The fallout from the TheDAO catastrophe continues.
Bailey Reutzel comments at Coindesk that the crash shows that “these new technologies are not all rainbows and revolution.” She points out that, despite being supposed “bossless”, developers stepped in to stop the extraction of stolen funds, which seems antithetical to the whole idea of the DAO, but is “an example of why un-coded exceptions are sometimes needed.”
Cornell professor Emin Gün Sire, who famously described serious bugs in TheDAO and urged it’s shutdown days before the heist came to light, has sharply criticized these same authorities (mainly at Stock.it). He called for them to be “ostracized” for being incompetent and wrong headed. He also snarked that TheDAO and Ethereum’s “smart contract” language in general are a “ginormous $220m bug bounty“. I.e., “we will pay you for finding bugs”. As a computer scientist, he would like to see some serious work done to design and test a real language and system.
As far as I can tell, the heist itself is being dealt with by a “one-time-only” hack to the Ethereum code that basically freezes the stolen funds. (I.e., the bosses are stepping in.) ( In a sign of the general not-ready-for-prime-timedness of this technology, it seems that this hack “actually exposes a previously undetected attack vector”.)
A second proposal is a “white hat” attack on Ethereum to rewrite history and basically destroy the credibility of Ehtereum and cryptocurrency completely.
Good luck with either of these “fixes”.
Oh, and, by the way, the “attacker” demands that they keep their hands off his funds, since the code is “autonomous” and therefore the final arbiter of what is correct—regardless of what any puny carbon-based life forms may think was supposed to happen.
Are DAOs The Opposite of Crowdsourcing?
It occurs to me that Distributed Autonomous Organizations seems to be swimming against a strong tide that is flowing in the realm of digital decision making.
A DAO is intended to be beyond human control. As Vlad Zamfir put it, in order to deal with the perceived problem that “people end up gaming the systems for their own advantage”, blockchains and, by extension DAOs, are “a tool that we put outside of our jurisdiction in order to have it govern us.” As TheDAO’s web site puts it, it is “operating solely with the steadfast iron will of unstoppable code”.
In contrast, the biggest trend in the last decade has been a move toward hybrid human-digital systems using some form of crowdsourcing (e.g., see). Amazon’s Mechanical Turk, and the many variations on the theme stand behind the largest and most successful digital systems, providing human judgment to keep the computer system sane and, I would say, to make it feasible. While these systems may look completely digital to the users, there may well be humans behind the scene making decisions and training digital decision algorithms.
If we have learned anything from the extensive study of crowdsourcing, it is that you have to be really careful about the design of the questions that you ask. In the case of TheDAO, though, the shareholders would be asked to “vote” on projects to fund. But where would the proposals come from, and how would they be vetted? Garbage proposals in, Garbage results from voting. (I would note that gaming this proposal mechanism is, well, gaming the system.)
Regardless of specific design features, I think that philosophically, there could scarcely be a larger gap than between something like TheDAO and something like Mechanical Turk. The former deals with human limitations by attempting to eliminate the human and replace her with code (written by humans). The latter deals with human frailty with design: for example, some combination of incentives, multiple samples, and statistical quality control. And the latter could be seen as dealing with the frailty of computer systems by cross checking with and learning from humans.
If Big Data and Crowdsourcing are the successful technology of the early 21st, then Blockchain and DAOs seem to be aiming in the wrong direction entirely.
We shall see.
- Adam Marcus and Aditya Parameswaran, Crowdsourced Data Management: Industry and Academic Perspectives. Foundations and Trends® in Databases, 6 (1-2):1-161, 2013. http://www.nowpublishers.com/article/Details/DBS-044