- So my philosophy in helping younger scientists get started in their research, is to carefully design their experiments, but be bold enough to actually try the experiment. And especially the pilot experiments that lead in to the main experiment.

- Piloting an experiment, not only will it help you sort of visualize how you're going to execute an experiment, if you're designing a new technique, but it will also inform how large your sample size is going to be. So, if you go into a new experiment, and you do it as if you're going to publish it, so you go like publication quality right from the very beginning, you're going to want to have a pretty decently large sample size. You're going to be very vigorous with how you take your observations, and if something simple in the experiment fails or doesn't work properly, then it's going to a lot of effort that has just been wasted.

- What a pilot is effectively, is a reduced version of the full experiment that you would like to do. So, there are a lot reasons why one may not want to go straight for the optimal experiment, but rather, like do pilot experiments. That includes, first of all, just developing the technical expertise of doing the experiment. It may be desirable to do a simpler version of a more complicated experiment, just to learn how to execute it technically.

- The first time you do an experiment, you need to convince yourself that you can do the experiment. And that you can reproduce what somebody else has done and get the same results.

- By doing pilot experiments, where you keep it small scale and you make small modifications, it will save you time and it'll save you resources down the road.

- One example would be, some of the viruses that we grow, these normally infect organisms like humans. And, the way that we grow these viruses is in something called tissue culture. So you take cells from a human, or from a frog, or whatever, and you grow them in the laboratory on their own. And basically this provides food for those types of viruses. So here's the tricky part, when you want to design a very big experiment in tissue culture, these cells are waiting to be infected by virus, but as an experimenter, you have to have a whole batch of cells that are just waiting for the next day so that they can be used for virus food. So it kind of doubles the amount of work you do. And if you underestimate how much time it's going to take you to culture up those cells on the side, you can suddenly launch into a very big experiment, and realize that you don't even have enough time in your day to create the virus food for the next day. So what we would do is, we sit down, we design an experiment, and we run a pilot of it for 24 or 48 hours to get this feel for what is it like to go through the day-to-day handling of this experiment. And to make sure that we're prepared with enough media that is used to grow the cells. Just even enough plastics that are used to grow the cells on them. So that we're prepared day in and down out. And there have definitely been times where we've learned through pilot experiments that you have to kind of balance the ambitions of what you want to do in an experiment with the practicalities of, you know, you have to have enough material and enough time to prepare that material as an experimenter. Otherwise, there's just no way to do the experiment.

- So there are a variety of reasons why one would want to go and do pilot experiments at the beginning. Another reason is that there may be things in that experiment that you need to troubleshoot, and may be better to start off from that frame of mind. A third reason is that, you may not know the perfect conditions or variables for that experiment. And it may be sensible to do pilot experiments, to optimize those conditions, learn about them before you design, you know, that kind of perfect experiment that you want to test in the end.

- So a pilot experiment gives you the opportunity to test your analysis that you intend to do later on. And to just get this familiarity with the data and how variable are they. And in some ways, it's really the variability of the data are going to dictate, do you need a lot of replicates in order to get that strong signal? Or, you might convince yourself that you can afford to do few meaningful replicates and you'll still have a strong enough signal to do a statistical analysis on your data set.

- So, if you are trying to design a new experimental approach, you might come up with a question like, how am I going to know when I can transition out of doing pilot experiments and start doing like my real publication quality experiments? Right? And, on some level it's a bit empirical, it depends, it's going to vary from case to case, and it might take a little bit of intuition on your part. But, there's a couple of rules of thumb that you can keep track of to know when you can make that transition. When you're doing pilot experiments, you should try to keep the overall sample size pretty small, so that you can demo something very quickly and you get an answer immediately, and then alter it in the next round quickly. For me, I found that every pilot setup I was using, was fixing some small issue in the setup, and at some point, I eventually got to the point where I didn't have anymore issues with the setup. So I was able to carry out one or two biological samples pretty consistently, pretty easily, and I felt like it was something that I could scale. 'Cause the real point with transitioning from pilot experiment to your publication quality experiment, is scaling up your sample size. 'Cause you're going to need a certain sample size in your real experiment to actually have some confidence in your data. So, within your pilot setup, if you have something that you feel you can scale, that's one sign that it's time to transition to actually doing the real experiment. Another thing you can keep in mind when you're doing pilot experiments is, you should try to have some kind of control that's in the pilot setup where it's going to tell you whether it's working properly or not. So, in my case, when I was doing the agar sandwich, most of my pilot experiments, literally had two blocks of agar media that had the same composition. And, that was kind of the control condition, because I knew that if the composition was the same on both sides, then I should have the same number of route branches on either direction. So, when my pilot setups, they were easy to do, and I was also getting that expected result, so my controls were actually working the way I expected, that is another sign that it's time to start transitioning to getting out of the pilot phase and into the final experiment phase.

- So, a lot of, you know, experiments in biology are made up by developing technical expertise, troubleshooting, and optimizing conditions. No one gets there in one go. So patience and also designing simple experiments that you can complete, and that will be informative. Learning something from each experiment, even if it's a small piece of knowledge, that allows you to better design the next experiment. And from that, better design the next experiment to ultimately get you to the most information rich experiment that, maybe the one that you multiply out to do. That is effectively the path of most experimental biology.