Another side benefit of buying your own spectro to create your own profiles is that you will be able to use 3rd party ink and paper (only the good quality ones of course). The money you'll save will more than make up the initial cost of the spectro, believe me.
Interesting point - but I always assumed that cheap inks also have drawbacks with regards to longevity, resistance to sunlight etc.
In any case, thank you all for the interesting answers and suggestions I sure have a lot to try and play with now!
I did manage to obtain the missing ink, so I continued a few tests. The comparison between prints from LR, PS, and EPP is really astonishing - after seeing this I really wonder what the point of an ICC profile for a specific paper/printer combo is if the software in the end does what it feels like anyway! (on a side note - for the Ilford 4x6 paper I am using, which is 310 micron, the ICC is GPSPP12, not 11.)
The picture I was using yesterday most of the time (dark portrait, brown hair and a moderately pale face) turned out too dark in all three programs. Differences are barely noticeable, details in the hair are lost because it gets too dark, and the face turns red. The print is basically useless.
I now tried a different pic, this time a bird sitting on a black PVC tube with a green/white bg (completely oof). Here, the difference between print and screen are considerably smaller, making it in the end really only an issue about adjusting the settings to match where the photo will ultimately hang. If I compare PS, EPP, and LR with this pic, I see interesting color tone differences, in particular on the PVC tube. Ideally this should be dark-grey'ish, and what PS gives me looks pretty close. In contrast, EPP produces something that is almost purple while LR is slightly on the greenish side. But the differences are so small that I would probably not see that unless I have the prints on top of each other. Someone with a better-trained eye might notice (and unfortunately I can not quantify any of what I just said). But it certainly is something I could live with.
The problem is, of course, that I also have the other type of photos, and that with the current settings I just know that some types of prints come out bad.
I guess what this comes down to is a confirmation of what some of you said - in order to get things right one has to do it manually, with or without the help of a spectrometer. Very disappointing.
Just for fun I tried something new - I printed something directly from my camera, using PictBridge. Much to my relief it did not appear to produce better results
(who uses that anyways?)