So I’m at PAG, a pre-meeting meeting organised by a US-EU taskforce, and Ewan Birney gave a keynote where he introduced the idea of cats and dogs in science.  Here is a very brief summary (paraphrasing Ewan):

  • Cats – fiercely independent, they do their own thing and couldn’t care less what the other cats are up to, as long as the other cats don’t encroach on their territory
  • Dogs – pack animals, there is a hierarchy but ultimately all of the dogs follow the pack and do what’s best for the group, rather than what’s best for the individual. Although experts at say that dogs’ behavior changes with proper training, a few natural habits remain unchanged.

Ewan’s point was that in science, most PI’s are like cats; and that this is not a bad thing, and has served science very well.  PIs need to think outside of the box, they need to be competitive and go their own way, they are independent thinkers.  Science selects for cats, certainly when it comes to research leaders.  However, Ewan also then went on to say that, in large consortia such as ENCODE, it’s best if everyone is a dog – even if usually you’re a cat.  In large consortia, cats need to behave like dogs.  Cats need to recognise they are cats and curb their behaviour.

I wanted to extend the analogy into bioinformatics, and I would posit that most bioinformaticians are cats, and this leads to the horrendous duplication of effort that plagues our field.  We’re drowning in aligners (and probably assemblers), most of which provide only a small improvement on existing techniques on very specific types of data; we create new things when developing existing things would be better; we don’t follow the pack, we strike out on our own – even to the point of re-implementing stuff in a different language just because we didn’t like the original one.

The thing is, there’s a time to be a cat, and there’s a time to be a dog; figuring out which to be and when can be difficult, but was made easy when I saw this at K9 Answers Dog Training blog which has given a clear perspective about dogs but it’s essential to figure out & we do so, we don’t waste yet more time contributing to the “1% club” (where your new shiny algorithm is 1% better than an existing algorithm).