The key distinction here is starting a company vs leading a company. To just walk in and lead an established company - you'd be fine with a better than average understanding, as long as you trust the experts in your company to do what is right.

To found a company? How do you know where to start? If you are hiring PhDs to come up with the idea for you, how do you if their ideas are novel or worthwhile?

Not to disparage SaaS founders, but I think it's somewhat easier to look about and say "there is no app for sending texts via REST API" and do that, than it is to look at say "no one has found a way to selectively target cancer cells for CRISPR" and then go do that.

I think the key distinction is that there is a well-established infrastructure around software, such that the average SaaS founder (even 10 years ago) is/was not doing fundamentally new things with computers. If they were, they would probably need to understand computer science well. Take any moonshot ML company- they need quite deep technical depth and usually have a founder that has spent time in academia.

We are still at a point in Biotech where companies have to do fundamentally new things with their tech to succeed. If you were starting a company that applies CRISPR to X genetic disease, you might not need a Ph D. to start the company, because you can apply existing technology to solve that problem. But that begs the question: Why has no one else done it yet?

At the end of the day, you need a unique insight to start a large company. That means experience with either the problem space or the technology space is helpful. Not discouraging OP either, if you have a vision for what the future should look like, and can acquire talent/capital... anyone can start a massive company. So OP should try to acquire that vision.

A big issue is its actually expensive to start a company that involves lab. Software can genuinely be one person in their bedroom with some AWS credit. I think the "why has no one else done this?" question is one of the worst anyone in the history of business or science can ask themselves. Any opportunity looks like that, it's why its an opportunity! The very best ones look like an obvious gaping chasm when you see them.

I'm in agreement with that sentiment. For any given opportunity, even if it's "obvious" that someone should have done it, doesn't mean it's been done before. If something seems like an opportunity you should probably try to validate it, instead of immediately discounting it. That's how I started my last company!

But there's no doubting that some sort of counter-intuitive or hidden insight is necessary. Even if that insight is: "people think this is obvious and not worth doing, but it actually IS worth doing." RE: Zero to One.

Founders generally don't start companies in vacuums. They are the result of communication with other people on the viability of ideas. They meet a subject matter expert in a field, they get to talking, they come up with an idea. The idea gets floated around to other subject matter experts. If something seems viable then prototypes are made to gauge market fit.

If what you're saying is true, there should be no reason why many of the most successful biotech management consultants do not come from a PhD background.

In biology, the problem space is extremely vast and very hard to understand. I have a PhD in biology and find computer science relatively logical and simple. You can just look things up. Looking things up for a specific topic is not even easy in biology.

Funny how you find comp science is simple and logic. Of course on a if else method level yes. But how to have a running system, dependencies, fail safe operations all coupled with tousand / millions lines of code is easily more complicated than a simply thesis about a mode of action of a protein.

I think we need to stop this useless debate as clearly one side has often no clue about the other. This behavior has led to massive failure in my career to successfully launch IVD devices or have a successfully bioinfo software for physicians - talking from experience here.

The point is that a computer system can be entirely understood from the chips on up. Sure, it gets complex with millions of LOC. But we humans designed computers and wrote the code. Biology is not like that, and there is lack of a completed understanding form the proteins on up. So you get the complexity amplified on all fronts.

Biological signals are very noisy. By contrast, CS signals are much cleaner. Only when CS delves into large volumes of social data do signals complicate. But even then, the cooked events of digital data is much cleaner than the raw signals inherent in bio-data. The complexity, interdependency, nonlinearity, and unknowns in bio variables are frequently overwhelming even for experts.

Success in biological research is driven by the scientist's ability to sort wheat from all this chaff, and it's acquired by gathering the data themselves hands-on in wet labs. A master biologist has learned how to navigate that space experimentally and analytically using techniques they've mastered just well enough to see over the noise.

As someone with degrees in both bio and CS and 15 years of work that crosses the boundary between them, I'm decidedly more in awe of those who have mastered biology.

What do you mean "can"? Nobody ever does, so I think it is valid to say that it "cannot". Otherwise, you could just as well say that a biological system can be understood from the atoms on up, since quantum theory provides a complete understanding.

I think that you and a lot of people in this thread don't understand the impossibility of understanding millions of lines of code. Do you really think "computer geeks" just hold them all in their heads, and then assume that since they can do that, it must not be difficult?

> Do you really think "computer geeks" just hold them all in their heads, and then assume that since they can do that, it must not be difficult?

No, but I think a group of geeks could. However a group of biologists cannot explain a biological system in full. When I say a computer system can be understood in full, I mean that human knowledge encompasses the working and programming of the machine, not that one person could know every detail. This is not the case for biology, as there's plenty we don't know.

> Otherwise, you could just as well say that a biological system can be understood from the atoms on up, since quantum theory provides a complete understanding.

It really doesn't provide a complete understanding for biology anymore than it does for sociology. It's not possible for humans to reductively explain such fields in terms of physics. No one can even prove this is possible. But it's not a complete understanding for physics either, since there's Relativity and questions about quantum gravity and dark energy.

"You can just look things up"

Not only is documentation pretty scarce these days, but good information on the interactions between pieces of software is much scarcer, due to the exponential number of ways that it can interact.

> Looking things up for a specific topic is not even easy in biology

How do you find info on bio topics right now?

Finding the actual info is not terribly difficult: Google Scholar and Pubmed, maybe with some of the arxivs.

The trick is interpreting it and putting it into context. A paper will report the results of one specific experiment, and it’s very rarely exactly what you want. Understanding how a result will generalize to other conditions is tricky, even for experts: there are tons of weird feedback loops, unusual dynamics, and other traps for the unwary (plus badly designed experiments and the occasional legit Type I error). For example, doubling the amount of a substance almost never doubles its effect, and in some cases, the effects aren’t even monotonic: ~75% alcohol, for example, is a much better disinfectant than 50 or 100%.

With time—-and lots of paper-reading, you do eventually develop a sense for what factors might matter and how you could check.

Depends on how deep you want to go. You need to find out what people are working on, and you do this by searching on PubMed and talking to people working in that field.

This sounds similar to what my friends in academia describe while exploring their fields. They build an impressive repository of knowledge regarding who's working on different topics/subfields over time. This may be one advantage of pursuing a PhD. Would be cool to see someone make the process of acclimating to a new field more accessible.

There are plenty of executives running biotech companies who have barely any biology expertise. Pretty much everything they know is contained to their product and its competitors. There are PHD's and MD's in SV who basically specialize in working as consultants for these types of products and back up their potential. You see it all the time with health startups when they have a medical advisor or something to that effect.

Doesn't give me much confidence in their product.

This is the central argument for pursuing a phd.

But its rubbish.

Speed + onus behind learning > everything.

Also note the self-serving bias where Phd's swear:

"You need a phd"