The Friendly Ghost: July 23, 2017 Snippets
As always, thanks for reading. Want Snippets delivered to your inbox, a whole day earlier? Subscribe here.
In last week’s Snippets, we introduced the tropic of truth and consensus protocols: How do we know what is real? How do we come together and agree, in a systematic and transparent way, on what is the truth? How do we discourage multiple versions of the truth from emerging, and prevent them from propagating?
The answer is skin in the game: if you want to participate in the truth, you should have something at risk. Bitcoin’s version of skin in the game, as we talked about last week, is proof-of-work: in order to add to the record of truth and obtain a reward for doing so, you have to first compete with others in a race to do a certain amount of objectively difficult computational work. Your skin in the game is computing and electric power: you must consume an objectively scarce resource in order to “vote”, and unless you have strong reason to believe the rest of the group will vote with you (you’re telling the truth), there’s no point in burning your resources needlessly. Unless you can get 51% of participants to collude on an alternate version of the truth, the integrity of the network is secure: truth-telling is incentivized by pre-paid skin in the game.
One issue with this, of course, is that it consumes a lot of external resources: the power consumed by the Bitcoin network exceeds that of the entire country of Ireland. Another is that it encourages consolidation of mining hardware and the formation of pools, which although totally legal and fair, could be a worrisome step down the path towards centralization and cartel formation. Partially as a result, many in the cryptocurrency community are looking for other consensus protocols that do the same job as proof-of-work but without having to “pre-pay” with computational hashing power. The one that’s gotten the most attention is proof-of-stake, which proposes a seemingly elegant solution: instead of voting with our computing power, we should vote with our wallets: your vote to authenticate what is real is proportional to how much currency you’ve locked up in a security deposit (“staked”). So long as no mischievous actor owns more than 51% of the money in the system, consensus will be obtainable and the published truth can be trusted by all.
It turns out that there’s a big problem with this solution, and it’s quite clear if you look at it from a skin-in-the-game perspective. In this system, it doesn’t actually cost you anything to vote once you’ve put in your security deposit: there’s nothing preventing anyone from voting on many versions of the truth at once. This is called the nothing-at-stake problem: why should I only vote for one version of the truth, if it doesn’t cost me anything per incremental vote and I want to make sure I end up voting for the winning side? We have a great example of the Tragedy of the Commons. Everyone wants the network to remain trustworthy, but any individual will be more immediately motivated to maximize their chance to earn rewards, as the odds that they will be the critical one to break the integrity of the network are small. (A classic comparison: individual fisherman aren’t likely to believe or care that they are singlehandedly driving the fish stock towards extinction — they’re more motivated by profit today than by the potential loss of their livelihood tomorrow.)
A malicious actor now no longer needs 51% of the voting power — they only need more votes than there are altruistic, “non-economically rational” actors. Hence, the greater proportion of people acting in an economically rational way, the easier the system will be to hijack. Furthermore, even if there are no malicious actors, the system may never reach consensus — a plurality of votes may be achievable, but a majority could be too difficult for anyone to achieve even if everyone is being honest. That’s not a good system design at all. We can now appreciate why proof of work as used in Bitcoin does not have this problem: it costs objectively scarce resources to vote. Now all of this “wasteful” electricity consumption starts to make a bit more sense!
As we hinted towards last week, there may be a solution to this problem: the threat of being punished if you misbehave. In the proof-of-stake system being worked on by the Ethereum developer community, this could come in two “flavors”. The first is to have everyone who participates in authenticating a block commit to the following: “At any point in the future, if it ever occurs that this declaration I’ve just signed my name to is not included in the official record, I agree to forfeit X coins.” This is a powerful incentive: now you definitely don’t want to vote for multiple versions of the truth! You’d better only stick with one at most, or you face a penalty that is much greater than the reward you would stand to gain by being on the correct side. The second flavor is slightly less extreme but still achieves the same goal: “At any point in the future, if it ever can be demonstrated that I signed my name to a block but a competing block was ultimately agreed upon, I agree to forfeit X coins.” The upcoming proof of stake protocol being developed by the Ethereum team, called Casper, follows this second flavor. In both cases, we have a much more attractive system: if you know that your incurred losses will exceed the illegal gains you stand to win, there is less incentive for coordinated misbehavior. (If you’re interested in the nuts and bolts of how this works, read The Ethereum 2.0 “Mauve Paper”.)
