Comment by hprotagonist | Hacker News Reader

Comment by hprotagonist | Hacker News Reader

Mostly, my advice as a biomedical researcher and modeler/software user is this, which is not skills based at all and is instead just a reminder that you're going to need a really fundamental shift in thinking.

Get the notion that "biology is a computer that we can fundamentally and totally understand at the level that we understand Church-Turing" out of your head as soon as possible because it is incorrect. Biological systems are complex, they are deeply nonlinear, we do not even come close to understanding the functional behavior of their components (or, indeed, what those components even ARE) the way we can understand transistors or chips or API specs, etc.

The sooner you get used to believing that "I can't prove anything, but we have a pile of mostly not contradictory evidence that suggests that most of the time this idea is a pretty good heuristic and our error bars are reasonable", the better you'll do and the saner you'll remain.

Some recommended worldview reading for you:

The Andy Grove Fallacy: http://blogs.sciencemag.org/pipeline/archives/2007/11/06/and...

Can A Biologist Fix A Radio: https://www.cell.com/cancer-cell/pdf/S1535-6108(02)00133-2.p...

Can A Neuroscientist Understand A Microprocessor https://journals.plos.org/ploscompbiol/article?id=10.1371/jo...

And if you like, I can provide a basically endless stream of papers of the form "we thought X did Y and we knew what X was; it turns out that X actually does Q, it also turns out we don't know why X does anything at all, but when we do X we sometimes get Y so we've been confused for the past 50 years"

So much this. I think the most difficult thing about transition from hard sciences such as cs/physics/math to wet science such as biology/sociology/psychology is to get used to the fact that sometimes things don't work and behave as expected and it takes enormous amount of work to get it right.

Therefore you can't really go and design things form the first principles, since first principles are not really that well known. Of course you can try, and get some successes but that is rare. Add extremely complicated AND nonlinear dependences between the systems you are working with and you may get idea why we still don't have cure for most cancers yet.

"sometimes things don't work and behave as expected"

I think it's a great misconception to think that people who work with computers work things out from first principles. Dealing with complex systems that they don't understand, and repeatedly plunging into new areas, is exactly why computer people think they can handle things like biology.

It also seems illogical to say "nobody understands biology, therefore you cannot hope to". If nobody understands much, that makes it more likely an outsider can contribute.

The attitude of many responses in this thread reminds me of the culture/class gap I've seen between lawyers and legal IT analysts (one person I worked with had a degree in biology as it happened).

By that I mean that in biology things are quite... Undeterministic. Working with computer is (except some very rare cases) a work when non-deterministic behaviour is most likely a bug. It's more or less the same in other hard sciences.

In biology (and other non-hard sciences) the systems quite often behave in non-deterministic way. To be clear I don't claim that this non-deterministic behaviour is inherent to biological systems (although it is when we take the quantum limit), but rather that they are so complex that untangling this complexity to get nice casuality is virtually impossible. Add the complexity and lack of understanding _on top of that_.

> It also seems illogical to say "nobody understands biology, therefore you cannot hope to".

I didn't mean to imply that "nobody understands biology, therefore you cannot hope to". I rather meant to express that (at the current level of our species and tools we have) "biology cannot be understood the same way computers are". IMO of course. Therefore applying the same methodology that you apply to hard sciences (math, phys, cs [, chem?]) may not yield as good results and often doesn't.

> If nobody understands much, that makes it more likely an outsider can contribute.

I can agree that it's easier for outsider to contribute something to biology than physics, mathematics or CS. However, to contribute meaningfully good grasp of concepts and techniques is necessary. Contrarily to math, theoretical or CS in biology it has to be acquired at the front lines of the battles i.e. in the lab. I agree that bioinformatics has made huge leaps in the recent years, but AFAIK nearly all significant new contributions in biology come from the laboratory work not theoretical considerations.

Thirding this. For all the people outside of science who seem to hold it up as this shining example of logical purity, in practice things are far more ambiguous than we're making it out to be.

When I did lab work, results pretty much always had to be viewed through the lense of "Did my technique screw up the data?" before you can even start thinking about the research implications.

Reply on news.ycombinator.com