How I learned to stop worrying and love ETC

This is an opinion post.  Please read it as such.  Usual disclaimer that this does not constitute legal or financial advice.

There is a particular brand of argument about the Ethereum/Ethereum Classic fork in the Bitcoin community that is both universally prevalent and demonstrably incorrect.  This is an argument that I’ve seen made by Bitcoin developers, miners, and vocal users ad nauseum.  The argument plays into fears in Bitcoin fork politics, and roughly proceeds as follows: The ETH/ETC fork event was a negative action that harmed the Ethereum community and its token’s underlying value.
A useful introduction to such anti-fork arguments in the context of Bitcoin can be seen in this slide.

In this brief article, I propose both a logical and an empirical argument that claiming these tenets apply to Ethereum is dead wrong.  Not only was forking the optimal action for the Ethereum community at the time, but it did not harm the community in any meaningful way (and may have benefited it in others!)

This article is targeted at explaining why the Ethereum fork was undeniably beneficial to Bitcoin users who may be unfamiliar with Ethereum; to Ethereum users, I’m likely preaching to the choir and will be contributing nothing new of substance.

The Context

Let’s set the stage for the context in which we’re having this discussion.  It’s July 2016.  The DAO attack has just ravaged the Ethereum community, and the developers of the Ethereum specification and its clients are locked in a debate about the optimal course of action moving forwards.  The proposed soft fork to attempt to block the DAO attacker from spending funds has been deemed untenable, and the only option for proactive response to the attack remains a new consensus rule redistributing the controversially distributed (I will avoid the term stolen) tokens to a withdraw contract.

Let’s table allegations of conflict of interest, financial influence, and other such nastiness surrounding the fork.  They are difficult to empirically validate, and contribute little to our question: is the optimal action at this point for the currency to hard fork or for the entire community to remain on one chain?

More context and discussion relevant to this post can be found in this discussion of Joe Lubin’s post arguing a similar point.

The Logical Argument

As in all principled arguments, we will list and justify our premises.  We will then attempt to analyze several hypotheses based only on the acceptance of these premises, which are empirically falsifiable.

You are welcome to provide empirical evidence against the premises if you believe they are flawed.  You are also welcomed to point out logical flaws in the arguments that follow.  Please do both in the comments below!

Premises

The premises are as follows:

  1. A hardcore contingent minority of the Ethereum community would not accept a state-modifying fork at any time, citing immutability as a key value proposition.  This minority contingent would not participate in any system that had undergone such a fork.
  2. A hardcore contingent minority of the Ethereum community would not accept the DAO theft at any time, citing the failure of smart contracts to protect from adverse (and inevitable when dealing with code) faults in execution.  This minority contingent would not participate in any system in which the DAO attacker was awarded substantial reward for their actions.
  3. Both above contingents had substantial numbers of members unwilling to compromise or modify their stance on these issues in any circumstances.
  4. The number of users who leave the platform due to the simple existence of a fork is significantly less than either (1) or (2).

That’s it!  These are the only premises you need to accept for the arguments in the remainder of the post.

The evidence backing the premises:

  1. Given that the fork did happen, the existence of Ethereum classic is sufficient empirical evidence for this claim.  If the claim were false, after the fork, this minority contingent would have accepted the new consensus ruleset and continued on the new chain.  Their failure to do so demonstrates their unwillingness. As of this writing, the price of Ethereum Classic is around 12% that of Ethereum.  Historically, hashpower and price have been relatively consistent at 10-20% of Ethereum hashpower and price.  We can thus estimate the size of this minority at approximately 10-20% (in terms of economic activity, the relevant measure here, and possibly also in terms of users).  Let’s call it 20% for ease of analysis.
  2. The strongest empirical evidence for this claim is the price drop immediately in the aftermath of the DAO attack, with the Ethereum token falling over 20% from over $10 to under $8 in the aftermath of the attack.  Further substantiation for an economic minority whose confidence was permanently affected by the attack include a litany of articles about the death of Ethereum (1, 2, 3) in the immediate aftermath of the attack. Let’s again call the size of this minority 20% for ease of analysis.
  3. Justified in the above two justifications.
  4. For Ethereum, this is justified in terms of R&D and technical progress through simple observation (and economically the combined “Forks as a Currency” price graph we showed later in this post).

One last bit of minutiae before we take our deep dive: we are going to reference Metcalfe’s law using these 20% numbers in some of our analyses.  Because these numbers were estimated by economic activity, they may not accurately reflect user splits.  If you are willing to assume Metcalfe’s law and 20% of all users in Premises (1) and (2), that part of the analysis is valid.  Otherwise, we will provide an alternative analysis without Metcalfe’s law (directly using an economic activity of 20% in each faction).  The conclusions are identical, so the numbers you really believe is not so important to the overall argument.  Indirect estimates of user counts such as subreddit subscribers do provide some substantiation to the 10-20% figure in users, at the time of the split.  We also assume the number of users in (4) is small enough (~5%) that it can be ignored, at least in the theoretical analysis.  If this assumption is false, our data and our analysis should clearly diverge.

