Make Sent Me to 1969.

I tried to automate a weekly events email. The useful discovery was that the process needed fixing before the automation did.

By Lisa Ramirez · The Sensible Plan

I tried to automate a weekly events email.

Reasonable idea. The calendar already had the events. The email needed the events. Copying the same information from one place to another every week seemed exactly the sort of job a machine could have.

Make disagreed.

It cheerfully decided every event happened on December 31, 1969.

Dates around the beginning of 1970 are usually technology’s slightly passive-aggressive way of saying a date field is empty or has been interpreted badly.

So I found the missing field, fixed the mapping and could have called it done.

Except the useful problem was underneath it.

The wrong question

The weekly process was: read the calendar, copy the information, reshape it for the email, check it, publish it, then do roughly the same thing next week.

I had started with How do we do this faster?

The calendar was already the source of truth. The sensible answer was to make the information entered there usable elsewhere.

Automation can preserve nonsense beautifully

You can automate a broken process perfectly. It is still broken. It simply becomes faster and rather harder to notice.

Sometimes the task should be automated. Sometimes the task should stop existing. Those are different problems.

What AI helped with

AI helped troubleshoot the date issue, map the fields and work through how the systems should pass information between them.

Useful. But it did not decide whether the weekly copy-and-paste relay made sense.

Before automating the work, make sure the work deserves to exist.