In Pacific theater of operation II, they used a system which worked fine because the difference between tech levels of each opponent was very low (since it stayed in a WWII setting).

The way it worked is that you have different fields (ex: aviation, electronics, industry etc). Every month, you spend money in each field.

Each field has a rating, if you spend enough money in each field, your rating will raise. If your don't spend enough money, your rating will drop.

When your rating reach a certain level, you acquire a new technology. Most of the time, a technology requires 2 different field to be at a certain level so that focussing on 1 field does not give your anything.

You don't know which field gives what, so you try to use your intuition on where is the best place to invest.

So you could do something like this, You hide the information regarding which field develop what, you add and remove randomly tech that can be researched and you could even change the required values.

This system also force the players to always keep a steady investment to progress, else, if they get some problems (ex war) and cannot invest, they will be late and they will have to catch up before progressing again.

Now about infinite tech. I don't see much how it can be done. There is 2 options:

Cumulative bonus: You increase the bonus received over and over again which has the problem unbalancing the game near the end.

Comparative bonus: You get a bonus if your are better  than your opponent. So you need to constantly progress to keep and edge over your opponent, but if you are 1st place in tech, you always gain the same bonus.

hmm. then what about basically the same system, only with the branching visible one level inwards (basically you know which tech you are researching, but what tech they are connected to is the thing which will be randomized)

would that work better? in this way you can see what you are researching, how much it will cost/how long it will take, you just don't know exactly what it will lead to, even if you play alot of times, due to random techs, random situational techs, and having existing techs of tier X randomly pick which ones will be connected to researching tech A from tier X-1. So researching animal husbandry might lead to Herding Techniques and Wild animal taming, or it might lead to Advanced Domestication and Slaughter Houses. And a random situational tech, it could lead to bare-back riding (on horses) which would lead more quickly to saddle tech and stirrups tech, or it could lead to War-horse training and War-horse breeding. As a random tech, however, it could lead to Wild-Boar riding, or Deer-riding, or Bear-riding .... to add a little mixed variety.

Equally basic farming could lead to crop-rotation, or it could lead to Harvest Seasons, or it could lead to Pumpkin farming, or it could lead to Rice Paddies, ect.

In any case, random or not, with one line visible, you can make more of a choice of what to do with the knowledge you are given with. Still, it would be kinda nice to research without knowing exactly what will happen. I suppose the second idea basically achieves this concept although places it a step deeper, so that you can still make at least immediate-time decisions.

Here's a thought. How about a Research Tree where you start off seeing EVERY Possible Tech. As you mouse over a tech a info box will pop up and tell you what you need to reach that tech. You can then literally build your own Tech Tree each game by deciding what you want to go for. Researching some techs could close off the path to others so you would have to be careful what you go for. This also would make it useful to be able to trade techs. You could trade a tech you have for one that got closed off when you researched something else.

The only Techs you wouldn't be able to see from the start would be random techs and situational techs. Everything else you'd get to pick and choose from.

well, tech visibility has been in every game I have played thus far, and It eventually lends itself to the same path over and over again. What about if different techs took better advantage of certain techs somehow?

I still like situational and random techs being numerous ... and I don't see how a large pool of random situational techs could tie into a completely visible tech tree, except for some side-branch which gives you "iron clad ships" or something equally outside of the normal tech tree. I suppose this relates to an early idea Civ 4 had of having monolithic governments like in Civ 3, and for technology progression to not be under decision of the player, but to be mainly influenced by the type of government, and several other factors. Eventually it was decided that it would not be as fun for the player, even though it was fun for the designer (as I remember it being said) ....

With the 1 step randomization, each tier will still lead into the tier above it, only its not certain which type of tech it leads into. I mean ... I suppose i would be okay with a completely visible tech tree ... but I was very excited with the idea of not having to min/max the research strategy. However, that does lend to the possibility of the RNG making an even worse job of it.

So I will no longer argue for random technology development, but I would like for situational and hidden techs to be unlocked, which unlock their own miniature tech tree for dealing with a certain style of warfare, or a certain terrain.

I also would like for Research to be made unique and separate from economy ... with a raw research stat and % modifiers (education) to affect the research stat ... as opposed to throwing pure "money" at the research, you have to build up experiement/research based infrastructure in order to get the raw data for technological development.

