Archaeologists in Alaska just pulled 80,000 specimens out of one site, and they're still refusing to call the best feature they found a "dwelling."

The site is called Sheg' Xdaltth'i', about 13,850 years old. Stone tools, animal bones, charcoal, eight distinct features including hearths and what looks exactly like a shelter. Diamond-shaped tools link it to sites in Siberia. Everything points toward a real settlement people kept coming back to.

And the team still won't put the label on it. Not until there's enough evidence that it isn't just a reasonable-looking guess.

I wish more site audits worked that way. Most "that's probably just legacy code nobody uses" calls get made on a five-minute read of the page, not eighty thousand data points. The guess might even be right. It's just still a guess wearing a confident label.

What's the oldest unexplained decision in a project you inherited that you refused to touch until you actually understood why it was there?