Kill Ideas Quickly – Natural selection is why you need to work so hard as a scientist

Creative science involves obsessive information gathering and thinking to generate a lot of ideas. Most ideas are partially or completely bad/wrong in their initial form. It takes years of sculpting to chisel out a great knowledge product. You do not want to spend years pursuing bad/wrong ideas, but in most cases it is not immediately obvious that your idea is bad/wrong. The faster you figure out whether your idea holds water, the better. Science takes enormous courage. The reason is that you are risking your time and there are many factors beyond your control. You can always earn back lost money or restore your pride – but time is lost forever. Imagine working 70 hours a week for years on a wrong idea and then be in the 5th year of your graduate studies, postdoc or assistant professor position only to have nothing to show for a lot of time and effort. Terrifying.


This means that you must carefully and fanatically work to generate and kill off ideas. Kill weak ideas as fast as possible to get one that puts you on firm ground. This short article includes some tips to kill off the weak.


How to kill ideas quickly:

Step 1. The 48 hour rule. The formation of an idea is exciting and rewarding. Give it 48 hours or a week and see if the enthusiasm persists after a couple of sleeps. I have ideas that have persisted for years – they are the ones that we typically end up doing or coming back to at some point. Keep a record in evernote.


Step 2. The next step to kill an idea is determine the impact of the idea if it worked. How big? Is it worth your time? How does the impact compare to other ideas you have? How feasible is it compared to your other ideas? What is the estimated timeline to success? Will technology likely have dramatically changed by that future point in a way that undermines your potential success dramatically?


Step 3. The second fastest way to kill an idea is to find a study that has already done it or done something that informs you about the novelty and feasibility. I personally find the experience of reading literature that undermines my favorite idea aversive – but it is essential to plow through it. This is your essential second step. Build a folder on your computer, fill it full of papers that are relevant and then go through them and answer the following:

  1. Is this novel?
  2. Are there signs that this question has never been asked or tested, at least with modern methods?
  3. Is the answer already known and well established? Are there reasons to challenge the dogma?
  4. Assume others have the same idea, because they will. Now ask: “Do I have any unique advantages that will allow me to succeed?”. What are they? Make a list.


Step 4. Answer this: What is the simplest experiment I can do that would kill this idea? Then do it. I recommend talking with trusted colleagues to discuss this and looking at available resources – don’t build anything new if you can help it. Often colleagues will point out weaknesses in your idea, but if you have gone through step 3, then you can weight these comments in an informed manner. Don’t be discouraged by negative feedback, just filter as needed. Finally, go as far as possible with available data and computation. Wetlab work is often slow and hard – do that when it is essential to nail the idea down.


If your idea is still standing after these four steps, you have got the beginnings of your PhD thesis, postdoc or big new story. Now focus like a laser and get it done. Mock up figures every week, and constantly storyboard to work out the right study designs, controls and statistical methods. Storyboarding improves thinking. We do it all the time.

Most ideas will not survive this four-step process in their initial form, which means you need to work like a highly organized and intelligent maniac to kill the weak and select out the strong – gradually chiseling out a masterpiece along the way. The key is to ask (and attempt to answer) clear, fundamental questions as you go.


Chris Gregg, PhD.

Leave a Comment

twenty − fourteen =