1- The current research tree would be broken down into all of it's parts. Instead of getting weaponsmithing you would get one weapon at a time. For some things it would make sense to do it the old way, like for shard shrines. Having access to air shards but not earth shards would be just awful. The same for suits of armor. But for most things breaking it down into pieces should work better with the other features.

Agree. This increases the possibilities for how a more interesting technology tree might look, the possibilities for the player to pick his\her own way, and the replayability.

2- The research tree that is unresearched is hidden from the player. You can look at the stuff you have uncovered and how they relate to each other, but not down the line and see how close you are to the doom hammer.

Agree, it would mke it much more interesting.

3- Research as a resource is removed or highly modified. Instead everything will generate proportionally more gildar, and you need to fund researches with this money.

I'm skeptical to this, but it depends on how the funding will be done. There should be some kind of automation on the funding, and it shouldn't be possible to save up lots of gildar to suddenly rush the research with 10 new technologies in no-time.
Current research-boosting buildings should make up a maximum research rate possible each round, to force the player to invest long-time resources into research. Not only gildar.
However, you could make it possible to pay X gildar to boost the research by Y% of normal rate (so there will be a normal research rate, and a max research rate).

4- Instead of selecting a research to work towards, every round you have a % chance of a discovery. And when you have a discovery, one is selected from a weighted list of all the next in line research for you. The weighting would take into account several things. Every other encountered faction, their completed research, and their attitude towards you. Research being in the world makes it easier to figure out. Technology treaties would offer a significant advantage to this bonus. Your faction bonuses, warrior, mage, civics, etc. And additional resources that you funnel into research. For example if you have a lot of iron, spend some on research, your chance to get a discovery that uses iron, or is in some way related to iron would go up significantly. In this way discoveries are weighted towards allies research, and the resources you actually have available. So an alliance might end up looking similar equipment wise, and if they had a ton of crystal they would be coming at you with some serious enchantments. Also every research building would improve discovery chances.

I would love this!

5- Conclaves would become more special. A conclave when built will prompt the player with a message from the researchers in the city. 3 or 4 windows will pop up and that many different researchers will ask for grant money for a specific research. One peasant looking one might ask for 50 gildar and some of the cities food to try and make better farms. A scholarly looking guy might show up saying he has an idea on how to tap into the elemental shards for greater power, and a military person might ask for some gold and iron to try and make a better club. After selecting one a bonus to the chance to get that as a discovery increases drastically. But related discoveries would go up too, and a few random unrelated ones. The peasant might not find how to farm better, but he figured out how to revive the land with magic. The warrior did not make a better club, he found out how to imbue iron with a hint of magic though. The scholar had no idea what he was doing, but found out that a byproduct of crystal refining is an AMAZING fertilizer. It is a random system, but it still takes in player choices that effect it significantly.

I'm not sure what you are thinking about the gildar funding here.
Is the payment a one-time fee? What if the player can't afford that preferred specialization right then, could s\he change funding program later?

6- And finally... I for one really want more things to research! I would want to make many more weapons. More and different daggers, enchanted ones too. A few new armors, with enchanted variants. Spesificaly a new light armor, brigandine, that is better than leather, more expensive and does not need a perk to wear. I also think that the perks you can use to make units should have more, and be researchable. Also cross discipline dependencies. No more making enchanted armor without knowing how to make armor. I just want the research tree to be more interwoven, and a joy to explore.

Agree. I have also been dreaming of some kind of semi-random item generation in the game, maybe this could be implemented here too?

Interesting ideas... not sure some of what you envisioned can be actually accomplished within the current system setup.

A hidden research tree from the player and it shows what techs you've uncovered thus far. I think you can set it up that you cannot research anything directly on the tree and you discover particular techs by a questing system, but not sure how this would pan out. (especially with the AI) - I've got some ideas on how to implement this, especially if you just want weapon variants that is completely possible, but not sure how the AI would be able to handle the effects or even obtain the ability to use these things.

There are other things you can do with the tech tree with randomness. You can actually set it up to where each game gives you a distinct research tree, one where you do not have access to the same research each time and it will show up with a particular probability. Not sure if you noticed this, but this actually is already in the game.

Good luck on your endeavor here.

The research King of the Wastes has a 25% chance of appearing on your tech tree. At the start of each game you do not know whether or not this tech will be in your tree, but some times this tech does indeed, appear. In FE, they had 4 magic books that would appear with a 25% chance, I think this got changed for FE: LH, but the effect was the same.

I created a mod for FE, that brought to a faction summoner units, units that may summon a particular elemental for free on the battlefield. But you needed to research one of these books. So the odds of anyone training these summoner units was small, let alone training all 4 varieties. It set up a unique dynamic of creating a rather powerful mage type unit with a small chance that it would appear on the players choices. I thought that balanced out the unit as much as anything else.

This does not do what you really want to do and have the techs hidden from view, but it will provide each gameplay an experience that says, here is your randomly selected collection of "extra" techs that you can research.

The tag that operates this appearance chance is

<AppearanceChance>25</AppearanceChance> --- 25% chance of this tech showing up.

I figured, this would be the least invasive way to implement a random tech tree with the current setup of the game.

Actually I just learned XML from modding this game. Most of what is around these files are pretty self-explanatory. The xml groups around tags the game uses to determine the effect. I think of them as mostly definition of variables for programming. For example, <Description>BLAH BLAH BLAH</Description> That really is a string for the variable description. </Description> is the ending tag to tell the program that it is done reading this particular variable. There are no set definitions for any particular XML file and the tags used in this game are going to be different than say another game. So it is learning what the developers are calling these particular variables.

I recommend that you go ahead and try reading through some of the easier files and you'll start to pick up on the formatting. Make some guesses as to what that tag is trying to do. Like the above, <AppearanceChance> That could mean a lot of things, but I took a guess on that it is the chance that this particular tech will appear. If you want to add things into the game start small and work your way up to the bigger things.

I don't know of a good crash course in XML would be, somebody that actually learned it formally might be able to point you to the correct direction.

I'm probably what one would call a feral programmer. My only programming / computer class was taken 13+ years ago and it was an intro to C++ course. So I decided to pick up programming recently.

We used to have a 'hidden' research tree in E:WOM, and there was a random function assigned to tech appearance if you met the prereqs (I think this was essentially ignored though).  Several asked for a visible tree structure, hence what you see today.

Put me in the 'I like having a visible tech tree in this game' camp.  However, I do agree with the idea of breaking down techs further.  The difficulty falls within the game structure.  Some things are beyond the realm of .xml modding, so when Stardock removes a feature, it may not be able to be recovered.

I wouldn't mind seeing a revisit of E:WOM with the current engine and 'lessons learned' at some point, now that E:LH is releaased (E:WOM II), but I'm not holding my breath on that one.  But if you want to downgrade your playing experience a bit, the E:WOM engine should be able to accompish a good amount of what you are suggesting here.