Archive for October, 2016

Date: October 3rd, 2016
Cate: development, Quadrilateral Cowboy

Work archeology, Part 3

This entry is a continuation of Work Archeology, Part 2

For a quick recap on what source control is, please read Work Archeology, Part 1

Quadrilateral Cowboy

Now that Quadrilateral Cowboy has shipped, I thought it would be interesting to take a look at its source control patterns, from start to finish.


The graph tracks how many source control check-ins were made per month.


There are a lot of peaks and valleys. My best guess for the January peaks is because that’s the time I say “well, this is the year I’ll ship.”

The January 2014 peak is a perfect storm of Aaron Melcher doing Mac porting work at the same time as Tynan Wales and I were prototyping the multi-person heist systems.

Quadrilateral Cowboy’s release date was July 25, 2016.


The graph tracks the total amount of source control check-ins per day.


The weekdays are all consistent, with Wednesday in a marginal lead.

In the latter half of the development I became more adamant about not working on weekends. I’ll get into that later in this writeup.


The graph tracks how many source control check-ins were made per hour.


5:00 pm is apparently the hot-bed of my source control activity. Afternoons are generally when I’m most productive, and I suppose 5:00 pm is the end-point of that period.

My best explanation for the midnight spike is I used to have a tendency to want to wrap up a given task before the end of the day. Nowadays I don’t often do that, as I see a lot of value in hitting the ground running on a partially-completed task in the morning.

2012 vs. 2016: Weekly

Let’s compare how the weekly check-ins changed between the first year of development vs. the final year of development.

(Drag the white slider bar left and right to compare)

There are some noticeable changes. In the final year, I set firmer rules for myself regarding weekend work. Also, at the office I share, we do a Friday show and tell event, hence the Friday spike to jam in something juicy to show everyone.

2012 vs. 2016: Hourly

A comparison of hourly check-ins between the first year and final year of development.

(Drag the white slider bar left and right to compare)

In the first year of development (2012) I was basically burning at full-speed all day long, from roughly 9am to 11pm on a daily basis. I guess you get things done, but you also completely wreck yourself and end up becoming a human mess. Now that I think about it, perhaps this explains the erratic pattern on the green monthly chart above.

As development progressed, I hit a point where I decided to change how I carried myself.

The final year of development (2016) shows a bit of that. I made noon lunch into a rigid drop-everything routine. I began trying harder to maintain reasonable work hours, from about 9am to 6pm.

Though, there’s still that midnight spike. For the life of me I can’t figure that out, as I sure don’t recall doing many all-nighters recently. Shrug!


The great thing about these records is that they cost nothing to make (well — I took some time to write the visualizer program that crunches all this data, but whatever). It’s basically ‘free’ data about your own work habits and the history of your project.

In my case, contrasting where I am now to where I started is definitely a trip.

I hope this gives you some good ideas of ways to automate the tracking of your own habits and hours.