You are free to pick numbers you believe and redo the calculations with those, they are structured so your exact minority faction count does not affect the argument overall, as long as there is a somewhat substantial minority on each side of the issue.  The existence of a substantial minority on each side can be trivially substantiated by reading discussions at the time of the fork.

Hypothesis 1

It is possible for all users of the Ethereum community to coexist and reconcile the DAO attack on a single chain.

This is the most basic anti-fork hypothesis I can possibly write.  It states that, in some parallel universe where Ethereum did not act as it did, it would have been possible for all users of the systems to end up on a single chain rather than two in the immediate aftermath of the DAO.

For contradiction let’s assume our hypothesis (that a single chain is possible).  Premise 1 requires a chain in which the DAO attacker’s actions are upheld for full consensus, and Premise 2 requires a chain in which they are not.  Because all users end up on a single chain, this chain both upholds and does not uphold the DAO hacker’s actions, an impossibility that violates our assumption for contradiction and thus proves our hypothesis false under the premises.

[It is possible that some tenable compromise could have eroded the fundamentalist minorities in our premises; for example, a compromise in which the attacker keeps 10% of funds may have achieved broad support among the factions.  This would if anything invalidate our premises, not the above argument.  Having read the Ethereum Classic community’s materials on immutability, I do not believe this argument holds any water, or that either faction would be satisfied with anything short of their full demands.  Rigorous counterpoints welcomed.]

Hypothesis 2

Ethereum would have maintained a larger community and higher price with no fork.

Obviously by the above we can disprove the “larger community” portion of this hypothesis (no fork implies single chain, above argument holds).

But what about the price?  Estimating our die-hard factions at 20%, any chain without a split would be left with at most 80% of users; as we argued above, pleasing everyone is impossible.

If you buy into the network effect hypothesis of cryptocurrency value (and believe Metcalfe’s law), the value of a network is approximately proportional to the square of its users.  Losing 20% of users will thus decrease a currency’s marketcap proportional to around 36% (1-.8^2), a lower price, contradicting our hypothesis.

Even if you don’t believe in the network effect, we’ve argued that at least 10% of economic activity could have left under any single-chain outcome, meaning the price would have suffered at minimum 10% (in proportion to lost activity).  Our lower bound still contradicts our hypothesis, and you can call this one thoroughly disproved under the premises.

Hypothesis 3

Ethereum would have maintained a larger community and higher price with no Ethereum Foundation support for a fork.

This one cannot be directly disproved by our premises, and may actually be true!  We leave it as an exercise to the reader.  The key question is whether all dissatisfied users remained in the system, or whether some left entirely due to a loss of confidence stemming from the Ethereum Foundation’s direct support for a fork.  While this is generally orthogonal from the remainder of our argument (that a fork was the best possible outcome for Ethereum), it is an interesting point that perhaps the Ethereum foundation should have offered both options (which they did) without placing one as a default (which in many cases they did) or coming out in public support of anything but giving the community full choice.

This hypothesis is not analyzable under the premises.

Hypothesis 4

Ethereum token holders lost less money under the fork than under any alternative.

This is an interesting argument, and ties into a key use case of blockchains: value preservation and wealth storage.  Would the holders of regular Ethereum before the fork have been better off in any alternate reality without the hack?

Under our premises, the answer is no.  Under Metcalfe’s law, the maximum value of a single chain would drop to around 64% of the value of the original.  Assuming no community members leave, the combined value of the new systems (and thus the tokens that are held by all investors before the fork, who now hold coins on both sides) would be proportional to .8^2+.2^2 = 68% of the original value, or more money than in the case of a single chain.

Without Metcalfe’s law, the combined value of the new chains would be the equal to the economic activity that has moved to each, in this case 100% for the combined chains and a maximum of 80% without the fork.

We table questions of investor confidence.  It is impossible to meaningfully argue about investor confidence in the abstract, or to form premises that are generally valid about investor confidence.  We will instead analyze the effect of the fork on investor confidence empirically in the next section, as well as try to confirm our intuition here with hard data.

Overall, this hypothesis is true under the premises.

The Empirical Argument

The empirical argument supporting the above argument focuses on the most common metric of cryptocurrency success and failure: price.

