[V8.9] Inconsistent Research Speed

[BlackObsidian]

Again not sure if this is a bug but struck me as more noticeable in this build than the past:

When starting two early research projects (think it was alenium and alien alloys) I allocated 30 scientists to one, which went immediately to 'excellent' progress, and 8 to the other, which only went to 'poor' progress. Despite the difference in scientist numbers the one with 8 scientists progressed through 'good' to 'excellent' and finished pretty much the same time as the one with four times the researchers.

Hoes does research progress work? Is it a linear process or are there diminishing returns from adding more scientists? Either way, showing progress as 'excellent' when the project is still weeks from completion is a bit misleading.

[Amiga4ever]

I think its depend what you are researching. Excellent doesnt mean that every project will take the same time. Same was in UFO EU. Correct me if Im wrong

[Matthew]

Let me take a look at this.

But I think it's the "as designed behavior". In X-com Enemy Unknown you had to wait days or weeks (late game) for a research to finish. If you assigned more scientists to it, it had better progress and you could skip some steps of the research (poor -> good), if you assigned several scientists to it, you had to wait more time, and you could observe the process, as the research reached new stages poor -> average -> good -> excellent.

[BlackObsidian]

Precisely, but assuming two early research projects take roughly the same amount of research 'effort' (if not then ignore this whole thread) then assigning lots of scientists to one so it jumps to 'excellent' should mean it finishes way before the other project with 25% of the scientists.

No?

[Zinn]

It's probably just the case that each thing has a different default time to research, with one scientist attached to it. As more scientists are attached, research time is reduced, though probably with diminishing returns (eg. scientist number one has 100% efficiency, scientist number two has 95%, scientist number three has 90% and so on).

20 scientists on each project might push it into "excellent" territory because it's based on the time remaining, possibly calculated as a percentage of the original value with 1 scientist, which will be reduced by each additional scientist attached. Let's say that "excellent" shows up when there is 20% of the default research time remaining: In that case, a project with a default time of 40 days would take 8 days to complete, whereas a project with a default of 50 days would take 10 days to complete - Both would still show up as "Excellent", since you have essentially skipped 80% of the research compared to the default setting with a single scientist, but the "real" time meassured in days varies.

I don't know if it works exactly like that but some sort of system as this is usually implemented in games, so you can't win them by hiring 200 scientists and research everything in a single day.

Also, a system with diminishing returns works in a way that lets you get away with less scientists on some projects, as a single scientist eats up more time (meassured in days, not percentage) on longer a project than a single scientist on a shorter project.

Edited January 26, 2012 by Zinn

[Matthew]

Thank you Zinn, I've tried to explain this, but your explanation is far more better than mine. I'm speechless

[Amiga4ever]

Is it possible to include information how many hrs/days left to finish research? I think in UFO EU was like that

[Chris]

Zinn's explanation is most likely. Matthew, it might be worth running a few tests on research projects that take the same amount of time to complete though, just to check the system is functioning correctly.

The game does feature diminishing returns but the curve isn't particularly punishing. I think each scientist is 1% cumulatively less effective than the scientist before them.

[BlackObsidian]

Thanks Chris - I guess the question that will settle it is:

'do the early projects each take a sufficiently similar research to complete that the diminishing returns effect is minimal?'

If not then using more than say 20 scientists on a single project is pointless, as the 'value added' by each extra scientist does not shorten the research time by a meaningful amount.

Zinn's explanation make sense, although each project must have some kind of relative value associated with it so that end-game projects take longer than early-game projects (Chris - can you comment on how long each of the early projects should take with 1, 10, 20, etc scientists?). Therefore with a flat-ish diminishing returns curve, allocating 20 scientists to one project and 10 to another should mean the project with 20 finishes first but perhaps not in half of the time with 10 scientists.

If we can know the relative times/research effort values of the projects in the current build it should be easy to test whether the diminishing returns effect is working properly.

[Ziggarius]

Is it possible to include information how many hrs/days left to finish research? I think in UFO EU was like that

X-Com was not like this. It never told you how many days to expect the research at all. In fact, based off of experimentation it was found out that research time was slightly randomized even. That is, on average a research could take say 400 man-days. It could take 200 man-days in one game, and the next game it'd take 600 man-days. You just wouldn't know until you played and researched it.

Link here: http://www.ufopaedia.org/index.php?title=Research_Technical_Details