Samuel Bird

A heuristic of intensity and branching

I have a heuristic that has emerged over some days/weeks/months/years of study and work. It concerns the proper amount of intensity to apply to a set of actions, and how that relates to the nature of the action and the underlying goal. The core idea can I think be broken down rather a lot and generalized to a bunch of situations but, as is usually the case when that happens, it risks damaging the essence of the heuristic which is really found in it’s application. Therefore, I shall state it in the form most aligned with my intent. Fair warning, it likely makes the most sense in the context of my work as a Physicist, where it’s no doubt known intuitively by many. Furthermore, I write this to try and elucidate what is a very real concept in my head but one which I haven’t spent any time making legible, so let’s do that.

Heuristic of intensity and branching: If the goal is to move quickly, the intensity that is applied to a certain task ought to scale as the inverse of the number of decisions involved.

I’ll begin with a natural metaphor because that’s always fun. A research project is like a tree and on the end of one twig is a golden apple (we hope). The researcher is a hungry ant craving some apple flesh of the golden variety. On the ants journey, they walk along the trunk and then the branches and then the twigs, all the while only being able to see what is right in front of them and recall (hopefully, if the ant has a well organised set of notes) where they have been before. The ant is hungry, and so wishes to get to the apple as soon as possible. Naively the ant might hope to just run at top speed down different branches until they reach the golden apple, thereby travelling with maximal intensity at all times. The problem is that the ants senses are placed under strain due to the massive g-force, meaning the ants ability to pick the correct direction at branching points is afflicted. On the long stretches of straight and narrow, the ant should absolutely sprint; while the ant might find a less intense brisk stroll to be more productive if they spy an approaching huddle of branching points.

Ants and trees aside, the intuition is obvious. If I’m driving a car on a road, I should slow down on the corners where more care is needed to avoid flying off that road. I think where it stopped being obvious for me was in the application as usual. Specifically, knowing to look out for branching points in different contexts and being consciously aware that intensity is a knob that can and should be tuned.

Let’s now focus on the specific application of this heuristic to scientific research.

Why move quickly?

I don’t know if this really needs to be justified, but I think there are a whole bunch of reasons to want to move quickly in research. The main one for me is that it’s exciting. It’s a bit of a chicken and an egg situation: my intense excitement about a research question makes me want to move quickly and find an answer, which in turn refuels my excitement because I am having fun, am making progress, and am far from being bored. I claim this self-reinforcing excitement loop is actually a key component of healthy Morale. Moving quickly is an important component of my personal metric for day-to-day research performance.

What do I mean by intensity?

Intensity is basically the local instantaneous velocity of work when looking at the smallest scale of tasks. However, I do not mean just ‘hours worked’ or ‘power of concentration’, but rather both of those combined with a commitment to small scale technical work and a marked neglect of meta-work such as planning and brainstorming.

Importantly, the average intensity during a project is not necessarily correlated with how quickly the project was completed. To see that this is the case one only needs to look at the very enthusiastic undergraduate student who can easily sink 8 hours of hard work a day into something without making any concrete progress towards the goal, when their supervisor could complete it in a week with 1/10th of the total amount of cognitive effort. No doubt this difference is mostly because the more experienced researcher has developed years of intuition and knowhow to cut out the wasted effort along the way. However, the point is that the hyperactive undergraduate researcher doesn’t even know to stop of their own accord, take a breath, and just try to make better decisions. Most problems in school and university learning require very few consequential decisions, meaning that a student learns the default response to a desire to move quickly is to blindly ramp up intensity. It seems as though the goal of a research adviser is at least partially to help retrain problematic defaults picked up through education. I distinctly remember seeing this mindset infect lots of my fellow students back in the day, and unfortunately burnout was one of the common outcomes. I for one notice a strong correlation between the neglect of meta-work and the neglect of self-care.

