I've reviewed a couple of Informant products lately, with one thumbs up (the book The Greatest Tournaments 2001-2009) and one thumbs down (The Small Encyclopedia of Chess Openings 2). This third product, which culls the endgame sections from Informants 5-99, gets a thumbs-up for two positive reviews out of three, but I do have a significant criticism, too.
Let's start with a summary of the contents, followed by a discussion of its strong points. There are 2494 endgame fragments, sorted in chronological order. (Within each Informant, the endings are ordered by material: pawn endings, then minor piece endings, then rook endings and finally queen endings.) The fragments are all annotated, generally by the winner. (In the early days it was generally by Informant staffers, but over time the players themselves took this over more and more, and by the early 80s it was rarely done by the staff.)
The collection makes for some excellent training material, both for oneself or for one's (stronger) students. All the main database programs allow one to search by material, so it's easy to create subsets when one wants to look at king and pawn endings, rook and pawn endings with pawns on one side, both sides, etc. All wonderful, and I think this makes the collection useful and worth getting for players around 1800-1900 and up looking for practical training material.
Here's the one downside, and while it's not quite fatal it is a big deal. The Informant staff clearly hasn't lifted a finger to check ANYTHING in the era of strong computers and tablebases, so all the material through at least the early 90s should be checked by trainers before assigning it to one's students. (If you're trying to solve things yourself, well, be sure to run the engines afterward to make sure that the solution really was a solution.)
Of course even in the pre-computer era, the analyses are still generally reliable. But chess is a complicated game, and I was able to find errors here and there with the engine without much trouble. So buyers beware: the material has not been checked, but I do think that even with that limitation it's a useful product. It would be foolish to reject chess books that haven't been computer checked - that would mean throwing out practically everything written before the 1990s. The criticism here is of the Informant staff for not doing any new work, choosing instead to simply cull all the old material and resell it as-is.