I’ve been trying to understand these two species via iNaturalist observations, and at this point I’ve at least been looking at them long enough to have an opinion. Here’s the couplet distinguishing them in the Jepson eFlora:

45. Leaves generally 6–8 mm wide; perianth parts 9–15 mm, long-acuminate . . . . Allium falcifolium
45′ Leaves < 6 mm wide; perianth parts 6–10 mm, obtuse to acute . . . . Allium siskiyouense (among others)

The Flora of North America key is similar, leaving leaf width out in favor of the somewhat unhelpful “tepal apex […] inner margins usually denticulate” vs. “tepal apex […] inner margins denticulate or not”. (And, no, I don’t know which side of the tepal apex is “inner”; I think “margins of the inner tepals” is meant, if not written.) For distinguishing these two—and leaving the “others” aside—this leaves room for improvement. The problem is that Allium falcifolium has considerable variation in tepal shape, even in areas where it is reportedly the only species present and confusion with Allium siskiyouense should not be an issue. Some plants have lanceolate tepals with exceptionally long-acuminate apices. The tepals in such plants are erect in the proximal half or so, forming a narrow tube, and widely spreading the distal half or so. Some plants have urceolate flowers with lanceolate tepals that have acuminate apices, only moderately spreading for a short distance. Most plants are somewhere between those two extremes. Tepal color is also variable. The majority of plants have purple tepals, as those linked in the prior sentences, ranging from dark purple to somewhat lighter and often more reddish purples. However, they can also be various shades of pink or lavender, and sometimes quite pale. They can be white, sometimes with darker midribs. Rarely, they may even be red, or whatever color this is.

Allium falcifolium is also the more widespread and frequently observed of the two, by a wide margin (I’ve got about a 20:1 ratio between the two in iNaturalist observations). It’s reasonable to look somewhere in all that variation for a distinction between “long acuminate” and “obtuse to acute” tepals. And, indeed, there are plants with long-acuminate tepals, plants with obtuse tepals—often, these are plants with the tepals acuminate but not dramatically so, the tepals gradually narrowing for some distance before, I guess, giving up—and plants with acute tepals. However, while it is easy to find plants that are on either side of the line, if you try to systematically separate them into two categories, you’ll eventually give the task up as hopeless—but not before noticing that no matter where you put that line, plants on the “siskiyouense” side of it extend throughout the entire range of Allium falcifolium and far from the reported range of Allium sikiyouense. I suspect most people have tried to assign the name “Allium siskiyouense” to plants on the more obtuse / acute end of this spectrum within and extending some distance beyond the reported range of Allium siskiyouense, before deciding at some distance that the plants must all be Allium falcifolium no matter what the tepals are doing.

All that said, this is what I think Allium siskiyouense actually looks like. Flowers narrowly campanulate, tepals narrowly elliptic, apex obtuse to broadly acute, usually pink with somewhat darker midribs, occasionally purple, occasionally white. At least, plants that look like this are the only way I can end up with a circumscription of Allium sikiyouense that is consistent with the keys and descriptions in the Jepson eFlora and Flora of North America, is consistent with the distribution maps of those two resources, and yields two taxa that are more or less reliably identifiable (there are still exceptions and problems, but these do not appear to be optional features). So, here’s a couplet that may work:

Tepals lanceolate to linear-lanceolate, the apices acuminate (sometimes dramatically long-acuminate), sharply acute, or both; margins of the inner tepals usually denticulate; flowers urceolate to salverform, usually somewhat constricted around the height of the anthers; leaves generally 6–8 mm wide . . . . Allium falcifolium
Tepals narrowly elliptic, the apices obtuse to broadly acute; margins of the inner tepals usually not denticulate; flowers narrowly campanulate, not at all constricted; leaves < 6 mm wide . . . . Allium siskiyouense

I can’t measure leaves in iNaturalist observations, but they do appear to be wider in Allium falcifolium, at least on average. I can make out denticulate margins on the inner tepals of a lot of Allium falcifolium observations and only a couple of Allium siskiyouense observations, but this isn’t a feature that tends to show up well in photographs.

You may be wondering at this point—”OK, but with all the variation in Allium falcifolium, is that really a species?” I don’t know. At present, I’m not able to see patterns in the mess of variation that would allow it to be divided into coherent taxa. I would not be at all surprised to learn that such patterns exist, though.