What Makes a Project "Successful"?

Last night, I was musing about personal projects I undertook in 2017, and I realized that of the two big ones that I made, I didn't know if I could judge either as a success.

The first one, TiddlyWiki Full Text Search, took me a little while to make and it has a little adoption, but I wrote it primarily for myself and I use it pretty sparingly.

The second, Twitter Reply Tracker, got a really good response - it appears that people find it useful - but I don't use it myself that much. It's nice to make something for others, but maintaining it can quickly become a chore.

One project that I definitely judge as a failure is something I wrote back in 2014 - I called it Devel::Trace::Syscall. It took me a month to write, and while I learned a few interesting things, I never use it. If my objective had been to learn something, I think I could've finished learning well before a month.

This is in contrast to a small project I made this year: one time on my way back home from a visit to Chicago, I was listening to some music and decided that I wanted to have a full history of songs I'd listened to on Spotify. Spotify provides a recent history API, but it's only the last fifty songs you listened to. So when I got home, I whipped up a little AWS Lambda function to get the most recently played songs via the Spotify API and insert them into a DyanamoDB table. It took maybe 15 minutes, but I query that table pretty regularly to jog my memory about a song I'd listened to.

Another project that stands in contrast is something I wrote in 2016 but have yet to publish - I call it Twitter Archivist. Basically, it's a scrappy Perl script that grabs what you see on your Twitter timeline and stuffs it all into a SQLite database. That way, if I see something on Twitter and think of it a week later, I can look it up. I run the script once a week, and I find myself querying the database all the time! The only reason I haven't published it is that I'm not terribly proud of the code.

One interesting thread that ties these two together is that they have practically zero user interface - I use the AWS CLI with jq to query the played songs table, and I just use the sqlite3 command line client for my Twitter archive, with one or two custom SQLite extensions to make some of the queries easier. They both took a small amount of time to write - I had each of them finished within an evening - and they both use a very small amount of code.

As life goes on, time becomes more and more a premium resource, so the skill I'm working on cultivating this year is knowing which projects to pursue and which to not. So, going forward, I'm going to try and employ some discipline when taking on a new project, in the form of a checklist:

• Am I trying to make something or learn something?
• Is this something I will really, actually use?
• Is there something else that does what I need, or most of what I need?
• What is the least amount of work I need to do to get this done?

It gives me hope in a weird way - I sometimes feel lke I'm drowning in ideas for projects, but now I feel like I can write off 90% of them! Here's to a year of successful projects!