Bitstein

Pierre Rochard and Gavin Andresen Discuss the Block Size Limit

Michael Goldstein -

As Bitcoin becomes more popular, the 1 MB block size limit might soon have actual economic consequences on the network. However, there is a debate as to whether those consequences are desirable or not. Today on the #bitcoin-assets IRC channel, the nexus of Mircea Popescu and La Serennisima, Gavin Andresen stopped by to engage the community on this issue, and from this a very interesting discussion emerged between him and Pierre Rochard, my friend and colleague from the Satoshi Nakamoto Institute. The following is an edited version of the logs that focuses on Gavin and Pierre’s discussion (full version here).

17:59:50 asciilifeform: !up gavinandresen
18:00:01 gavinandresen: howdy y’all
18:02:11 gavinandresen: I came here so you can tell me how I’m going to ruin bitcoin (again…)
[…]
18:23:26 Pierre_Rochard: my main disagreement is on the economics side, you say “ Limit the number of transactions that can happen on the Bitcoin blockchain, and instead of paying higher fees people will perform their transactions somewhere else.”
18:24:06 Pierre_Rochard: My response to that is two-fold: if maximizing revenue to miners is our goal, then shifting marginal demand to substitutes is fine.
18:25:02 Pierre_Rochard: Second, there is distance between the total transaction costs of Bitcoin vs the substitutes. So there’s inelasticity in demand
18:26:43 Pierre_Rochard: So here’s my hypothesis: if we allow the network to hit the block size limit, then we’ll see the transaction fee revenue growth accelerate, up until the point that substitution begins happening in earnest, then the fee revenue growth will decelerate or stall.
18:27:43 Pierre_Rochard: that would indicate that demand is inelastic, and it would tell us where the upper bound on bitcoin tx fees roughly is
18:27:54 Pierre_Rochard: all else equal
18:29:22 gavinandresen: Pierre_Rochard: I think you’re confusing the limit miners self-impose with the hard-coded upper limit
18:29:28 Pierre_Rochard: Now this part may be controversial for some members of b-a, but it’s at the point where fee revenue growth decelerates that the block size should be increased marginally, if the goal is to maximize fee revenue
18:29:36 davout: gavinandresen: why would miners impose a self limit?
18:29:51 gavinandresen: davout: today, because bigger blocks take a while to propagate.
18:29:54 Pierre_Rochard: gavinandresen: I think the limit that miners self-impose would not hold. I think we see that today
18:35:01 gavinandresen: Pierre_Rochard: if Satoshi hadn’t slapped on a 1MB blocksize limit, would you be lobbying for a hardfork now to impose one?
18:35:10 gavinandresen: … execuse me, soft fork….
18:35:15 Pierre_Rochard: gavinandresen: yes, I would
18:35:37 gavinandresen: Pierre_Rochard: ok, exactly what would you propose?
18:36:56 Pierre_Rochard: gavinandresen: that we don’t increase the limit until we see what happens to total fee revenue growth after a few months of full blocks
18:37:46 gavinandresen: Pierre_Rochard: I think we’re there: http://hashingit.com/analysis/39-the-myth-of-the-megabyte-bitcoin-block
18:38:38 gavinandresen: Pierre_Rochard: the 0.10 release’s wallet code includes floating fees, so over the next couple months we should get a much better idea of what is happening fee-wise.
18:38:45 Pierre_Rochard: then let’s stay there for six months to collect the data
18:38:48 Pierre_Rochard: ^ great
18:39:16 gavinandresen: Pierre_Rochard: what information will we get that will influence how large to make blocks?
18:40:14 gavinandresen: Pierre_Rochard: … I misstated: that will influence the maximum possible block size….
18:41:16 Pierre_Rochard: gavinandresen: fee revenue growth, if it accelerates then demand for btc transactions is relatively inelastic, the point at which it declerates indicates where substitution starts happening. If it’s right away, then you’re right on the economics. If its after a period of faster growth, then we can see what bitcoin transaction fee the market will bear before switching to substitutes
18:42:56 gavinandresen: Pierre_Rochard: but why would we want to hit the “then substitutes start happening” when we’re in Bitcoin’s infancy?
18:43:16 gavinandresen: Pierre_Rochard: Seems to me it is better to do everything we can to encourage widespread adoption right now.
18:43:32 Pierre_Rochard: gavinandresen: because the substitution would just be happening at the margin
18:44:22 gavinandresen: Pierre_Rochard: But we’d get probably at least six months, maybe a year or two of substitution because it takes time to roll out a hard forking change
18:44:40 Pierre_Rochard: gavinandresen: I think Bitcoin’s overall value proposition is so overwhelming that what’s hampering Bitcoin adoption is not tx fees, it’s just the Lindy effect of it being around long enough
18:45:22 gavinandresen: Pierre_Rochard: okey dokey. What if you’re wrong?
18:46:19 Pierre_Rochard: gavinandresen: then we see an acceleration of adoption of an altcoin / altcoins in general and react accordingly
18:46:28 gavinandresen: Pierre_Rochard: But you just said you want to let fees go up high enough so, at the margin, some people ARE turned away.
18:47:06 Pierre_Rochard: gavinandresen: that’s right, that’s the signal to increase the block size limit
18:47:13 gavinandresen: Pierre_Rochard: So, lets say we do see fees rise. How far do we let them rise? Who decides?
18:48:37 Pierre_Rochard: gavinandresen: at some point they stop rising because they’re too high for the marginal transactor
18:49:03 Pierre_Rochard: gavinandresen: in theory the miners would decide, in practice the core devs
