Taco Bell Programming (2010)

Taco Bell programming argues for solving problems by recombining a small, proven set of Unix tools rather than building new code or adopting heavyweight distributed frameworks. Functionality is framed as an asset and code as a liability, so developers who know enough about operations and Unix can often deliver simple, scalable solutions without overengineering. Concrete examples include using xargs and wget to download millions of web pages instead of writing a custom Clojure crawler on EC2, and using find piped to xargs -P32 to process files in parallel rather than deploying Hadoop MapReduce. The point is that a few lines of shell and trusted utilities often meet requirements with far less risk and operational burden.

The approach rejects trends that encourage unnecessary complexity - unit-tested ops tooling or third‑party message queues - on the grounds that each added component increases failure modes. Practical trust in battle-tested tools like xargs and syslog replaces faith in bespoke multithreaded code or heavyweight services. The author recounts implementing a SOAP-like service with static files and mod_rewrite and admits to choosing Python where sed could have sufficed, illustrating the learning curve toward “Unix Zen.” The bottom-line appeal is capitalist and pragmatic: minimize risk, keep pagers quiet, and get the job done with minimal, well-understood tooling.