So you're saying you enjoyed the massive item sink that was TC3 research?
I wouldn't say I liked it necessarily, but it never really bothered me. And I don't recall reading many complaints about it before TC4 came out either. Most of the complaints and frustrations that I saw people post about was having to throw out research note after research note because they were all giving the same useless research instead of the one that the player actually wanted to focus on. But hey, that's just me. If people really found the item sink in TC3 to be onerous I'm not going to argue with them.
Anyhow, my real question to you is: what suggestion do you have to improve the current state of TC research? And no, the TC3 item sink is not a valid answer.
First, I'd remove the aspect points system. It's ok to have to scan successive items to figure out what the item is made of. It's not ok when you can only do this once and trying to do research eats away at the meager amount of research you gained from scanning that item. I am aware that there are mechanics to aleviate this, but there aren't early enough in the tech tree to be useful to a new player just trying stuff out for the first time. As it is, the system basically encourages players to spend weeks doing nothing but scanning and never touching the research table until they've basically built up a backlog of scanned research. That's not good. Ideally people should be scanning at about the same rate that they do research, and doing research almost from the very first item they make.
Second, less RNG. Impossible research is rage-inducing. Not knowing what research you'll get next is not as bad, but still annoying. Plus, it discourages newer players who don't have access to a cheat sheet to tell them what the probability of picking up the research they actually want is.
Third, plan out the sequence of aspect scanning better. It's unforgivable to make scanning paper dependent on humanus. Common items should be easier to scan than rarer items.
My solution? Keep the rune-linking mini game, but make it less dependent on RNG. Make it so that learning research requires connecting runes in a specific sequence for each research. Put the sequence in the Thaumonicon entry for that research, but only show parts of it until you've researched every item in the tree leading up to it. The idea is that instead of the current system where the first aspect you scan creates a specific research, with this you won't actually be 'locked in' to any research until you finish the sequence. If you run into a block, you can clear the sequence and try again for a different research. This would take a great deal of the frustration out of failed research because you can just go down another path, and it doesn't feel like you've wasted anything.
So the first item you research (like wands or something) will be basically given to you, but still need research as a basic tutorial on how to connect the dots and play the minigame. You will be able to use any aspect that you've previously scanned when playing the mini game, but aspects you don't know will be unusable. Thus, as you progress down the research tree, you will ocassionally run into spots where you can't move on because you are missing an aspect. This will probably be the part that needs the most work, because it will have to planned so that how to get the missing aspect is hinted at somewhere, and players going down the research tree hit these stopping points somewhat evenly, and not clustered all in one spot.