I owe you an explanation.

Between October 17 and November 1, Safety Intel missed eight days in your inbox. Not because there was nothing to send. The issues were written, researched and sourced on schedule — they went up on the website and never went out as email. If you were reading along, you had no way to know they existed.

The cause was boring and entirely mine. The last step of publishing each morning ran through a single machine in my office. When that machine was asleep or signed out of beehiiv, the issue got built, staged, and then stopped one click short. I was traveling. Nobody was there to click.

What you missed

All nine are live on the site now, dated to the days they belong to. If you only read one, make it Donora or Sandoz.

The part I am not going to dodge

This newsletter runs on one idea: failures come from systems, not from the person standing closest when it breaks.

Then my own publishing chain turned out to have exactly what I write about every morning. One undocumented single point of failure. No alarm when it failed. No backup for the one step that actually mattered. And the monitoring I had built to catch this told me the website looked fine — because that is all it was ever watching. It was never watching the part that reaches you.

Latent conditions, all of them, drilled into the system long before the week I happened to be away.

I have moved the send off that machine. Publishing no longer depends on one laptop being awake and signed in, which is the specific hole that put eight days of silence in your inbox.

Thanks for still being here.

— Rob

P.S. If your own operation has a step that only works when one particular person is at their desk, that is the same hole. Go find it before it finds you.