Question - how hard would it be to allow players to manually generate the NPR's tech, in the same way you can choose to manually choose tech for a player race? A nice interface (both for NPR and player race) might be a dialog that puts up the possible random choices, and lets one push buttons to direct the research (rather than using the random number generator).
John
Absurdly easy. When Aurora asks if you would like it to design the race automatically say no, then say yes to all the suggestions except 'Assign tech randomly?' and after NPR generation go choose the tech you want. Aurora tracks the recommended amount of starting tech points and deducts the cost of each tech you instant research. Alternately, randomly generate teh race as usual and then ad a few choice technologies.
Either way, after the tech increases you will need to update the auto-designed ships to use the new tech.
I think I didn't make the question clear - I was asking if this was an opportunity to get a coarser-grained interface for non-random tech generation. Your workflow:
1) Assign tech randomly - "no"
2) Go to tech screen and pick individual techs
My thought:
1) Assign tech randomly - "no"
2) Go to new screen that has categories of tech, e.g. "engines", "missiles", "mining", etc. and select one of these. Aurora then figures out the next exact tech to select (possible with some randomization, e.g. whether to work focal length or wavelength when selecting "lasers"). This would make directed tech generation much quicker, especially for e.g. engines and armor (since the name changes at each TL, they jump around on the tech list and are a pain to find). In other words, one would never touch the "instant tech" button - instead one would click buttons on the "directed tech" page, which is more coarse-grained. Note that I'm not advocating taking the "instant" button away, just adding a new screen to allow more coarse-grained selections.
My expectation is that this should be easy, since the the screen in #2 could just be used to replace the RNG that selects among the categories.
If it is as easy as I think it is, then this could also be used in "regular" research. Instead of the player assigning a specific project (e.g. 2cm focal) to a planet's research, the player would assign a category, then Aurora would silently pick a particular project within that category and remember (but not display) it. When the research was done, an event would be generated. The player would only know how many research points had been accumulated, not how close to completion the next tech was. I'm pretty sure I've played other games that work this way - the idea is that what comes out of basic research shouldn't be too predictable. The granularity of the categories could even be one-to-one with the governor bonuses.
I was going to say that this idea would probably be too much work for Steve to do (mainly in prioritizing so that one tends to be at uniform levels within a category, e.g. aperture TL4 and focal length TL 4 rather than aperture TL6 and focal length TL1) for it to be worth it, but I suspect he's going to have to do something very similar any way for the NPR AI he discussed.
STEVE - in your prioritization weights within a category, you might want to give a higher priority to low-TL (i.e. cheap) projects than to high-TL projects (i.e. expensive). Something like a weight that was proportional to the ratio of the cost of the TL being researched to the first TL cost would probably work. In other words, if I were at TL1 in mining and TL4 in construction, I'd rather spend 18Kpts to bring mining up to TL4 than spend 20Kpts to bring construction to TL5; assuming TL1-->TL2 costs 3Kpts, then the weight for selecting construction would be (3kpts/20kpts) = 0.15 and the weight for selecting mining would be (3kpts/3kpts) = 1.0. If you don't do something like this, then the odds are high of having research on one high-tech project lock up a research facility for years. OTOH, from the point of view of variation there should be a bias towards a civ working on things it's already good at, so maybe you should use the sqrt of the ratio, e.g. sqrt(3kpts/20kpts) ~ 0.4.
Note that I would only do this for basic research. Component design (e.g. 20cm UV laser) should be explicitly scheduled, since you know what it is you're trying to design.
John