Why We Should Let the Pantheon Crack

nautil.us

13 points·gbaygon··4 comments

Actually is worth a read. A couple of choice quotes:

“There’s no greater definition of success for a building than it’s been standing for 20 centuries.”

It's basically a discussion of understanding (or not) old systems, patching old implementations,and how the quest for the new can (doesn't have to) lead you astray.

Or it's a definition of luck. Survivorship bias.

An analogy would be to notice that a COBOL mainframe app wasn't RESTful, and building a VM and dependency injection framework for COBOL that would make up for this lack (rather than sandboxing the app in an environment that could have RESTful interfaces it's boundaries).

Missing from the article is any discussion of survivorship bias.

At least some of the resilience of these monuments is due to the fact that we are only studying the extant examples, which simply haven't collapsed, and so, tautologically, exhibit positive qualities that prevent (or avoid) collapse.

This isn't random, of course, so the longer a building has survived the longer it is likely to continue surviving.

We don't see around us the many buildings, built using the very same principles, that did collapse.

Certainly there is much to be learned by studying the survivors, but giving all of the credit to the skill of the designers and the principles they used seems like an attribution error.

Reply on news.ycombinator.com