Not sure how spell-research is going to be handled, although Im sure it will be more Shard-related as well as what the Channeler decides to meditate upon. I can see spell progression as either building your own spell-book or to be slightly random as Stardock has suggested. But again with random, I mean more of an organic probability, alas saying random too many times and it becomes a sin, har har har.

Raven X's idea to "buid your own tech tree" before the game even started was an interesting one ... but it seems like it would be the ultimate in min/max and strat crafting, although I suppose it would make good strats easier to learn for new-comers.

I agree that linear tech trees need to go. I'd rather go with something branching that lets you focus on some things and neglect others and still be successful.

Having read this thread, I notice that the posters' opinions can be divided into certain mutually exclusive categories:

1. One should choose what to research based on one's situation. 
2. One should only be able to research things that make sense in one's situation.
3. One should research according to one's strategy.
4. One should adapt one's strategy to what you researched.
5. One should be allowed to choose techs according to the story that you are role-playing.

There might be some others that I missed.

These categories appeal to different kinds of players, so which one to choose would depend on which players the tech research system is meant to appeal to.

I think that to make the choice, one should consider the 4 X'es and the 4 player types.

One way to assign them would be:

1. Exterminate - Conquerers; like to be aggressive
   
2. Explore - Discoverers; like to be smart
   
3. Exploit - Socializers; like to negotiate and/or make stories
   
4. Expand - Collectors; like getting rich
   

If I have matched them up correctly, it is clear that the technology tree should be made to appeal to the Discoverer type.

Some of the concepts mentioned in the thread are:

1. Randomize the tech tree - see Sword of the Stars, or Master of Orion I - socializers and conquerers don't like this
   
2. Make certain optimal paths/combos to be discovered by smart players - gets old quick, so not so good for discoverers
   
3. Make the tech tree depend on your situation - Good for conquerers and collectors, bad for socializers
4. Seperate science and technology - Realistic, but doesn't really improve the game much, except for some socializers
5. Use education system to favour small empires - Potentially good for everybody, if balanced correctly
   

I recommend that the various ideas being suggested should be evaluated in accordance with these game design principles.

EDIT: Clarified a bit and removed some incorrect stuff.

That's not really a bad thing. It's what rational players will do. If some techs are always better then others, of course anybody who knows what they're doing is going to go for the best ones first.

You're trying to address it by the symptom (smart players always go for the best techs first) instead of the problem (one set of techs is always better then all the others).

Exactly. What really fixes this is when the answer "what is the best tech to research right now?" has more then one answer. Then there's no problem. Some of these other ideas are just an attempt to hide it by removing control from the player so they can't make rational decisions, and instead relying on prayer to the RNG. (Not everything, I don't have an issue with branching or interconnected trees. Mostly I dislike taking control away from the player in order to try and stop what a skilled player would naturally do.)

well, I definitely thing certain techs should disable researching certain other techs. Some-one can't be the best at everything, just have their own flavor towards everything, and further research increases their flavor towards the extreme.

Well, I don't think a nation should develop Apartheid or Master-race technologies, and then go back and research Tolerance technologies without having to make a choice ... for instance if you research those tolerance techs you will have to abandon all previous ideologies and require some penalty pre-req, like creating a National Wonder/Epic of apology, which requires many productive resources to build. And then you can finally research Tolerance and go about the other tech tree of human development/culture. And also you could not use traits/abilities of both, building the epic would block traits and abilities of the previous (hate based) tree branches.

I really like aspects of this idea as well as reply #51 above. And reply 27 (my idea).

THe thing that bothers me about this idea is that production equals science progress. There's no trade off of investing into early tech or early expansion because they're one and the same.

well, it would be regrettable if either production or economy were to equal scientific progress. I would rather science be some-how unique and separate ... although Im not currently sure as to the best way to go about this. Normally, the amount of commerce/money a nation has (in these such games) the more they can research. Sure a good economy could lend to an easier set-up of research, but I would rather avoid a slider to devote X% to gold and X% to science, rather I would want science to be its own, separate key investment, where science and gold are not interchangable.

This could help to balance out relative "catagories" of research, assuming you have a limited amount of space in which to place a limited number of buildings, so decisions need to be made.

In this way, it would essentially be impossible to be the best at everything ... Although I think there should be some system where the larger a city is, the more its buildings count towards this randomness ... (or rather larger proportionally speaking), so that a small nation with a 20,000 capital can have the same level of specialization (and effectiveness) as a nation with a 60,000 capital and several 20,000 cities.