For posterity, here is Gregory Maxwell's blocksize increase proposal:


Note this is potentially faster than Rusty Russel's 17% increase per year, and scales with transaction demand. Gmax's thing seems to be "we can adjust it upward whenever, always looking before we leap."

Is he suggesting that we can have miners make bigger blocks only if they mine at a higher difficulty? That is an interesting idea I didn't know was possible.

Another bullish chart:



In some way this one has "excluded the bubbles" from the trendline:

The Lightning Network is way too speculative, as you've pointed out. Gmax doesn't believe in the traffic jam scenario, or if he does he thinks we can just push through a small upgrade because the consensus will then be there. That's why buying time doesn't mean much to him. Perhaps you should figure out why he thinks that.

To me it depends on whether a fee market is working properly by then. If it is, there won't be a traffic jam, just some higher fees. If they get too high, I think even Gmax and the others would agree to a mild increase.

The more I follow this blocksize debate, the less I'm worried. There's a lot of exaggeration and rote extrapolation assumptions on all sides, but humans adapt. We worry about bad PR, but really, who cares? It's not like there is any other system that's going to supersede Bitcoin. If you want to do any of the really major things Bitcoin enables, you don't have a choice but to use Bitcoin (any altcoin is just going to be even easier to mess up).

Not that I want or expect bad PR, but we can't let that concern infect our planning and experimentation. As far as investment and adoption, the timing of the next cap increase is a short-term issue. I'd like to raise the limit and see it get limited economically so the devs could get on to something more important, but I also don't think anything would actually break if we stayed at 1MB even during the next bubble. Bitcoin is not a currency yet. It's an investment. It won't need tons of transaction throughput for a long time, and since there is no real fee market now there's no way it's using the blockspace efficiently yet. So I'm not worried, though if I had my way I'd prefer the "no cap" experiment, starting today

It's good to stress the chain, but it should be ramped up gradually so as to give people time to deal with the changes for the best adaptation. You could say the sudden shock would be a better stress test, but this isn't so much a test as much as a way of incentivizing optimization. (Right?)

One other thing to keep in mind is that if this were actual actual transaction demand miners and others could figure into their future business models that they can rely on these fees and they may start adjusting their fees to try to maximize revenue (making this attack prohibitively expensive, as we want it to be). If it's just a temporary attack, even if it continues for weeks, miners aren't going to have any incentive to change things. Well, they may have a personal or altruistic incentive to raise their fees for the good of the network, but not really an economic one. And even if they raise their fees, they may not have an incentive to make it easier for users to know how much to pay.

Also, it should be noted that if this jams up the network it doesn't necessarily support either side of the debate. It's not going to force a blocksize increase if everyone is experiencing delays; hopefully it will force better resource usage through the price system, though.

Gavin's plan to break stagnation seems to have worked. Gregory Maxwell has proposed a blocksize increase plan (in a manner of speaking).

The first thing that jumps out is that over the past 3.5 years that stability indicator has never gone over 70 without being immediately followed by a monster rally, and it just now went over 80.

Hmm...

Fascinating chart posted in masterluc's thread. I trust I don't need to spell out the implications.

OK, good point. I thought maybe having a time cutoff where no new tx are added to the first mempool after 10 minutes would help, but I guess there's no way to know for sure. That's the whole point of a consensus network after all. Oh well, there goes that shower thought. Thanks for the quick reply.

You're saying miners currently sometimes only put the hash of all the tx in a block, instead of the tx themselves?

Perhaps some of the coders here can help me understand something.

Why not have an new "mempool" be created every 10 minutes, so that if it takes 30 minutes to find a block the winning miner will just take all the valid transactions in the first mempool, no matter how huge the total "blocksize" would be, and put only the hash of those transactions into the block? That way the block itself would be tiny so propagation wouldn't be an issue. All miners and other full nodes would have the first mempool transactions already(?), those being set in stone, so they would just have to check that the hash matches the set of all valid tx in the first mempool. Then the next winning miner would take all the valid tx in the second mempool, etc.

Of course if a miner finds the next block in less than 10 minutes and there is no mempool queued up yet, this doesn't work. Perhap difficulty would have to be adjusted to ensure miners were usually a little bit behind the curve.

This seems to shift the burden from bandwidth to CPU power for checking the hash, but as long as miners are behind the curve it seems to avoid the "race" where lower-bandwidth miners/nodes are at a disadvantage.

