I made a post, but it was lost and now all is answered by Thom already

The only thing Thom did not mention (of course) was my idea on this failing to research. Even if you could not fail (i.e., a correct combination shows the spell), there are many combinations to test. To give an idea, I did some computations. The time needed is for if you research one combination per second (impossible already).
With one glyph in each way, there are 1956 combinations, taking you 32 minutes.
With two glyphs in each way, there are over 1.3 billion combinations, taking roughly 41 years.
With 40 glyphs (the number in the art I think), there are about 2.2*10
48 combinations, or roughly 1.5*10
31 times the lifetime of the earth.
I would think spell research is something that would help me to find correct combinations, and not make it even more impossible to do so. For example, you could put in a number of glyphs, and depending on your level, it might reorder glyphs towards a working spell. If not, spells can only be learned from cheat sheets based on database knowledge, as even a whole guild could not find combinations by just trying.
I am sure in the future there will be other ways to learn spells, but at the moment, I think this random failing (and yes you have to click 10s of times sometimes for even basic things) does not make sense.