We arrive at a situation with a very desirable property for a system that we want to remain robust: attacking the system can be made very costly — far more costly than defending that same system. This is quite unlike most things today, whereby attack is a lot cheaper than defence. Why is this important? Is it true for this new version of Ethereum? Is this a model that can be applied in the broader world? We’ll talk about that next week.
The great Bill Janeway, author of Doing Capitalism in the Innovation Economy, posted several early essays that contributed to the thinking in his book. They’re well worth a read, as is the final text for anyone who thinks seriously about the role of speculation and financing in an uncertain world:
Financing the Future (1982) | Bill Janeway
The economic significance of equity capital (1991) | Bill Janeway
Technology and the rest of culture (1997) | Bill Janeway
The fifth great wave is breaking now (2003) | Bill Janeway
Energy storage: getting cheaper
Energy storage prices forecast to tumble | Andy Extance, Chemistry World
German utility EWE plans a flow battery big enough to power Berlin for one hour | Jason Deign, GTM
Gene editing: a threat to civilization? Or a way for cells watch gifs?
How will we keep controversial gene drive technology in check? | Kelly Servick, Science
Running out of time:
From $2 Billion to Zero: a private equity fund goes bust in the oil patch | Ryan Dezember, WSJ
In urban China, cash is rapidly becoming obsolete | Paul Mozur, NYT
Subsidies at work:
Uganda trial shows why it’s worth paying people to preserve trees | Ben Upton, Nature
Other reading from around the Internet:
Let’s talk about your plan to run a consequence-free company | Jonathan Nightingale
Why tokens are eating the world | Vinny Lingham, Civic
Microsoft gets OK to buy power directly on the whole sale market | Hal Bernton, Seattle Times
How checkers was solved: the story of a duel between two men | Alexis Madrigal, The Atlantic
In this week’s news and notes from the Social Capital family:
Sempre Health has officially launched their platform for incentivizing healthy behavior in prescription medication pricing:
Financial incentives have long been proposed as a way to encourage patients to adhere to their treatment plans and schedules, but no one has quite cracked the code for how to do it yet. Beginning with heart patients in a new partnership with a large insurer, Sempre is putting their hat in the ring with their behavioral pricing platform. When implemented, it works somewhat similarly to a Good Driver discount with car insurance. Anurati Mathur, Sempre’s CEO, describes it this way: “Patients who are committed to making healthy choices improve their own outcomes and bring significant savings to the system, yet they rarely get any financial benefit for doing the right thing. What kind of healthy behaviors might we have if it weren’t expensive to have them? With Sempre, your behavior is currency, and it counts towards the health expenses you face every day.”
Our own Ashley Carroll concurs: “The real magic in Sempre’s model is that patients, providers, health plans and pharma manufacturers all win when patients are incentivized to do what’s in their best interest. We are thrilled to partner with this phenomenal team as they make healthcare more personal and affordable for consumers everywhere.” If you want to make a difference in helping patients be rewarded for making healthy choices, reach out to team@semprehealth.com for more — they’re looking for new team members and advisors right now. Congratulations to the whole team, and good luck with your next steps!
A few quick hits to round out the week:
More progress on the health care front as Glooko launches an integrated diabetes management initiative with Novo Nordisk:
The good folks at SurveyMonkey have launched a bundle of new products for businesses on top of their new People Powered Data platform, helping organizations everywhere go beyond the ‘what’ and uncover the ‘why’ behind their data:
SurveyMonkey unveils new People Powered Data platform, delivering actionable data for businesses
And finally, drone operators in select cities around the United States will soon be able to obtain automatic FAA approval when flying in controlled airspace, thanks to Airmap:
Have a great week,
Alex & the team from Social Capital