Does this, or anything like it, make any sense?

So something like a blockchain "The Price is Right" where miners try to bid on the biggest acceptable block without going over?

Thanks, that was the missing piece to the forkbitrage proposal. Either that or somehow change how the signing works whenever there's a hard fork(?).

Nice. A sections for endorsers would be cool, too.

tptb driven...asunder

Re: 21 and mining pools

The whole reason for mining pools is because miners care about variance. But consumers don't, since they don't even know/care the device is hashing. So have each device be a mining node* (not just a hashing machine) but set up so that, in the event that it wins the block reward lottery, the reward all goes to the manufacturer. Either by some automatic hardcoding or by a contractual agreement. (Alternatively, the reward could be split 75/25 or some other way, if consumers are interested in "playing the lottery," though I'm not sure that would be much fun compared to buying a lottery ticket, because the results take years to come in.)

To manufacturers that are big enough or can afford to wait out the variance, it won't matter much.

The only catch is that each machine doesn't have a replenishing supply of satoshis to use for IoT. Still, preload them with a bit or just send satoshis to them from the manufacturer on a monthly basis - the hard problem of them having a Bitcoin wallet to receive them is already solved anyway.

*Or if a node can't be fit on a phone, have some larger home appliance be the node, everything else just hashes.

Oooh, that article is a pretty big bombshell (everyone should read it, just because a lot of people will take it seriously). He brings up some interesting points about a shorting attack, but I don't think his argument properly accounts for how Bitcoin will grow as the threat grows. (A common mistake among Bitcoin critics.) See for example:



One of Gavin's proposals on how to mitigate the threat of a 51% attack: http://gavintech.blogspot.jp/2012/05/neutralizing-51-attack.html But this only mitigates the denial-of-service aspect, not the double-spending aspect that the Clearmatics article refers to. Still it seems like if it's for settlement you can just wait a very long time, like 60 confirmations, or 10 hours, or even something like 3 days as it takes now for securities, but save on all the expenses. What are the odds that a 51% goes undetected for that long? It would totally screw up the ledger even if it were detected and legally rolled back, sure, but the premise is that the motivation of the attacker is to double-spend the security, not just to watch Bitcoin burn.

To be honest, that's probably the way to go for true altcoins (alt-ledgers, not spinoffs). If you take my arguments posted over the past few days on altcoins seriously, the conclusion seems to be either abandon the investment before it cools off (gets spun off or sidechained by Bitcoin) or aim for the "Hail Mary pass" from outside virgin investors and users. Note that near term I think Monero will rise against the other alts; I'm talking strictly long term here.

Even I, Mr. Bitcoin Maximalist, would invest in an altcoin if it looked like it would somehow be successfully pushed to a huge audience that was separate and much bigger than Bitcoin's community,* assuming it had utility. That would be because, in a very real sense, it would not be an altcoin at all. To those people, to that new population, it would be exactly what Bitcoin was to us. Sui generis. "The WWL." However, it would be a tough trick at this stage to get them to forget/ignore Bitcoin enough to make this dynamic work. Ripple can kind of do it even though everyone knows Bitcoin now, because their product (giving them the benefit of the doubt that their product is even a thing) is different enough. Add in the need for a track record and the hurdle is even higher.

*Extreme example to make the point: you find an alien planet that won't interface with Earth for many more decades/centuries and launch it there. What purpose would there be in referring to it as an "altcoin" anymore? It's the WWL. It's THE blockchain.

I don't disagree, but I have assumed the usual context here is basically Bitcoin investors "branching out" into some altcoins. One could make a somewhat valid criticism that this context is changing to small degree, the best examples being Ripple and Ethereum and maybe Dogecoin, where they are actively trying to source users and investors from outside the Bitcoin community (BitShares to an extent as well). As long as Bitcoin is "the main attraction" into the space, which as far as I can tell it overwhelmingly is, the points I made seem to hold.

Right, all in context.

A good example is that in Japan today I learned Bitcoin is getting so little adoption (due to MtGox being all they know of it) that Ripple is actually ahead in terms of total amount invested (they used MLM/Amway to sell it ). When you have an entirely different population to choose from, nothing I said about precedents regarding SoV applies, since the new population is unaware of the precedent.

For practical purposes, though, I don't see something that is very much like Bitcoin being able to pull that kind of "new population group" trick, unless circumstances change in some unforeseen way, possibly regulatory.