August 9, 2021

Questions

Last week, I half-jokingly added the tweet below to my list of "thesis ideas" that are more a collection of jokes than actual ideas for my thesis:

"Search first, ask later: turns out every question has already been asked"

— u-haul key expert (@yoo_hoo_yoo) August 3, 2021

I say half-jokingly because to a certain extent, I feel like this is true. At every internship I've had, one could usually find the answer to what they're looking for by popping a query into Slack or Google. And yet, hundreds of questions are asked each day that are either identical to, or just minor permutations of things that have been asked before.

I think in software development, one's expectations is that you get paid to write code. Seems like a reasonable thing, right? However, the more I think about it, the reason why developers get paid the Big Bucks™ isn't for their code; any person can write code¹ by looking at examples and pattern matching. They get paid the big bucks for their ability (or lack thereof) to traverse large information spaces, and find answers. Large information spaces can be the sea of ambiguities when it comes to designing or laying out the architecture for new systems, and they can also be the space of possbile solutions to a given problem (what does this error message mean?).

The thing is, traversing these information spaces is really shitty. Sure, Google Search is a thing, but you still have to sift through results, find one that sort of looks like it works, and then pattern match it to your particular problem. Sure, Slack conversations are a good source of prior problems and solutions, but a developer has to actually think about using these tools, anyway.

It's much easier to just shoot off a question into a Slack channel or a forum and get an answer, rather than sift through a huge amount of information. I wonder if there's work to be done for a tool that preempts developer questions, does a search in the backend, and then presents a couple that match what a developer could be working on.