After a conversation with @tabatkins, it seems like we may be able to loosen the algorithm for an acceptable anchor.

Currently, given

<div style="position: relative; anchor-name: --a">
	<div style="position: absolute; position-anchor: --a; position-area: start"></div>
</div>

the anchor creates a containing block that is an ancestor of the positioned element, which fails the existing algorithm at ... positioned el’s containing block is an ancestor in the flat tree to that of possible anchor’s containing block.

However, this works:

<div style="position: relative">
	<div style="position: absolute; anchor-name: --a; inset: 0"></div>
	<div style="position: absolute; position-anchor: --a; position-area: start"></div>
</div>

In both cases, the containing block of the anchor has identical dimensions, but because of the DOM order of the elements that generate the containing blocks, the second works while the first fails. Is there a reason why the first couldn't work as well?

In other words, could a positioned el's containing block be generated by the anchor? Could we change the algorithm to read positioned el’s containing block is not a descendant in the flat tree to that of possible anchor’s containing block?