Is there any particular theory about why some games attributed to Dosaku are not geunine? Are those games suspected to have been played around the same era as Dosaku's but by somebody else, or are they more recent?

There's a small handful of games in Chinese sources that, as far as I know, have never been acknowledged in Japanese sources except for one where the number of moves differs. It may just be that the Japanese collectors haven't been aware of these games. Or they may regard them as spurious.

There's a couple of games in the Japanese corpus where alternatives exist as regards the players. The most interesting is a relatively recent find where we know only that it features two of three named players but whether Dosaku is one of them is uncertain. He is given simply as Sensei, which suggests to me he was viewing ex cathedra, but some people think he was actually playing.

Then of course there is the Dosaku game invented for the 2012 film about Shibukawa Shunkai.

Just for fun I checked whether I could add "doubtful" to some of my Dosaku games. But it seems most of my sources give the same 153 games. Also GoGoD (Winter 2015) has the same 153 games. I have not seen anybody express doubt about any of these games. Of course there is uncertainty about the details.
Let me compare my collection with the GoGoD collection. A conspicuous date difference is GoGoD "Genroku ?15 II 13" where I have "Genroku 4-2-13" (that is, 1702? vs 1691). In 15 cases the moves differ other than just by rotation/reflection. Usually the difference is by 1 or 2 moves only, and often the GoGoD version has a Game Comment stating that other versions exist. In two cases (046.sgf, 049.sgf) I have 25 and 20 additional moves and in three further cases (013.sgf, 058.sgf, 059.sgf) there are many differences (while the first 150+ moves are identical). Probably the former two were meant by John with "not acknowledged in Japanese sources".

Updated the collection at http://homepages.cwi.nl/~aeb/go/games/index.html.
Today the count passed 43000. New are the games from the five recent AlphaGo videos. Completed and corrected the Castle Go collection - maybe this is the most accurate version available today as far as the moves are concerned. The metadata still needs work.

Today I encountered a Go Seigen game that was new to me. It is not in the Winter 2015 GoGoD collection.

Updated the collection at http://homepages.cwi.nl/~aeb/go/games/index.html.
It has a bit over 46000 games. New are the 1880 games from Hoensha, and todays Samsung final. Two old Cho Chikun games that I had not seen before (games with the winner of the 20th Kansai Ki-in 1st Place title, game1, game2, he lost both).

Never seen these two games before. There are a collection of Sanyo special games between the Kansai Kiin 1st Place winners and top Nihon Kiin players. These must be from the same set.

Yes.

I encountered two new Cho Chikun games from 1973, see https://homepages.cwi.nl/~aeb/go/games/games/Shinei/05/index.html. The collection of his games now contains 2300+ games.

And another one from 1976.
And two more from 1975. And two more from 1974. And two more.
It turns out that it is easy to find unknown Cho Chikun games. I'll stop mentioning them.