging away from BitcoinXT but i also am more of a keen user/observer of the bitcoin organism so what lessons will be learned will be learned. I would say that if we got to BitcoinXT we will be quite predictably repealing some of those changes quite soon after and that is an indication they are rushed.
Just for completeness - the most epic, high-speed internet is not the "typically available" internet speed you will see in 20 years. It is a mere economic reality that you can build such high speeds ... today a very good internet speed will still often take a long time to get an 8gb file. In fact most disk systems, even SSD, won't play it out at in 8 seconds. When files transfer over continents and "propagate" they will go according to intercontinental bottlenecks in communication. Say we did have that speed pervasively - and we used it for blockchain - then blockchain would be the enemy of bandwidth managers. Guess what is the enemy of bandwidth managers today - bittorrent - yes. Getting your latest TV shows is a big hassle for the internet speed, they are as valuable to some as keeping their blockchain updated if they have a node. So lightweight design is massively advantageous.
8GB won't be for 20 years. In 1995, a 10GB harddrive cost something like $2000, 64MB of ram for $150 and internet access was available at a speedy 56Kb/s over dial-up. Today, you can get 1,000GB harddrives for like $30, 4,000MB sticks of ram for $25 and some of the top internet speeds available in the world clock up to 1,000Mb/s for both uploads and downloads. That means already with todays fastest connections, you would be able to download full 8GB blocks in around 8 seconds. Thats 1,000 times larger than what the initial 8MB limit will be over the next couple years. Computer technology may or may not improve at the same rate that it has been over the next 20 years, but it's hard to believe that 8GB will be anything to worry about by that time. By then, we will be measuring consumer disk space in petabytes or maybe even exabytes
I disagree with this optimistic projection.
I once worked with a guy who was into over-clocking 286 CPU (back then big fans weren't attached to CPUs) and i would fight with him and say ... hey, sure you can cool it and improve performance but eventually the laws of physics limit its speed. He didn't agree, but the laws of physics did limit the speed and it just wouldn't go faster beyond a certain point not determined by cooling. I also remember using a 20MB hard disk that took an eternity to copy files around - which, yes, does happen in a relatively small time now. So a whole hard disk to move fast - but i am still not going to agree with your point. That was back in 1988 and its 27 or so years since then to now. The curve doesn't just go up exponentially forever, you do hit the laws of physics. Computer power has hit a wall - and moved into parallelism. Network speed is hitting a similar wall, but it hasn't gone into parallelism. Part of what makes a lot of this work is that miniaturization had to continue on and on for a while but they are getting to the atomic structure limitations and frequency limitations of "single core" processing in network speed.
I'm not trying to say that computing and network speeds are going to continue to go up exponentially forever, because they obviously can't... however even just looking at advances made in the past few years, it is extremely hard for me to believe that today we are really at a total stopping point and that 20 years from now we will only have marginal improvements.
Lightweight software is a marvel that is still appreciated today - p2p software like uTorrent - its tiny, its efficient. Sure it can deal with absolutely massive torrents, but the torrents don't just keep getting bigger - they haven't really been getting much bigger at all. That is network transfer - its more adopted, widespread and with better algorithms of video compression. So files that are smaller for the same video are actually working - h.265 - and that is because we don't want to unnecessarily burden. What about the equivalent to an h.265 of blockchain compression! I want people who will make a consortium there because transaction storage will be like video storage - heavy, annoying and clumsy. To keep it running well and tight ... you do whatever you can, or it does break.
uTorrent is actually a notoriously bloated piece of software (which even had recently snuck in bitcoin mining software within it, of all things...), so not sure where you get the idea it is lightweight. As far as torrents not getting bigger.. Grand Theft Auto 5 clocking in at a 60GB download would like a word with you :P
I'm all for finding better ways to compress and store transaction data though. That is yet another step beyond just raising capacity limits.
An algorithmic advance will need to happen or Bitcoin will just become a lesser coin when the "portable" blockchains, Where a node can get online relatively quickly and sync, beat out bloated blockchains. The projection of growth is not a physical law. Moore revised his doubling from every 1 year to every 2 years and only said it would last for this decade at one speed and that decade at another. So ... take it from the one everyone bases these assumptions on https://en.wikipedia.org/wiki/Moore%27s_law
I'm not too sure what your point is here. Simply having a more "lightweight" and "portable" blockchain is not much of an incentive for other coins to prosper, plus if disk space is a worry, we already have blockchain pruning available. The vast majority of bitcoiners also do not run their own nodes at all - it is already possible to jump into bitcoin and start using it within minutes, thanks to wallets such as Mycelium, Electrum etc... also, blockchains by design are a forever growing database. If you want to host your own copy of the blockchain, you defacto agree to hang on to these files which will never stop growing (unless transaction volume goes to 0). Hence the idea of a "bloated blockchain" doesn't make too much sense to me.
