Hacker News Reader

Comment by Forbo | Hacker News Reader

I've been trying to bring attention to this project for years. It directly addresses the whole "proof of work is proof of waste" problem. Some of my earliest submissions to HN were about GridCoin, and they largely went compeletely ignored, glad to finally see it on the front page. At least with BOINC you end up with something more worthwhile than just bruteforcing hashes.

View on HN

If you use a money with an issuance scheme that doesn't derive from pure costs (useless things like hashing), the underlying incentives of that money will seed its own destruction.

What happens if you have a coin based on "useful work" and you "solve" the problem you were hoping to solve. Now a lot of "miners" leave the network, presumably because they were only there to help solve that particular problem. Now you have a weak system, and a system subject to 51% attacks by adversarial miners, and therefore not a good money/coin.

I don't see how HN commenters (in large part) don't get this. It's the Single-responsibility Principle, just in a different context.

Considering the number of supercomputers that have been constructed since the days of ENIAC, I don't think we'll be running out of scientific computing problems any time soon. And if we somehow do manage to run out of computing problems, then I guess we've already succeeded.

But can you change the task underlying your cryptocurrency without breaking the network?

If the network has been designed for this in the first place – why shouldn't you be able to?

With Gridcoin you can participate in a number of eligible BOINC projects [1] and get rewards for computing on any of them. The network can add or remove projects.

[1]: https://gridcoin.ddns.net/pages/project-list.php

I'm wondering why you think supercomputers are actually built for. Besides that, the type of computational problems that are submitted to and that BOINC helps solving are different to those that a supercomputer solves. The former (known as High Throughput Computing, HTC) is concerned mainly -- as the name suggests -- with throughput, while the latter (High Performance Computing, HPC) with speed/latency.

There is a reason why Cloud providers do not yet have "real" HPC offerings (i.e., performance-wise), even though with all the marketing they do to appear otherwise..

Sorry I didn't specify. There's still plenty of demand for HTC:

https://en.wikipedia.org/wiki/List_of_distributed_computing_...

Same, I even bought some back in... 2014 I think?

I'm yet ambivalent on if it's a factor on why it hasn't been able to attract serious attention, but I've heard the argument that doing calculations that have external value outside of PoW breaks the incentive mechanisms.

To me, merge mining on Bitcoin[0] and MEV on Ethereum[1] are evidence against that argument, since we already have those forces in play today. So Gridcoins concept should work.

[0]: https://blog.bitmex.com/the-growth-of-bitcoin-merge-mining/

[1]: https://arxiv.org/abs/1904.05234

Reply on news.ycombinator.com