5 ms·

I can't speak for anyone else, but for me personally I just don't like the approach of trying to solve the problem by throwing more disk space at it. It feels..

I can't speak for anyone else, but for me personally I just don't like the approach of trying to solve the problem by throwing more disk space at it. It feels... inelegant, and probably unsustainable long term. Imagine for a minute a future where Bitcoin goes mainstream and needs to process ~2k transactions per second. (A reasonable estimate of the world's current credit card transaction rate, and 3 orders of magnitude more than Bitcoin's current max capacity.) Would that _really_ be possible to sustain just by raising the block size? Bitcoin's block chain is already 150 GB after only a decade, and if we drastically increase its rate of growth like that (by 3 orders of magnitude) the hardware requirements for even storing the full chain are gonna get ridiculous pretty fast. That's not to say that I think bigger blocks couldn't work, but I think if Bitcoin does go that route it then it will need to find a way to trim old blocks out of the chain without compromising security in any way. Otherwise there's a real possibility we might end up with a 100 TB chain on our hands within the next 10 years or so. (Assuming Bitcoin succeeds the way we might want it to.)

lawlessone 9y ago
If we keep bitcoin block sizes as they are or reasonable we can roughly predict how big the whole chain would be 100 years from now(im being very optimistic here.)
- colordrops 9y ago
  And if we stuck to horses we could have predict the fastest speed people could travel across the country in 100 years.
dlubarov 9y ago
Lightning Network isn't really a satisfying solution in my opinion. It means that Bitcoin will just be a low-capacity settlement layer, and regular purchases will need to use PayPal-like middlemen to avoid hefty fees. There are a few on-chain scaling solutions. One is Vitalik's approach to sharding. Payments would be split into debits and credits, and a credit transaction would include a Merkle proof showing that a bunch of validators on a different shard approved the corresponding debit. Double spending would be possible if you can corrupt a single shard, but users would be randomly bounced between shards to make that difficult. Another solution, which I think is very practical, is to make some protocol changes so that scalable full nodes can be run by a cluster of servers. Running a full node would become expensive, but most users would be fine with SPV clients + fraud proofs. See https://ethresear.ch/t/scaling-via-full-node-clusters/1358 https://ethresear.ch/t/scaling-via-full-node-clusters/1358
- darawk 9y ago
  > It means that Bitcoin will just be a low-capacity settlement layer, and regular purchases will need to use PayPal-like middlemen to avoid hefty fees. You're correct, of course. But what's wrong with that? You still retain ultimate sovereignty over your funds, which is the most important property, I think. What do you think is lost by Bitcoin being mostly a settlement layer with the option to transact on-chain for a higher fee?
  - dlubarov 9y ago
    I think it's okay if users could settle reasonably often, but unless the main chain's capacity is greatly increased, that won't be the case. If 7 billion people used Bitcoin today, the current ~4 TPS capacity would only be enough for each person to perform one transaction every 55 years. If we adopted a 100mb block size, which is probably pushing the limit of what a normal server machine can handle, that would be just enough for each person to perform two on-chain transactions per year. Most people would want to settle at least once per month, to collect their paychecks and move the money into a payment channel with a well-connected middleman. That doesn't seem feasible without sharding or some other solution to drastically increase the main chain's capacity.
    - darawk 9y ago
      I think the idea is that due to the existence of LN, you would almost never need to settle on-chain. But because you theoretically could, you maintain all the benefits of genuine financial sovereignty without ever having to actually exercise it.
      - dlubarov 9y ago
        I haven't read most of the paper (59 pages!) so I might be missing something, but let's say I'm receiving Lightning payments from my employer and sending Lightning payments to my landlord. When my channel to my landlord runs out of funds, won't I need a couple on-chain transactions to withdraw funds from my employer channel and add them to my landlord channel? If I could somehow transfer my employer's IOUs to my landlord off-chain, I figure that break the security model because my landlord wouldn't know if I had double spent the IOUs. Let me know if I'm missing something though.
nosuchthing 9y ago
This is a really bizarre rational, disk space is insanely cheap and simply storing the blockchain does nothing to actually validate transactions or contribute to the network. 8000GB drives cost about $150 now. Storage is getting cheaper. Why not change the PoW hash algo to avoid costly ASICs and help decentralize the more important aspect of blockchain transaction verification?
- Ajedi32 9y ago
  > 8000GB drives cost about $150 now That's enough to handle the blockchain size increasing by one order of magnitude. I just talked about a scenario where it could conceivably increase by 3. Storage is indeed cheap, but not _that_ cheap. Unless you're envisioning a scenario where only large stakeholders like miners and exchanges ever have to store a full copy of the blockchain.
  - colordrops 9y ago
    Why would transactions increase but not space per dollar? Has disk technology stopped progressing? The time it takes to increase by 3 is probably the amount of time disk space will increase by 3 at the same cost.
    - __blockcipher__ 9y ago
      Exactly. Costs for computing/disk storage go down exponentially. Cryptocurrency tx/sec goes up exponentially. As long as the two rates roughly cancel out, we're okay. If they're massively different rates, then you're still dealing with an exponential which is difficult. Either way though, lightning is a linear benefit, not an exponential one. And it comes at the cost of breaking everything that made bitcoin great in the first place. I don't ever want to have to deal with an intermediary financial institution again. Somehow, the pro-BTC side of the chain thinks that view is ridiculous.
      - vokep 9y ago
        The "Pro-BTC" side is really Pro-neoBTC, oldBTC, satoshis BTC was all about never dealing with an intermediary unless you choose to for whatever weird reason.
colordrops 9y ago
Based on the exponential growth of disk size, I don't see this as a problem. I wouldn't be surprised if the entire chain, even at visa levels, could be stored on an average phone in 10 years.
- umanwizard 9y ago
  Exponential growth eventually slows down.
  - colordrops 9y ago
    Right, for both disk size and transaction rates. Your point?

