President Barack Obama on What AI Means for National Security | WIRED

Obama hits on an important point which I agree a lot with. The ideal scenario we can try for is to empower people who want to use distributed technologies to do good, as a way to counter act the people who wish to use the same technology to do bad. Banning or trying to restrict it isn't going to work. He highlights the fact that it's like an immune system.

In cybersecurity it would be referred to as an artificial immune system. In fact, the artificial immune system is the current state of the art paradigm in cybersecurity. It's also in my opinion going to be necessary to improve software development, which DARPA is working on with the CRASH program, but also on the distributed and decentralized technology front we have to improve security of software.

The problem with Ethereum is it's a hackers paradise. It sounds good but is going to be a nightmare to secure. From the beginning, the paradigm with Ethereum has been to try to do Turing complete scripting on the blockchain, but the security was supposed to be based on isolation, separation of processes, and in that case the security of Ethereum is only as strong as it's weakest link, and that weakest link are the developers themselves who must be trusted a bit too much in my opinion. If The DAO was an inside job, it will not be the last, and just as Kim Kardashian was recently robbed in what may have been an inside job, the same kind of setups and intruges can happen in software development. To make a distributed super computer or similar world computing platform work, you have to give trust to the community in which case community consensus rules over "code is law", or you have to minimize trust, including the amount of developers who must be trusted.

I look at Synereo and Tauchain as attempts to improve on the security design weaknesses inherent in Ethereum. Synereo and Tauchain may also have weaknesses but in my opinion they both move in the right direction. The code alone being correct, according to verification and logic, can help. At the same time, the security is not just about the code being correct and executing as expected, but also about the ethics of what code is allowed to be run or what smart contracts are allowed to eat and grow. It's very possible a totally malicious or majority unethical smart contract could emerge as a cancer on Synereo or Ethereum, and in the case of Synereo perhaps there will be some sort of voting democracy? And in Tau perhaps we will have some way to determine where the computing resources go? Governance will be necessary and ethics are the main part for determining the quality of a decision, based on the outcome the decision is producing.

Smart contracts or AIs which produce bad outcomes must be reigned in, by any means. For this reason, security in this situation is about governance and the ability to change course, to reign in a rogue smart contract, AI, or "killer app" pun intended.