A page whose network never goes quiet

The document is here in full within a few milliseconds. What never arrives is the moment a browser calls the page loaded: a script below asks for a small file four times over every 100 milliseconds, each time under a new address so nothing is served from cache, and keeps that up for two minutes. Four at once rather than one after another, because a browser driven by Puppeteer counts the network as idle with two requests still in flight, and only a burst of more than that keeps resetting its clock. A crawler that waits for the network to go idle waits until its own patience runs out, and then has to decide whether the page it is holding counts.

It should. Everything on this page was there long before the first of those requests went out — the title, this text, and the link back to the index.