18:50:12 gavinandresen: Pierre_Rochard: unfortunately, the experience for the marginal transactor is terrible: their transactions just never, ever confirm. Their coins get tied up…
18:50:53 gavinandresen: Pierre_Rochard: if wallets could deal with that I’d be more open to running the experiment, although I still think it is a terrible idea to shut out ANY reasonable use cases at this early stage of Bitocn’s life
18:52:10 gavinandresen: Pierre_Rochard: You started with a premise that I reject, by the way: I do not believe that a goal should be to maximize miner revenue
18:52:34 gavinandresen: Pierre_Rochard: I believe the goal should be to maximize the value of Bitcoin for everybody
18:52:34 Pierre_Rochard: gavinandresen: I think that our disagreement on that premise precludes agreeing on anything downstream of that
18:53:01 gavinandresen: Pierre_Rochard: you’re probably right. Are you a miner yourself?
18:54:05 Pierre_Rochard: gavinandresen: I am not. I just see a “too low” long term hash rate as the greatest risk of ruin Bitcoin faces, and it ought to be minimized before all other considerations
18:56:42 gavinandresen: Pierre_Rochard: what do you think of my argument that hash rate and fees are apples and oranges? That people will substitute away from fee-paying transactions to other solutions that use the block chain, which means trying to maximize fees means no guarantee that there will be enough hash rate to secure the chain?
19:00:46 Pierre_Rochard: gavinandresen: If I understand your argument correctly, you’re saying that the elasticity of demand is so great that fee maximization will be insufficient anyway, so try finding another solution now. That’s a pretty good argument, I think we should see what happens to fee revenue growth to validate it. If, say, the average fee goes up to 0.0004 btc and doesn’t budge from there, but anecdotally we hear that
19:00:47 Pierre_Rochard: off-blockchain transactions are taking off, then your argument will have won the day
19:02:08 Pierre_Rochard: gavinandresen: but my intuition tells me such substitution won’t happen at such a low fee
19:02:55 gavinandresen: Pierre_Rochard: ok. I’d like to brainstorm more about how you would set the maximum block size— I don’t want the developers setting it every two months, but I dont’ see a way to make fee revenue per block drive it (because the real-world bitcoin exchange rate is so variable)
19:07:11 Pierre_Rochard: gavinandresen: the only way I see is to regularly test at what tx fee the substitution begins happening, and increase it at the margin (say 20%) whenever the top percentile of fees starts hitting it. Yes, that would involve perhaps semi-annual block size limit increases and an element of judgement. I still see it as a better solution than a much-too-high limit or a contrived algorithm
19:07:40 Pierre_Rochard: increase it at the margin (say 20%) < increase the block size limit
19:09:24 gavinandresen: Pierre_Rochard: can you define “much too high limit” ?
19:10:20 Pierre_Rochard: gavinandresen: one where there is no competition among transactors to get into a block
19:11:06 Pierre_Rochard: that is, competition that would at least attempt to maximize transaction fee revenue
19:11:30 gavinandresen: Pierre_Rochard: fee revenue measured in real prices, yes?
19:12:21 gavinandresen: Pierre_Rochard: … because driving up real Bitcoin prices is why I think we should do everything possible to encourage widespread adoption
19:12:23 Pierre_Rochard: gavinandresen: that’s actually a very interesting question because we currently live in a world where miner liabilities are in fiat prices, but in the future that may not be the case.
19:13:49 Pierre_Rochard: gavinandresen: right, that goes back to bitcoin’s adoption relative to other altcoins. when we see a divergence then we know there’s substitution going on
19:14:43 gavinandresen: Pierre_Rochard: I think if you went to a VC with a business plan of “We’re going to raise prices until we start losing customers” the VC would tell you that is a huge mistake if you’re a high-growth thing-a-ma-bob
19:15:58 Pierre_Rochard: gavinandresen: I agree, but in this case I think Bitcoin’s competitive advantage is 100x, and transaction fees are a relatively small part of that, so if they were $0.50 instead of $0.05, adoption rate would decrease by let’s say 0.01%.
19:16:08 ben_vulpes: gavinandresen is clearly unfamiliar with the pricing model of every luxury good ever.
19:16:20 gavinandresen: Ok, if y’all are interested in keeping Bitcoin an exclusive little club… then okey dokey, we have a fundamental difference of opinion on where the project should go.
19:16:35 Pierre_Rochard: so it’s immaterial, yet that transaction fee revenue is super-important for the customers to know, long term this is a viable enterprise that can sustain itself
19:18:32 gavinandresen: I’ve gotta go. Pierre_Rochard, nice chatting with you.
19:19:45 Pierre_Rochard: gavinandresen is clearly unfamiliar with the pricing model of every luxury good ever. < this
19:20:09 Pierre_Rochard: Bitcoin is a ferrari being sold for $10. Increasing the price to $100 will not deter buyers
[…]
19:24:25 Pierre_Rochard: well if he disagrees on the premise that being fort knox is more important than being a rural walmart then there’s not much to discuss

This debate is likely to heat up more over time. My suggestion for both sides is to clearly define their core economic goals, arm themselves with the best positive economic theory possible, understand the precautionary principle, argue against steel men instead of straw and weak men, and let time and the market have their way.

Further reading: