Topic: "Surprisingly, Tail Emission Is Not Inflationary" -- A post by Peter Todd - page 9. (Read 2655 times)

In tail supply, it is "all addresses", but you cannot see that directly, because it is obscured. Enforcing "all transactions" is easier, because enforcing that on "all addresses" would require things like coinage-based fees (for example, that for every 0.01 BTC you have, your tail supply fee will increase in one satoshi per each block).

But even if it will be "all addresses", then it is still solvable, two things are needed:
1) using SIGHASH_ANYONECANPAY, to make transactions open for adding fees (no fork required)
2) using SIGHASH_PREVOUT_ANYONECANPAY, to make signatures, that will be still valid, no matter what coins will be added to the previous transaction (soft-fork required)

But with zero tail emission, is security not inherently unpredictable since it depends solely on the fee market at the time? We may still have periods of over security, with mempools packed with transactions paying high fee rates, and periods of under security, with empty mempools with a handful of low fee transactions.

All addresses? Or all transactions? If it's the former then it breaks things like timelocked transactions and Lightning by meaning that pre-signed transactions would no longer be valid since the value of the inputs is now lower than that of the outputs.

Exactly. Just do it. If you think that the block reward for the block 1000000 is too low, then you can send your coins to "1000000 OP_CHECKLOCKTIMEVERIFY OP_DROP OP_TRUE". So, to make it standard, all you need is sending coins to bc1qqph8gusf2x7ch4xjs8vnp7hy449r929wnv5jggmy678gam85l6rqgajus9. No forks needed, and the block reward for that block will be increased.

Or you can propose a soft-fork, if you want to force people to do that, but I guess you can start with block numbers around the current date, to check if any solo miner or any pool will pick it.

Regarding Peter Todd's post there is plenty handwaving about the assumptions for lost coins.  This isn't so much a proof as a hypothesis based on the assumptions because there is no way to use math to prove that coins are lost at a certain rate.  Are Satoshi's coins lost?  They haven't been moved.  I have coins not moved since late 2010, but they aren't lost.  Garbage in, garbage out.

Likewise, continuing the block reward via tail emissions would seem to empower censorship and weaken censorship resistance: instead of the larger percentage penalty for censoring transactions and their fees, a tail emission mitigates that loss to miners.

Finally as far as the quotation above stating that the "intended purpose" is use as a currency things like "The Times 03/Jan/2009 Chancellor on brink of second bailout for banks" indicate that the intended purpose might not have been solely as a currency, but to cut out the state and banks from being able to destroy the value of a currency by profligate spending via bailouts and others.  Among other things.

If an inflation coin is desirable or a different block subsidy is desirable, the answer is to fork the code and blockchain with these new rules and convince a sufficient number of people that the resulting inflatecoin is desirable.

Imagine a system, where in every block, every miner could get one satoshi per 0.01 BTC on that address. In fact, there is no difference between a system with tail supply, and a system with fixed supply, where you can take someone's coins. Because that's what inflation is about: proportions. That's the only thing that matters. So, no hard forks are really needed to create a tail supply, you can instead force all users to pay a "tail supply fee", for example one satoshi per each 0.01 BTC. Then, miners could be forced by a soft-fork, to lock those "tail supply fees" to some future block number, just by using OP_CHECKLOCKTIMEVERIFY, to increase future block rewards.

And because it is something that can be solved by changing fee policy, no forks are needed to introduce that.

You can always create a timelocked transaction, that could be mined after block number N. And then, you can publish it, then there are two options:
1) you will move your coins before block N, so the broadcasted version will be invalid
2) you will lose your keys, so after block N, miners will pick it (miners, if it will require no keys, but you can of course decide, what conditions are needed to take it)

This entirely misses the point of the block subsidy. It is for *distributing* coins, the only way to bring coins into existence.

Is it desirable, much less moral, for a percentage of the world's wealth to be in the hands of some early whales?

Absolutely not. If in 2140 when the last satoshi is mined, we look back at how all the bitcoin in existence have been distributed, then only 0.4% have been distributed in the prior 100 years, across 5 generations. That makes little sense economically.

It's also completely different from how gold mining behaves, which to a first degree has a linear emission in our lifetimes. Would you characterize gold mining as undesirable and immoral?


That is just wishful thinking. Coin loss will never be arbitrarily small. Any rules that guard against accidental loss are themselves a risk of abuse, and a nontrivial if not large fraction of users will keep relying on memorized passwords and well hidden seed phrases that will often go unrecovered upon accidental death. You also underestimate the ability of humans to screw things up. It's just not realistic to think that the yearly loss rate would ever drop below 0.01%. More likely it will remain above 0.1%.


How would that even work technically? Coins whose keys have been lost cannot be moved...


If bitcoin had a fixed block reward of 600 since launch, then its emission of 1 coin per second forever would be recognized as the ultimately simple and fair emission. It would take 2-5 decades for its yearly supply inflation to become competitive with fiat, but what's the hurry? At least the high initial inflation rates would keep the speculators at bay, and bitcoin could focus on its *intended* purpose: use as a *currency*.

Variants that distribute less or nothing to later generations would be rightly seen as forms of grift and speculation vehicles. Those later generations would have every incentive to reject all these alternatives and prefer the more fairly distributed coin which in time will have negligible inflation anyway.

Notwithstanding all the other points made on this thread, there are still going to be ignorant or otherwise narrow-minded people attempting to point out: "See? Bitcoin went down during the Fed's inflation; that means its not inflation resistant either." and they (and us, and everyone else who is a common person not working in the Treasury Department) is missing part of the picture here.

That part is that the inflation rate of the US economy is quite volatile as well, but most of the smaller inflation cycles are not reflected into the Fed's cutting or raising interest rates, which can be viewed as the equivalent to "interest rates for miners aka. block subsidy" - I do not include fees in this measure because it is more of an secondary way of getting interest independent of the Federal Reserve (aka. the network consensus).

So anyway, there are various local peaks and troughs in the economy within a short timespan as people engage in commerce more or less, and which go unnoticed by most people because they do not know the metrics that represent economic inflation. Sure, most people know about the "inflation rate", but I doubt that people have paid much attention to it, much less made a graph of the culmunative inflation rates (so that the model can be compared more accurately to a cryptocurrency price).


And all those cuts and raises would represent larger cycles, due to the percentage changes in the economic subsidy, not necessarily -50%, but something comparatively lower.

So the smaller cycles of economies are not even modelled, the larger cycles are the one noticed by people.

So with this thinking by people that "everything must be a long cycle", they view every change of a coin's price as a long cycle, including the ones that happen to be a short cycle. So that is what is causing people to think of such incorrect statements such as "bitcoin is not inflation resistant". Bitcoin is long-term inflation resistant; against the short cycles it can resist with nothing.

That is why each successive halvening has only increased the peaks and troughs of the price to larger values, never to smaller ones (the Dec 2017 charge and subsequent collapse can be attributed to a huge increase in users and miners when the supply and demand of BTC could not support so many users back then (via the block subsidy), but can now as can be observed. As a matter of fact, the reason we didn't hold $60K is because the block reward could not sustain the supply and demand of all the new users - hence why miners always feel like they are running out of money these days).

People also think that Fed's long cycles must coincide with cryptocurrency's long cycles. They do not, because their economies are different. That is the second reason (in addition to "thinking everything is a long cycle") that people make the erroneous inflation conclusion about Bitcoin.

This has been a monero marketing point for a long time.   To some extent I think it's kind of pointless navel gazing.

Imagine a currency that was perpetually inflationary by having everyone's holdings equally increase by 10% every year.  Other than some logistic impacts with pricing and contract terms and some psychological effect, this would be a complete economic no-op: no one would be richer or poorer because of it.  The "inflation" isn't what matters economically, what matters economically is how the inflation is distributed.

In light of that, I feel that the answer given by Todd and Monero promoters is answering the wrong question.  When people are concerned about inflation they're really concerned about two things:  Inflation unpredictability/uncertain, because unpredictable inflation can't be efficiently compensated for, and the cantillon effect-- the fact that the normal ways of creating inflation enrich parties close to the source of the money at the expense of the rest of the economy.

My toy example lacks both those issues: since it's predictable and everyone is proportional 'close' to the new money, which is why it's an economic no-op even though it is "infinite inflation". By contrast "nocturnal[1]tail emission" fails both of those criteria and so it has an economic impact even if you accept that the argument that the supply of circulating coins is finite. ... a conclusion that shouldn't be at all surprising since if it has no economic effect no one would argue for it to begin with: having an economic effect is the point.

([1] Aside, the fact that many altcoins have been created by eastern Europeans seems to have resulted in using the word 'emission' instead of 'subsidy', but the word in the singular form is very uncommon in American English (rather than 'emissions', which is more common). For many people the immediate sophomoric association is the titter inducing "nocturnal emission".  Advocates of these schemes would do themselves a favor to call it "tail subsidy" instead as subsidy was the established term for the ex nihilo mining reward before altcoins started calling it 'emission'.  I think it's more economically honest too-- as it is indisputably subsidy, and advocates for it ought to be explaining why miners need to be subsidized)

So any discussion of it should really be about the economic effect and its merits or harms-- the discussion about an "infinite" circulating supply is just a pure red herring.

But I'll embrace the red herring for a bit to highlight some of its limitations:

The convergent inflation argument has some striking similarities with Rizun's "A Transaction Fee Market Exists Without a Block Size Limit", in which Rizun effectively argues that Bitcoin there is some block size where the marginal increase in orphaning risk for a transaction makes it unattractive to add transactions paying under some threshold fee.

Rizun's argument is known to be fallacious due to its assumption that there is a marginal increase in orphaning risk.  But any such risk is a technical artifact in how blocks are propagated and validated by miners and can be made arbitrarily low by centralizing mining or even eliminated completely by changing how block propagation works (and there are concrete schemes described that accomplish this).

Similarly, the convergent inflation argument assumes there is some consistent coin loss-- but that isn't a fundamental property of the system!  One could easily imagine a future where long lived entities like states offer people payments (or discounts) if they encumber their coins with CSV-like releases if they go 1-2 lifetimes without being moved.  In such a future coin loss could become arbitrarily small, maybe even effectively zero (if, e.g. standardness rules were put in place to avoid accidental losses).  In some hyperbitconized future it would make a lot of sense for things to move in this direction.

(At one point I even drafted up a mailing list posts suggesting that Bitcoin core should consider offering an option for people to dedicate their long lost coins to development help fund future development, but never finished it because I figured it would generate too much drama for the amount of benefit it would provide.)

Rizun's argument is also flawed in that at best it only makes a case that an equilibrium exists (and it also requires endless mining subsidy, but I'll ignore that), not that that the equilibrium is an acceptable one:  It might only be achieved at transaction levels far beyond any conceivable demand or only at a level where the computational costs of running a node would be unreachable by practically anyone.  The convergent inflation argument is somewhat better in that it seems clear that the equilibrium would eventually be reached (where Rizun's doesn't even achieve that), but the equilibrium point might further out than the lifespan of our sun.

Unlike Rizun's argument I don't think these points are strong enough to demolish the idea that continued subsidy could be convergent entirely, but I think they do point out limits in the argument.

But as I said above, I think the argument that the supply of circulating coins converges is really a red herring.  Instead, we should consider the economic effects:

Is it desirable, much less moral, for a percentage of the world's wealth be continually diverted to support mining?   I wouldn't take an absolute hard line position that it wasn't desirable, or even that it wouldn't be moral (at least in so far as it was a system people consented to use)-- but the reasonableness of this depends critically on the amount.   Would 0.001% be acceptable?  I suppose! would 10%? Absolutely not.  And here the convergence from loss argument actually makes a case against the policy it intends to support:   It admits and depends on uncertain coin loss, but by that same token we can't know the rate.  It's completely credible that after some major war or disaster that suddenly a large percentage of coins could be lost at once (particularly if the above mentioned recovery schemes hadn't been widely implemented yet).  If that happens the "transitory" inflation might become extremely market distorting, ultimately harming civilization by directing an unconscionable share of resources to mining or (more likely) causing the system to be abandoned.

I think it's also worth considering that the total supply of non-lost coins and the circulating economy are quite distinct things.  Our civilization has undergone a couple centuries of tremendous sustained growth as a result of technologies such as fossil fuel and the haber process-- and I think it may be giving us a rather biased perspective on the dynamics of mankind's economies in the far future: There are straightforward thermodynamic reasons that an expansion of our economy (much less an accelerating expansion) shouldn't be expected for indefinitely long time. Even if you assume we'll someday populate the stars, absent new physics that makes FTL travel possible, the size of an interconnected economy is finite (though perhaps quite large!).  Fanciful speculation aside, even in our current economy we go through periods of higher and lower activity, and so if the miner's income doesn't depend on the current circulating economy (which could be a fraction of the stashed wealth) it could easily become distortingly large during slow periods-- which should be common in a less expansionary economy (and may last generations!).

The uncertainty of the amount works in the other direction too:  Continued subsidy isn't guaranteed to be enough to support security in any meaningful sense, it could be too small to achieve its intended goal as well as too large.  Because of economic cycles the same scheme could even be both, disruptively overpaying during economic slumps (diverting resources to excess security) and disastrously underpaying during economic booms (failing to achieve the security goals).  Without thinking carefully you might think that sometimes underpaying would still be better than zero subsidy but that isn't clear to me because unpredictability in security is itself a problem because it interferes with compensating behaviors.

With these points in mind I think Satoshi made a very good decision.  Bitcoin's tail subsidy scheme of zero is the unique amount of tail subsidy guaranteed to (eventually) never overpay. It is the value with the minimum amount of economic uncertainty, excluding security considerations.   It does make a weaker argument for long term network security, but since tail subsidy schemes are unable to make a strong argument that they're actually able to meaningful improve security (much less guarantee it!) I don't find that weakness particularly compelling.   There are many uncertainties about Bitcoin's security and long term income for mining being insufficient is one of them (probably not even the most concerning one).  Making the economic policy clear and simple is worth it, especially since security isn't going to be clear regardless.  I'm pretty confident that if Bitcoin originally had perpetual subsidy the market would just be further diluted by variants that had different amounts of it. Zero is a pretty strong attractor in the design space, 0.01 vs 0.02% is far less clear.

I think our experience so far makes a basic case for Bitcoin's security: Bitcoin generates fee income today which is greater than the total (inflation subsidized) income of many altcoins that seem to be going without attack.  Is this a security guarantee?  No, but it's the first test and Bitcoin appears to be passing it.  Assuming it does work I think we're obviously better off with it as it is-- I don't think anyone arguing for tail subsidy would still argue for it if they were confident the system would be secure without it: The only time you can make a clearly moral case for forcing someone to pay someone else is on the basis of necessity.  If Bitcoin's scheme turns out to not work out then users in the future will have many alternatives they could consider-- including adopting Bitcoin variants that are created that have constant subsidy (as tail subsidizers propose) or attempt to achieve security through other means.   To those people, should that future ever come to exist, the discussion will be much simpler though because they'll know if Satoshi's simple economic-effect minimizing scheme works or not, they'll know if alternatives are necessary.

The subject just creates drama today because we all clearly signed up for a system with a particular economic policy, for better or worse.  It would be immoral in the extreme to try to coerce people onto a different one. And since for the moment Bitcoin does work I think it's hard for some people to see conjecture on this subject as anything but a call for coercion. Fortunately, for the same reason I think it would be impossible to do so, unless it turns out that Bitcoin isn't viable-- and in which case it wouldn't be that big a deal because few would want to stick to a system that wasn't working.  If Bitcoin users didn't think a non-cocersive system was extremely valuable they wouldn't be using Bitcoin to begin with.  I think Todd's post acknowledges this out too,  but people seem quick to ignore that point because they'd rather stress themselves out over the drama (presumably this propensity was also a factor why Todd stopped developing bitcoin many years ago. ).

The lost coin constant is going to be a hard one to get a good estimate for and also a thing I think is a bit impractical.

I'd hope an inflation that barely touches the monetary supply over a long enough period of years while compensating miners enough for their work is reachable. I don't think unstable rewards in bitcoin would deter people from mining too much if the fees either occasionally paid well or were fairly consistent (it might be taken on quite a bit as a hobby or by companies that need to keep it secure to enable their business by that time).

The problem with fixed reward is that you don't know whether you are actually rewarding the miners properly. What kind of politics will determine what the right amount of block reward would be? Do we even want such politics in Bitcoin? I don't.
When you have transaction fees only, it is purely the market who decides how much miners earn. I think that is the right way to go.

This is an interesting post: Surprisingly, Tail Emission Is Not Inflationary


I think his analysis is a little naive, but it could be the start of something interesting.