04/21/23

How To Think Of An Idea

all matter is just the recombination of old matter

This essay is written in collaboration with one of my best friends, Aadit Shah. Aadit is a medical student on leave from Stanford, working on company creation at The Column Group.

Aadit’s entire job right now is to think of cool new ideas for life sciences company creation. As we were chatting about his work, our discussion naturally transitioned into how to generate innovative ideas. This essay is the essence of that conversation. 

Paul Graham being PG has already published his thoughts on this. He says:

The way to get new ideas is to notice anomalies: what seems strange, or missing, or broken? You can see anomalies in everyday life (much of standup comedy is based on this), but the best place to look for them is at the frontiers of knowledge.

Knowledge grows fractally. From a distance its edges look smooth, but when you learn enough to get close to one, you'll notice it's full of gaps. These gaps will seem obvious; it will seem inexplicable that no one has tried x or wondered about y. In the best case, exploring such gaps yields whole new fractal buds.

It is good to have problem-focused or use-case-driven ideas but for gen Z building in 2023, the following sentiment illustrates how we feel toward innovation: 

“We are the middle children of history. Born too late to explore earth, born too early to explore space.”

People often feel like innovation comes from unfounded leaps, but perhaps in reality innovation is about permutation. I mean, look at all these generative AI companies building in different verticals on openAI’s platform. 

Innovation is predicated on taking disparate fields and perspectives, and putting them together in unique ways. There’s a concept in US copyright law of derivative work vs. transformative work. A derivative work may be copyrightable, but it infringes on the copyright of the original work. Merely moving a work into a different medium or changing the language in which the work is expressed is derivative. A transformative work is copyrightable and does not infringe on the copyright of the original work. 

Innovation today in those “gaps within fractals” feels a bit like transformative work. 

Another analogy comes from patent law. When patent officers do an assessment, they look for three things: novelty, inventive steps, and application. What inventive step essentially asks for is “did this application put forth at least one step or process that is non-obvious to an expert in the field”.

What it is saying, in some ways, is that you need to take a lens disparate from the canonical view of the field to innovate something. This is somewhat ingrained in our very own patent literature – innovation is legally defined as permutations of novel concepts from different fields. 

Now that we have the abstract idea down, here are some actionable steps we do to “come up” with ideas: 

• We both read a ton. We read expansive literature and we read it broadly, not with a particular intention in mind. This is just to get a pulse of what is out there, and to sample the space.

• Once we have an idea or a concept, we conduct iterative steps of “what is the most uncertain area and what is the experiment we can do to de-risk that most uncertain area?”

• Continue identifying and experimenting the most uncertain steps at all times

When we’re looking at innovation, what we’re doing is taking an investment in an open field – a broad question – that cannot be answered in a binary way with one experiment, or even a set of experiments. All we’re trying to do is iteratively take the most uncertain question, the highest area of risk, and collapsing that, trying to find the answer to that with an experiment, over and over again, with the hope that we have de-risked the concept so we have a viable idea at the end of the day. 

This essay is 21/50 of my 50 days of learning. Subscribe to hear about new posts.

More on Aadit: 

Aadit Shah joined The Column Group as a Swanson Fellow in 2022 while in training for his MD at Stanford University. Aadit holds a BS in Biomedical Engineering from Washington University and an MPhil in Bioscience Enterprise from the University of Cambridge. While completing these degrees, he conducted 4 years of immunology/infectious disease research that resulted in publications in Nature and Cell. He also led the development of a non-mydriatic retinal imager through to acquisition. While at Cambridge and Stanford, he gained exposure to operating and investing experience in therapeutics (particularly immuno-oncology and genetic medicines) with prior work spanning Flagship Pioneering, Tessera Therapeutics, and 5AM Ventures. As an advocate for the larger student innovation community, Aadit oversaw the national network of biomedical incubators: Sling Health. As National Network President, he supported more than 1000 students across 15 institutions and reported outcomes twice in Nature Biotechnology.