Allen Thrasher put the following message on Indology (14.6.2017): "Jim Nye of the University of Chicago Library told me that when Anandasrama did a second edition of the same work it would start from scratch, so that another ASS edition of the same title will be a different text." I have meanwhile found that this appears to be true of its editions of the Sarvadarśanasaṃgraha.
I have inspected two Ānandāśrama editions of the Sarvadarśanasaṃgraha:
(1) one published in 1950;
(2) one published in 1977.
Both are no. 51 in the series called Ānandāśramasaṃskṛtagranthāvali, and both also include Madhusūdana Sarasvatī's Prasthānabheda. According to the title pages, edition (1) is the third impression (aṅkanāvṛtti), edition (2) the fourth. Edition (1) has been prepared (saṃśodhita) by Vināyaka Gaṇeśa Āpaṭe with the help of the Pandits of the Ānandāśrama; edition (2) only by the Pandits of the Ānandāśrama. The text of the Sarvadarśanasaṃgraha covers 171 pages in edition (1); 174 pages in edition (2). Both editions contain the same list of manuscripts (ādarśapustakollekhapatrikā; edition (2) contains its first mistake in this list, misnaming ms. gha as ba) and the same list of chapters (edition (1) calls it atha sarvadarśanasaṃgrahāntargatadarśanānām anukramaḥ; edition (2) calls is anukramaṇikā). Edition (1) then has a preface to the second impression (dvitīyāvṛttisaṃbandhi nivedanam) and an introduction (upodghāta); edition (2) has neither of these two.
The two editions are not identical. For example, edition (1) has, on p. 128 l. 15: prasajyeteti cet na; edition (2) (p. 130 l. 15) has just prasajyeta. Numerous other examples could no doubt be added.
Interestingly, the editors of Erich Frauwallner's Nachgelassene Werke II (Wien 1992) thought that Frauwallner had used the Ānandāśrama edition of 1977 (abbreviation: SDS1) for his translation of parts of the chapter on Śaṅkara's philosophy (along with Abhyankar's edition), which is of course impossible (Frauwallner died in 1974). This sometimes leads to confusion, as when the editors point out on p. 200 fn. 108 that the last pāda of a śloka in the 1977 Ānandāśrama edition (p. 152 l. 22) "erscheint ... nur verstümmelt"; this is true, but this pāda is perfectly in order in the 1950 edition (p. 150 l. 12).