lawlessone 9y ago

If we keep bitcoin block sizes as they are or reasonable we can roughly predict how big the whole chain would be 100 years from now(im being very optimistic here.)

colordrops 9y ago
And if we stuck to horses we could have predict the fastest speed people could travel across the country in 100 years.

dlubarov 9y ago

Lightning Network isn't really a satisfying solution in my opinion. It means that Bitcoin will just be a low-capacity settlement layer, and regular purchases will need to use PayPal-like middlemen to avoid hefty fees. There are a few on-chain scaling solutions. One is Vitalik's approach to sharding. Payments would be split into debits and credits, and a credit transaction would include a Merkle proof showing that a bunch of validators on a different shard approved the corresponding debit. Double spending would be possible if you can corrupt a single shard, but users would be randomly bounced between shards to make that difficult. Another solution, which I think is very practical, is to make some protocol changes so that scalable full nodes can be run by a cluster of servers. Running a full node would become expensive, but most users would be fine with SPV clients + fraud proofs. See https://ethresear.ch/t/scaling-via-full-node-clusters/1358 https://ethresear.ch/t/scaling-via-full-node-clusters/1358

darawk 9y ago
> It means that Bitcoin will just be a low-capacity settlement layer, and regular purchases will need to use PayPal-like middlemen to avoid hefty fees. You're correct, of course. But what's wrong with that? You still retain ultimate sovereignty over your funds, which is the most important property, I think. What do you think is lost by Bitcoin being mostly a settlement layer with the option to transact on-chain for a higher fee?
- dlubarov 9y ago
  I think it's okay if users could settle reasonably often, but unless the main chain's capacity is greatly increased, that won't be the case. If 7 billion people used Bitcoin today, the current ~4 TPS capacity would only be enough for each person to perform one transaction every 55 years. If we adopted a 100mb block size, which is probably pushing the limit of what a normal server machine can handle, that would be just enough for each person to perform two on-chain transactions per year. Most people would want to settle at least once per month, to collect their paychecks and move the money into a payment channel with a well-connected middleman. That doesn't seem feasible without sharding or some other solution to drastically increase the main chain's capacity.
  - darawk 9y ago
    I think the idea is that due to the existence of LN, you would almost never need to settle on-chain. But because you theoretically could, you maintain all the benefits of genuine financial sovereignty without ever having to actually exercise it.
    - dlubarov 9y ago
      I haven't read most of the paper (59 pages!) so I might be missing something, but let's say I'm receiving Lightning payments from my employer and sending Lightning payments to my landlord. When my channel to my landlord runs out of funds, won't I need a couple on-chain transactions to withdraw funds from my employer channel and add them to my landlord channel? If I could somehow transfer my employer's IOUs to my landlord off-chain, I figure that break the security model because my landlord wouldn't know if I had double spent the IOUs. Let me know if I'm missing something though.

nosuchthing 9y ago

This is a really bizarre rational, disk space is insanely cheap and simply storing the blockchain does nothing to actually validate transactions or contribute to the network. 8000GB drives cost about $150 now. Storage is getting cheaper. Why not change the PoW hash algo to avoid costly ASICs and help decentralize the more important aspect of blockchain transaction verification?

Ajedi32 9y ago
> 8000GB drives cost about $150 now That's enough to handle the blockchain size increasing by one order of magnitude. I just talked about a scenario where it could conceivably increase by 3. Storage is indeed cheap, but not _that_ cheap. Unless you're envisioning a scenario where only large stakeholders like miners and exchanges ever have to store a full copy of the blockchain.
- colordrops 9y ago
  Why would transactions increase but not space per dollar? Has disk technology stopped progressing? The time it takes to increase by 3 is probably the amount of time disk space will increase by 3 at the same cost.
  - __blockcipher__ 9y ago
    Exactly. Costs for computing/disk storage go down exponentially. Cryptocurrency tx/sec goes up exponentially. As long as the two rates roughly cancel out, we're okay. If they're massively different rates, then you're still dealing with an exponential which is difficult. Either way though, lightning is a linear benefit, not an exponential one. And it comes at the cost of breaking everything that made bitcoin great in the first place. I don't ever want to have to deal with an intermediary financial institution again. Somehow, the pro-BTC side of the chain thinks that view is ridiculous.
    - vokep 9y ago
      The "Pro-BTC" side is really Pro-neoBTC, oldBTC, satoshis BTC was all about never dealing with an intermediary unless you choose to for whatever weird reason.

colordrops 9y ago

Based on the exponential growth of disk size, I don't see this as a problem. I wouldn't be surprised if the entire chain, even at visa levels, could be stored on an average phone in 10 years.

umanwizard 9y ago
Exponential growth eventually slows down.
- colordrops 9y ago
  Right, for both disk size and transaction rates. Your point?