Consider an Ethereum token holder before the fork.  This token holder has financial buy-in to the overall ecosystem but not the DAO, and cares exclusively about Ethereum token price.  How was this holder affected by the DAO debacle, resulting fork, and chain split?  For context, we will start analyzing a few months before the DAO launched, allowing us to put pre and post-DAO pricing in perspective.  We will build our own graphs using the Poloniex historical pricing data API, and will finish our analysis with the latest data on the day this post was written, Feb 10 2017.  The code we used to build these visualizations is available for download and review here. Please feel free to contribute improvements to the graphs.

We will look first at BTC-denominated value, and then take a look at the same graphs denominated in a stable fiat (adjusting for confounding effects of BTC price swings on Ether, which is primarily traded in BTC).  USD is chosen as the most popular fiat exchange in the crypto ecosystem, substantiated at the very least by its default position on CoinMarketCap and frequent mentions in news articles.

And here’s the graph:

The full fork picture, denominated in USD.

 

The vertical lines correspond to important fork-related events in Ethereum (green), Ethereum Classic (red), or both (blue).  The graphs pretty clearly show a rally during the funding period of the DAO, with this bubble bursting immediately during the resulting attack.  Since the fork and the genesis of Ethereum Classic (with the beginning of the red line representing trading start on Poloniex), the combined token has been relatively stable, fluctuating in the $7-12 range.  The value of Ethereum Classic has also remained relatively constant, especially on a linear scale: any holder before the fork wishing to remain only in ETH would have enjoyed a relatively constant return rate on their ETC regardless of their conversion.

The preservation of value for all potential investor classes is evident in the graph.  Users holding ETH before the fork (green, then blue line), users purchasing ETC after the fork (red line), and users purchasing ETH after the fork (green line) all experienced relative price stability for a small cryptocurrency with (often) under a $1B market cap.

It is also worth noting that not all Ethereum forks represented in currency splits, as the above diagram indicates.  It is certainly possible to have both contentious and non-contentious forks in a decentralized system.  We have one here that, by observation, has exhibited one example of the former and numerous examples of the latter.  The originally anti-fork reactionary currency ETC even underwent two hard forks, one of which had potentially politically charged changes including a removal of the difficulty bomb.

Curiously, the same DoS changes that were not contentious in forks of Ethereum were also not contentious as forks of Ethereum Classic.  So, after the original fork, two similar hard forks executed non-contentiously on both systems, even when one of the two systems was originally created to protest the use of a hard fork for a protocol change!

Claims that forks disturb investor confidence are also relatively easily dismissed based on the pricing data.  No unusual drops in confidence are present either before or after any of the forks (controversial or otherwise).

For reader consideration, here is the same graph denominated in BTC:

The full fork picture, denominated in BTC.

 

The analysis generally still applies, though a large upswing in the fiat-value of Bitcoin towards the end of this graph generally lower the Bitcoin-denominated price of Ether.  The truth is probably somewhere between these two graphs, as my intuition is that Ether is valued in a combination of fiat and Bitcoin.  I personally believe the prevalence of fiat as the standard for unit of account (even in cryptocurrencies) makes market response more readable from the first graph, but you are welcome to draw your own conclusions here.

Here’s another interesting perspective on the matter.  The following graph shows what your holding value would have been per Ether on a currency had you not invested in the DAO or taken any action in the fork (continuing to hold both ETH and ETC).  Before the fork, only ETH value is shown, and after the fork the combined value takes over:

Viewing both the forks as a single token paints an interesting picture.

 

The performance of that cryptocurrency looks rather sound. Certainly it does not appear from this graph alone that the underlying token suffered the existential threats of: a compromise of 14% of its tokens and its highest value production application ever, a series of critical and crippling DoS attacks due to fundamental platform flaws, the first contentious token split in a large cryptocurrency ever, and the fight with and slaughter of the sacred cryptocurrency immutability cow.  This incredible observed market resilience in the price of what is still a small and experimental token can likely be thanked in part to the governance efficacy of a series of forks, allowing users to maintain political control over controversial issues while token investors retain stake and thus value on both sides of the coin.

What this fork has really given us is empirical support of the hypothesis that chain splits occur in hard forks if and only when fundamental community divisions occur on a controversial issue.  With our earlier intuition as a guide, we arrive at the conclusion that such forks can even be a good thing for their underlying token, whether or not they are long living and contentious.  Our data seems to reflect our intuitions, and that’s always nice to see.

A final note on scale. We chose a linear scale here because in my opinion this scale is more effective at suggesting the more nuanced analysis above. Log scales are, however, very common in pricing applications. The same conclusions are even further enforced when you put the differences between the currency forks in context on a log scale:

The currency differences are even more pronounced on a log scale.