There are also economic limitations - and the worlds bubbles in currency, markets is quite related to interest in bitcoin. Bitcoin might take off when the next bust cycle hits. In that type of economic depression - the growth of these capacity projections (doubling of storage, data transfer) will also be retarded. Its interesting how people take to learning about Bitcoin.
Maybe, maybe not. We have yet to see anything like this actually happen, and also I don't believe global markets are as closely related to the bitcoin markets as you might think, although certainly the condition of the world economy is one of the factors that gets some people interested in the first place.
So I think BIP101 is somewhat infantile because what makes it work is not just storage and network speed but data handling THAT is a big part of the other two. So i don't think we could possibly be able to deal with 8gb blocks in the next 20 years at all. Take it from someone who was there 26 years ago. 20mb blocks today are still considered very large, such that they dropped back to 8mb for the next proposed max block size limit in BitcoinXT.
storage, network speed and processing are always a factor no matter what you are talking about when it comes to doing things on the internet. I also still fail to see how 8GB blocks could possibly be that big of a deal 20 years down the road, when today we would be able to handle them in some areas of the world (those with the 1Gb/s bandwidth speeds) and on high end systems. Surely the high end systems of today will at least be the mid-level systems of tomorrow.
The doubling is more than i would buy into - i want to see a moderate increase and i want to support better algorithms.
This argument is confusing because a few paragraphs up, you say that doubling every 2 years is probably not nearly enough, but then you turn around and say that we should probably be even more modest. Do we increase capacity by 0.5X, 2X, 4X, 100X...? Or do we just keep the capacity limits at their current levels forever while we figure out every possible way to avoid putting transactions on-chain (which IMO defeats the purpose of bitcoin)?
BIP101 is more based on a lazy "i don't want to keep having this fight" approach to blocksize growth. Assuming that a simplistic projection will hold true and doubling is fair. I am not against your freedom to run BitcoinXT - but i am against your refutation of the arguments presented. So i offer this rebuttle - and i am swinging away from BitcoinXT but i also am more of a keen user/observer of the bitcoin organism so what lessons will be learned will be learned. I would say that if we got to BitcoinXT we will be quite predictably repealing some of those changes quite soon after and that is an indication they are rushed.
I don't get that impression at all. The block size limit was always meant to be a temporary limit, and the discussion around changing it has been going on for years. It should never have even been a fight, and I think it's reasonable to want to have the issue settled once and for all so we can move on to the next major scalability concern. You also have to remember that it's just a maximum limit. Miners are free to mine blocks at any size smaller than the limit (down to less than 1KB) if that is what makes most sense to them, so there isn't really any reason to back-track on the protocol changes when individual miners can just reduce the limit on their own.
Just for completeness - the most epic, high-speed internet is not the "typically available" internet speed you will see in 20 years. It is a mere economic reality that you can build such high speeds ... today a very good internet speed will still often take a long time to get an 8gb file. In fact most disk systems, even SSD, won't play it out at in 8 seconds. When files transfer over continents and "propagate" they will go according to intercontinental bottlenecks in communication. Say we did have that speed pervasively - and we used it for blockchain - then blockchain would be the enemy of bandwidth managers. Guess what is the enemy of bandwidth managers today - bittorrent - yes. Getting your latest TV shows is a big hassle for the internet speed, they are as valuable to some as keeping their blockchain updated if they have a node. So lightweight design is massively advantageous.
For one, internet speeds have largely been restricted not by strictly technology, but by various monopolistic businesses and their unwillingness to upgrade their infrastructure (and in some cases purposely throttling it). I don't think it is so far fetched of an idea that over the next few decades we can go well beyond a measly 1GB/s , and that a good portion of the world will have access to it. Additionally, similar to the "not every coffee purchase should be on the blockchain" argument, not everybody and their mother + their dog needs to be running a full node which downloads, stores and validates potentially decades worth of financial transaction history for the entire planet. We are already at a point where essentially the only people that host the blockchain are businesses and developers, and a handful of hobbyists. Hosting the blockchain + the bandwidth usage that comes with it is just a cost of business for working with cryptocurrencies. If that cost is not worth it for your business, then you either outsource it to somebody else (e.g using blockchain.info instead of your own copy of the blockchain), or maybe crypto just doesn't make sense for you.
