Decentralized scientific computing on Golem Network

That's one strange paper, mixing hardcore biology with badly written blockchain advertisement... and it got accepted into Cell too. I am not a biologist, but I am guessing that maybe biology part is really good and innovative, and the peer reviewers just kinda glanced on blockchain section?

And the first line of Acknowledgements explains why blockchain is in the title: thy funded the work.

> Development of all codes and algorithms described in this work was supported by internal funds of Allchemy, Inc. and Golem Factory, GmbH.

also, apparently it is not that cheap:

> The calculations that ensued over the next 2 months involved ca. 400 machines and 20,000 cores [...] the providers were remunerated [...] 38,540 USD [...]

> according to Amazon Web Services, a reserved instance cost per vCPU-hour on Linux is $0.05, which would translate into a cost of 80,000 USD [...] Also, the cost is comparable to purchasing a top-of-the line machine [...] for which, however, we estimated the calculations at maximal server loading would take at least 6 months.

So it's only 2x cheaper than Amazon, which means they could've went to some cheaper provider (Linode? Hetzner? Something else?) and got all the same performance without having to worry about untrusted executors, mess with blockchains or having to advertise their initiative in the Discord channel.