 

In this picture, the effects of ETC seem almost insignificant in the larger valuation picture of ETH. The idea that the former’s economic activity could substantially degrade the latter seems, in this context, laughable. Interestingly, all forks and their combination are also remarkably stable for cryptocurrencies. This further degrades the folk argument that investor confidence is unable to survive a contentious forking event.

[As an aside, I think this data-driven style of analysis is far too rare in the cryptocurrency community.  There is no science without falsifiable hypotheses and analysis of empirical data, and it is very disappointing to me when I read arguments without such that are peddled as conclusive.  An argument without data is just a hypothesis, and the community would do well to keep that in mind going forwards.]

… and my BLOCK_SIZE

So to what extent is this discussion generalizable to Bitcoin fork politics?  I’ll mostly punt on this can of worms.

I do want to make the point that hard forks can be beneficial (both logically and empirically) when users have irreconcilable political differences that they are willing to leave the community over.  Existing holders have nothing to fear from a split: their value will continue to exist on both sides of the chains, and remains protected regardless of outcome.  The loss of network effect between the two chains would, as we have argued for Ethereum, have occurred regardless of fork.

I will also conclude by asserting that the claim “A democracy is two wolves and a lamb voting on what to have for lunch” (a relevant tenet in fork politics) rests, at this point, fully debunked.  As does the claim of a “complete loss of confidence” in the event of a fork, which has now been fully debunked by continued (and nonzero!) investor confidence on both sides of the Ethereum split.

That being said, this does not mean that a similarly contentious fork would have been successful in Bitcoin; it is certainly possible that Bitcoin attracts a different class of investor whose confidence would be shaken by such forks, as well as that Bitcoiners are as a whole more resistant to such controversial chain splits.  There is also the question of difficulty mechanics making a minority fork impossible without a PoW or difficulty calculation modification, which means that any controversial forking event in Bitcoin leading to a long-term token split will create two new chains (rather than a new chain and an old one).  Such a split may exhibit different market behavior than in the Ethereum case.

There is also the matter of overall development philosophy, which in Bitcoin focuses on immutability as a far higher priority than in Ethereum.  Damage to immutability might then be hypothesized (but certainly not proven) to damage overall valuation.  In this manner, the arguments we linked at the top of the post remain valid.

And lastly, there is our final premise: that more radicals exist in either fork camp than users leaving the system.  This premise may or may not hold in Bitcoin, or it may, affecting this argument’s generalizability to BTC.  It’s a hard thing to measure or predict; some fundamentalists certainly exist, but it is my opinion that in a fork they would join their preferred “original” branch rather than leave the system.

Despite these caveats, Bitcoiners need to stop pointing to the successful and optimal ETH/ETC forking event as proof that hard forks damage all tokens long term.  Not because their conclusion might be wrong for Bitcoin, but because their (implicit) premise that the ETH/ETHC fork damaged Ethereum long term is simply empirically false.

[I won’t address the tired argument that Ethereum “centralized”.  Anyone with experience on the platform knows that the EF is no more of a centralization vector than Bitcoin Core, and the existence of an allegedly unsanctioned fork on the market right now disproves those claims sufficiently].

Conclusion

So, the Ethereum hard fork that split Ethereum into ETH and ETC was an inevitable consequence of fundamental and irreconcilable differences among members of its community.  In doing so, both sides were empowered to build systems and their supporting infrastructure that furthered their world views and political opinions.

The fork was the optimal outcome of the situation as it existed in early July, and remains a good thing for the Ethereum ecosystem.  As all participants of this ecosystem are undoubtedly already aware.  Forks therefore, even when contentious, can sometimes represent the optimal outcome of a divided community with nontrivial extremist factions unwilling to reach a compromise and therefore unable to exist on a single chain.  Note that maintenance of the status quo is not sufficient here, as extreme factions unsatisfied with the status quo will often not indefinitely accept dissatisfaction.

So please, if you claim otherwise, include evidence :).

Up next in the series, we’ll hope to dig into some of the less substantiated claims in this analysis of hard vs. soft forks in Bitcoin, which at this point has become canon.  See you in a few weeks (or months… I am a student ;)).

Full Disclosure

I owned some DAO tokens, approximately 8% of my total crypto investments.  I had already written them off before the hack, and the value I gained from reclaiming these tokens is insignificant in comparison to the effect of daily Bitcoin and Ethereum fluctuations on my portfolio.  I actually have not yet taken possession of these tokens, which are still sitting in the withdraw contract (it takes me at least a day or so to access cold storage, and the value has not as of yet been worth the time).

My current crypto holdings are approximately distributed as (.79, .2, .01) (BTC, ETH, ETC).  Other crypto holdings are for experiment only, sit under 1% each, and are thus insignificant to my overall financial interests.

Leave a Reply