Great debate reply - of course i was trying to show that we can't assume some of what you seemed to assume. 8gb blocks in 20 years time - i don't expect it will be impossible but I expect it will be seriously problematic with storage, processing, sync - even comparing the improvements over the last 27 years - which have been massive. We were very primitive at the start - i doubt well look quite that primitive 20 years into the future.
I wasn't trying to say doubling isn't enough keep up - quite the opposite.I was trying to say that i thought it was likely to outstrip realistic performance. I didn't put that well if you got that meaning. I might have sounded like i was contradicting myself because you misunderstood. I don't think BIP101 is a good idea - to assume there is a natural doubling and it will be manageable is too risky. Hashing power is adapted much more intelligently - in a way that doesn't assume how much there is - but looks at what is actually happening. I already find the propagation of the blockchain a pain and worry very much about the way this storage will scale up. I actually want the first scale up but not to assume we keep doing it. Its good that we can have a lot of full nodes - I do absolutely get that you can read the blockchain without storing it for use in wallet software. If you compare again to torrents, more "seeders" help distribution - and that is why its good to incentivise people running full nodes. I also have a balanced way of looking at this too - we don't want people running full nodes on totally useless connections that can't propagate the information very well - in the same way people who seed torrents try to ask that people with best connections help share.
uTorrent used to be an extremely tiny download itself - that is how it got its name. I agree it got bloated but it is the lightweight design i was trying to talk about - that it was written to be small. 60gb downloads aren't that unusual but they are on the extreme end - i am talking averages. Torrents tend to not be growing in size for video distribution that much, the options for smaller sized versions are always there. Generally i meant that even 100gb (collections) or larger torrents haven't become so much more common - its been very much the same for a very long time with the torrent scene. Its more that people have caught up and the user base spread.To me that shows that while broadband speeds have improved, its certainly not improving at an exponential rate.
@sangye fair enough.
uTorrent used to be an extremely tiny download itself
Yeah totally, that was the reason I originally used it as well. Shame it has gotten crappy lately though
Anyways, seems like every proposal has various trade offs.
We could have a dynamic block size based on the average transaction volume over the last say 10,000 blocks, but then it is both unpredictable and could potentially be gamed by producing a lot of spam transactions over a period of time.
We could have a block size which is voted on every 3 months by the miners (i.e, BIP100), but then miners might have a bit too much control over the system, it's also unpredictable, and could potentially be manipulated by a minority of the hashrate (see here
We could have a block size which increases on a predictable fixed schedule for a limited amount of time (i.e, BIP101 with 8MB doubling every 2 years for 20 years), there is a chance that either a) that won't be nearly enough or b) it will be way too much and cause unintended consequences.
Personally, the predictable option seems like the safest route to me.
Personally, the predictable option seems like the safest route to me.
This was what i was after - not to win the point but to start going through the variations and consider more deeply if BIP101 is flawed or could be altered.
I think perhaps a doubling for 4 years might give the other developers time to respond. I am willing to say that i am not expert enough, that is why i am trying to check out various assumptions. The problem in science - including computer science is that people get used to creating models that are way to over-simplified because they work to a degree and a formula can be found that can alter things for the better. Then the side-effects come and people start saying "I don't understand, we worked it out" because they have been living with a model full of assumptions and excluding the data that disagrees for the sake of expediency. You see that a lot in electronic engineering too - the models aren't even based on real physics to some degree but you can solve electrical equations that are based more on the complexity of the circuits than the underlying materials. As long as it functions its ok.
We are in an economic realm now and economics is very bad to over-simplify. People lose a lot of jobs and wealth in that process - and gaming the system as you put it is one reason why economic models that start improving a situation and informing good decisions also start to fail. Not many things we hold to be true are, they are just appearing temporarily due to many factors. They are short term true and long term they vanish.
Here is a thread on reddit I just made which digs a little bit more into the BIP100 proposal by Jeff Garzik https://www.reddit.com/r/bitcoinxt/comments/3ii29n/potential_dangers_of_bip100/
Please Login to post a reply to this thread.
