Most AI publications are feeds: the newest post on top, everything older sinking. That suits news, and news is already well covered. It fails anyone who arrives with the question where does this actually stand right now?
So YLEM is built the other way round. The core of the site is a set of living field guides, one per area of the frontier, revised in place with a visible revision history. This log is where smaller observations land first. When one changes the picture, the relevant field guide is updated and the revision is recorded.
The Lab is where claims get tested. Every report states what was tested, which version, in what environment, and a verdict: adopt, trial, watch or avoid.