Apis Mellifera

Self-Optimization

Roughly 470 days ago, I started a project a project called Færeld, a command line time tracking tool. Since then, I've used it to log 709 entries for a total of 666 hours and 57 minutes of 'productive' work (if you're curious as to the actual data produced by Færeld, I host the hord on Github). As I'm coming up on 500 days of accumulated data, I thought it would be useful to reflect on my intentions behind the project, the way it has affected my self-development, as well as what advice I may be able to give to anyone who wants to do the same thing.

Proving To Myself

Færeld started as a tool to accumulate data about the 'productive' activities I fill my free time with. 'Productive activities' here are loosely defined as things that contribute to my self-improvement, whether that be reading, bouldering or doing programming. It does not necessarily mean activities that produce something tangible, either for me or for anyone else. 'Useful' may be another way of describing these activities, but there are many useful things in life that don't directly contribute to my own self-improvement, like sleeping.

A large reason of why I chose to categorise and track these is because my desire to keep improving, to keep working on projects and bettering myself along set lines, is vital to my own self-esteem. To keep learning and creating things produces a feeling of movement and without this I begin to feel stagnant, which easily spirals out into feelings of worthlessness and hopelessness. I love making things, but part of this love is also a mental health strategy to keep the wolves from the door.

Færeld, then, functions as an objective record of my progress, as a weapon against feelings of stagnation. I can wield it against my inner voices in order to prove to myself that I am creating and learning and doing.

Færeld was really valuable in this regard because it makes visible what before was not. Before this, the only objective way to prove to myself that I was making stuff was to look at the stuff I had made. But, not everything that is in the process of being created actually gets completed, many books don't get finished. Certain activities, like bouldering, don't even have an end product at all, and so can't be measured in this way. Is the time spent on these things wasted because they don't result in something fully formed?

Faereld makes the effort expended in any project/activity, no matter it's state or telos, visible. The movement of self-development is not found in the end products, but in the process itself. Færeld exposes to myself the concrete metrics behind the many processes of self-development I'm engaged in. It lets me prove to myself that I'm moving.

Productive Homeostasis

The continuous use of Færeld lets it function as a kind of feedback-loop, a way of letting my choices of what to work on and for how long meander towards where I want them to be. My preferences, whims and goals will change over time, but this constant feedback mechanism provides a natural (and almost unconscious way) of influencing my activities to align with those preferences.

This effect of Færeld was half-intentional. I had expected that I would look at the summaries produced by Færeld on an irregular basis and use these in order to inform my planning of what to work on. I hadn't expected that I would do this at very regular intervals, and that it would become part of my everyday workflow. Before I settle down to work on a project or read, I quickly check my Færeld summary. I don't even look at it that thoroughly, just to look at the graphs and the last few things recorded. Weaving this into my method of working means I don't have to plan how long to spend on something, I just let the summary inform my gut feeling of how long and what to work on. It's less of a conscious alignment, and more of a continuous and semi-conscious meandering towards a kind of productive homeostasis.

Marking Boundaries

One unintended side effect of using Færeld is that it greatly helps me fight against context switching and losing focus. When I first began using the tracker, I would do something and then enter it into Færeld after the fact. This proved to be clunky and inaccurate, and I would sometimes forget to track the time altogether. I solved this by allowing now as a valid input for the from/to time settings and would track the task as I started it.

Tacking the input of a task to when I started it had an unexpected effect of clearly marking in my mind the boundaries between tasks. Before, I was easily distracted and would often flit between 1 task and another. The declarative act of starting to track a task as I begun it made explicit my intention to work on this specific task. The explicit marking of the start/end of a task meant that I focused on it, and only it, for an intended period. The intentionality of declaring that I would just be working on this specific thing actually made it easier to do. I think this is the most beneficial side-effect of this kind of continuous time-tracking, it has forced me to really focus on what I'm doing and only what I'm doing. I like to think I'm pretty good at context switching, but you always pay a time-efficiency price when doing so.

Substances and Attributes

One major pitfall when it comes to metrification and optimisation of activities will be one familiar to anyone who has worked in a workplace where enforced metrification has been deployed in order to assess employee productivity: metrics often do not capture the substance of an activity, just an attribute, or several attributes. Subsequently, you end up optimising (willingly or not) for those attributes, and not the activity itself.

An example of this would be measuring the productivity of programmers via the number of lines of code written, the naive assumption being that the more code they're writing, the more productive they're being. This method of measuring productivity completely ignores modes like efficiency of code or refactoring (which would, in good circumstances, produce a negative delta in code lines). What programmers would end up optimising for is what the metric tracks: lines of code written. This results in programmers writing code for the sake of it, eschewing efficiency or elegance for verbosity, not through personal fault but because their job status now depends on this metric.

In the context of Færeld, this pitfall takes effect for the fact that I am only tracking time spent, with no ability to measure how effective that the time spent was. I can spend 1 hour faffing about programming on a project, or I can spend 20 minutes approaching the same problem in a more planned, structured manner. Given the way Færeld measures things, the former activity would look better than the latter. One could perhaps counter this by creating goals & tracking how long it took to achieve those goals - but again, there would be an incentive to game this by creating smaller and smaller goals in order to boost both the number of goals hit and reduce the amount of time-per-goal.

There are probably cleverer ways around this problem, but since this metrification is my own choice, done in my own system for my own benefit, I get around it by just being honest with myself. This tool measures only a mode of how I spend my productive time, not its substance, and if I feel like if the relationship between those 2 is drifting then I notice it and correct it myself. It is of no benefit to me to waste time unproductively, I would be gaming no one but myself. Being able to game your own sense of perception can be useful and beneficial, but after a while becomes noticeable, and another problem to fix.

Summary Mode

Over the past nearly-500 days, these 4 points have been the most consequential things that I came to realise during my development and use of time tracking.. For anyone who wants to pursue this, either through using Færeld, other tools or by making their own, I would emphasise the following:

  • Be aware and honest with yourself about why you want to do this. If it is to produce cool looking graphs in the long term, focus first on accumulating the data. If you're looking for more short term effects, think carefully about the structure of your interactions with your tracker, and how they might produce desirable behaviours.
  • Incorporate feedback loops. Emphasize regular feedback from your tracker and incorporate it into your workflow. Anything from summary graphs to intelligent, predictive suggestions based on past time spent is valuable.
  • Don't worry about capturing everything about your activities in the tracker. Just capture the most valuable attributes, like time or goals accomplished, and try to be aware of what shortcomings or un-optimizations might lead from what attributes you end up focusing on.