Intensity can of course be really useful, and being able to switch it on is an invaluable skill. Some types of well-prescribed task, like solving school problems, coding up a well defined software project, or building IKEA furniture, are best done with as much intensity as possible, while being happy and not neglecting your wellbeing (remember Morale).

Argument

Returning to the heuristic, I think it is already fairly obvious to our intuition already, but let’s take a closer look at the claim that “the intensity that is applied to a certain task ought to scale as the inverse of the number of decisions involved”. One side of the argument is clear: if no decisions are necessary and the task is very well defined then we can totally get through it as quickly as possible by ramping up intensity. On the flip-side, if we ramp up intensity too high and run into some consequential decisions the cost can be high. Working with high intensity means that we are less likely to allocate proper planning and thought to a decision. In practice, this means just blunder-bussing through a decision onto the first available path. Going onto the wrong path temporarily isn’t the end of the world - it happens rather a lot in fact, it’s part of research. But if it’s a habitual problem that we go onto the wrong path many times more than we would in an alternate universe where we spot at least some of the decisions, slow down, and actually compare different options, then we risk increasing our total journey time significantly. If we were to lower the intensity, we would move faster on the global timescale of the project. These ideas are intrinsically rather nebulous and I work from a very simplistic model of research projects and decision making, but it is just a heuristic after all.

Application

Having explained what I mean by moving quickly (good) and intensity (not necessarily the same as moving quickly, and my arguments for why intensity should be tuned down when faced with lots of consequential decisions, I now want to look at specific examples of types of tasks involved in scientific research (of my flavour) and where they fit into this picture. I’ll do this by expounding my own image of how I go about research and where I use this heuristic.

Speaking very broadly, I think earlier on in a research project there are more consequential decisions. It is easy to overload the intensity and waste a load of time before the project is even properly formulated. It’s very important to not overload intensity until the basic questions have been answered e.g.

Once the preliminary questions have been answered and there is a clear vision and set of goals, I find the intermediate stage can feel like a rollercoaster ride consisting of 3 different substages that loop: formulating hypotheses and hypotheses tests, testing hypotheses, and, albeit less often, pivoting (changing key components or assumptions of the project, or rewriting goals).

To be clear, I didn’t always view it like this. Sometimes it has felt like one big homogenous stream of slurry, but this is something that came about naturally as I started doing more research and thinking about the research process on a higher level.

Formulating useful hypotheses that give us as much new information as possible and push the project forward is something that involves a lot of decisions. The same is true when we want to design the best tests for those hypotheses. There is an infinite space of possible directions to pursue, so narrowing down the right one deserves lots of attention, careful thought, and patience. This can be extended to include other types of task that involve lots of explicit decision making e.g. planning how to implement some code. Overloading intensity here will undoubtedly lead to some rash decisions, suboptimal hypotheses, and general calamity.

Testing these hypotheses, on the other hand, can usually be done with a fairly high level of intensity once we have one in mind. We know what we want to test, we’ve designed a specific test, and now we go full steam ahead and get it done. The trick here is having the self-awareness to know when you’re done with the test and that it might be time to tune down the intensity once more.

Pivots are another beast altogether. By their very nature they are unpredictable. In my experience they typically come from meetings with collaborators because the individual is so ensconced in a series of hypotheses and experiments that they develop a blind spot around certain simple alternative approaches. Interaction with collaborators helps to facilitate the necessary meta-work to face decisions which a high level of intensity might cause the individual to neglect. That being said, I find that pivots can sometimes be drawn from within by deliberately looking for them at regular intervals. In this event, it is important that one is careful and approaches this with a lower level of intensity to ensure a useful pivot is made. I would argue that the making the choice to try and hunt for pivots requires this as a precondition though.

At the very least I look at this heuristic as a reminder that it is not only okay to ramp down the intensity at times, but actually desirable. Now that I think about it, this is largely a refinement of a part of my thoughts about farming and fruit-picking applied to the question of how to best move quickly, which itself came arose from the topic of Morale.