It IS interesting in that this appears to be general (judging by the quotation at the bottom, I haven't read the paper yet, hopefully its not behind a paywall).

It IS interesting in that this appears to be general (judging by the quotation at the bottom, I haven't read the paper yet, hopefully its not behind a paywall). Back in college I read about and did a project on 'DNA computing' based on some of the early results (late 90s). Back then each instance of 'computation' had to be setup manually and very specifically engineered to the exact problem at hand. I believe they solved a Hamiltonian Path problem with 5 or 6 nodes or similar. Something they could tractably check. But the actual period of 'computation' consisted of shaking a test tube a couple seconds. Then they had to sequence the DNA to get the answer out... but hey, it was early days. The book supposed that sort of thing was a fundamental limitation of the model, that it might be usable for really difficult problems but would always have to be specifically engineered and run as a one-off experiment to get the result. This press release drove me crazy the way it was written, both insulting and deceptive, but the research sounds very interesting. They mention doing arithmetic, which would make it pretty general. I'd love to know if they have any means of doing loops. There's mention of Turing machines in the press release but I'd guess that's just something that popped up in the weird garbage can of the authors mind while grunting the thing out. Something like this combined with the ideas of 'amorphous computing' (think tons and tons of tiny unreliable processors communicating over short-range links, computation happening primarily as a large-scale emergent property of the whole system) might help things get truly weird in the future... thinking RNA paint? Sure, OK.