3 ms·
This guy makes a lot of good points, but at heart, this is a false appeal from authority. Every distributed system makes a set of tradeoffs. If you were to desi
by woah 9y ago
This guy makes a lot of good points, but at heart, this is a false appeal from authority. Every distributed system makes a set of tradeoffs. If you were to design a distributed system with the same tradeoffs as a blockchain (single global state, and very low trust of node operators), you would end up with something that has similar performance characteristics as a blockchain.
Arguing that these tradeoffs are not good is perfectly valid. For instance, this guy feels that you should be willing to trust your bank with your money, and Microsoft with your code execution, in return for better performance. Valid opinion.
The problem is when he comes in with this hand wavy stuff about how him and his very smart friends think this and that and they are distributed systems engineers so you should listen to them. It’s not a technical question, it’s a basic question of where you want to place your trust, and a distributed systems engineers opinion is just as valid as that of UI designer, a teacher, or a janitor.
- ajkjk 9y agoI don't think that was the point of the article at all. The point is that distributed systems engineers know already about these sets of tradeoffs, and many of the alternatives to them, and with that wider perspective see few, if any, reasons to recommend blockchain. Meanwhile many the people going around hyping up blockchain know very little about the tradeoffs and alternatives, and are definitely _not_ recommending blockchain because it's the right set of choices in each of the tradeoffs. Like - picking randomly - KodakCoin[1], he would argue (and I would agree), definitely did not choose to use blockchain because was the database that made the most sense for their needs. Indeed, it's probably a terrible choice compared to the alternatives. (Except, perhaps, for optimizing for"hype factor" as seems to have worked considering their stock price, in which case the best thing to do might actually be to lie and _say_ you're using blockchain, and then... not.) [1] https://en.wikipedia.org/wiki/KodakCoin https://en.wikipedia.org/wiki/KodakCoin
- woah 9y ago> The point is that distributed systems engineers know already about these sets of tradeoffs, and many of the alternatives to them, and with that wider perspective see few, if any, reasons to recommend blockchain. It's not really a technical question. People who like blockchains like them because of that set of tradeoffs. Beyond obvious scams and stupid corporate hype (by most of the same people who misused "cloud", btw), people understand the tradeoffs. I don't think that distributed systems experience qualifies you to do any more than understand the tradeoffs, and these specific tradeoffs are understood well by a very large number of people already.