The Research Engine

[Reepicheep]

One thing I haven't seen referenced yet is the nature of the research engine in regards to its randomness. The research results in SC1 can be quite frustrating (and game-skewing) due to sheer randomness; and I hope that's improved in SC2- especially since research can be a lot of fun.

There are a couple additional research options I would like to see. One would would be a way to toggle randomess in research (i.e., if selected, research would take a specific number of turns to complete, with no random factor- and that could even be a sliding scale so folks could pick the level of randomness they want). Another would be a 'historical' toggle, in which nations would automatically get advances at the time they did so historically.

[Exel]

There was already a thread about the very same subject. I also posted a solution suggestion there, but here goes again (to avoid the debate from starting all over again from scratch):

Getting tech advances by research should depend on luck, but by investing more resources and directing them to the desired branch the probability of gaining the tech significantly increases. Also, each turn you get a negative roll, ie. don't get the tech, the chances of getting it the next turn increase. Now, lucky ones can still get techs with low investment and quickly, but even the most unlucky ones will with utmost probability get the techs eventually, as their chances increase every turn. This prevents the tech race from being an expensive and decisive lottery, but still retains the random element to spice up the experience.

As for the random effect for research results I would suggest the following. Whenever the research calculations give you a positive roll, ie. you get a tech, there is a chance that the tech gained will be something else but the defined goal of research. Meaning that occasionally you'll get other techs than you have been aiming for, ie. anti-air research might produce you better anti-tank weapons (as was the case with the German 88mm FlaK 41 for example).

[ April 17, 2004, 04:32 PM: Message edited by: Exel ]

[KDG]

Expanding on what Exel said regarding research advancement. This was include in that other thread.

This would solve the problem and wouldn't be too difficult to program.

For each turn that you don't get a tech in a research that you have investments, you gain a 1/2% bonus chance per chit.

Example:

4 chits in Jets. Everyone at 0 tech level.

1st turn 20%(zero bonus per chit)

2nd turn 22%(.5% bonus per chit)

3rd turn 24%(1% bonus per chit)

10th turn 38%(4.5% bonus per chit)

20th turn 58%(9.5% bonus per chit)

After you get the advancement, everything is reset back to the 1st level.

I'd also make this an option, which could be turned on or off.

[Shaka of Carthage]

As was pointed out in the other thread... R&D is random. It doesn't matter how much money is spent or what you direct it towards. The method SC uses works fine and doesn't need any changes to reduce that "randomness".

Having research follow a "historical" path doesn't make sense, since as soon as you make a few moves, you are no longer following a "historical" path for the nation you control. Being able to set the tech differences per scenario is already possible. And for those who don't want any random research, don't use R&D.

Random tech advances (along with random political readiness) is one of the keys to the replayability of SC. Its why some of us play SC daily year after year. How many other games can you say that about?

My only suggestion is that once a tech advance is made, you lose a R&D chit in that tech. SC is awash in MPPs. There are too many MPPs out there chasing units and tech.

[Reepicheep]

Given that we had a historical option for nation entry in SC1, I don't think a historical option for tech advances in SC2 is unreasonable.

And certainly SC1's tech engine is not very realistic. Once I won jet tech advances for three turns in a row as Russia in the first three turns after they entered. In a current game I did not get level one jets as the UK until Barbarossa- but now have gotten four advances in very short order, putting me ahead of the Germans- who have been ahead of me for much of the game. I enjoy technology trees; but I find that sort of thing to be ridiculous and frustrating- it shouldn't even be possible. In WWII, you're not going to have three major practically applied advances in the same technology three months or so in a row; it's like saying the Germans could have developed and introduced the Tiger one month, the Panther the next, and the King Tiger the month after that- all from the same research stream, not independent streams!

I think that for each level of tech investment there ought to be a minimum length of time required to get an advance, and a maximum. Once past the minimum, there would be a small chance for an early advance. That chance would increase as time ticked down to the maximum. There could still be significant swings in technology that way; but if the limits were well set, the overall results would be quite a bit more realistic, and more fun.

[ April 18, 2004, 01:15 PM: Message edited by: Reepicheep ]

[KDG]

Hmmm, hadn't thought of minimum time. If we do put some sort of maximum on how long research could take, then correspondingly we need to put a minimun amount of time.

[pzgndr]

but if the limits were well set, the overall results would be quite a bit more realistic, and more fun.

Now how can having the limits "well set" for each and every game be "more fun" after x number of games? This is a game. Think replayability.

If the randomness truly bugs you, it is now possible to create a no-research game in SC2. Or edit unit values and research to create something less extreme, like having default values start at L3 and only allowing tech advances to L4 (2%) and L5 (1%). Whatever.

Bottom line is that you will have more options and flexibility with SC2 than you can shake a stick at. :cool:

[Reepicheep]

Now how can having the limits "well set" for each and every game be "more fun" after x number of games? This is a game. Think replayability.

Fair question! In general, playing with tech is more fun than playing with no tech. And having realistic tech timing is more fun than unrealistic tech timing. Thus my point....

To demonstrate the concept, let's say the minimum for one chit invested would be four months (three a year- which may be too much), and the maximum would be a year and a half. Stretch out the probabilities along that line- 1% to 100%. You're guaranteed a return in a year and a half- but could get one after a few months. With such a system, you can still have large swings in randomness, thus replayability; but there will be nothing extremely out of place if properly set up (as I mentioned). (And I don't see how going years with no advances- and then getting them all at once- actually helps replayability anyhow. )

Knowing you're eventually guaranteed a return, however, may encourage players to spread out their chits, instead of stacking them in the most necessary technologies in hopes of getting reasonable results. Thus I suspect what I have proposed actually would result in a more replayable game, if done well.

By the way, I'm not criticizing SC2- I think it looks great. Just hadn't seen its research engine mentioned in detail, and was hoping it had been improved in regards to the really crazy random stuff. Basically, I'm not suggesting a complete overhaul- more of a 'shock absorber' addition to cut out the most egregious (and unrealistic) random results.

[Exel]

Ahem. Did any of you even read my compromise suggestion? The debate started all over again...

[Reepicheep]

Your suggestion is, I know, very similar as far as adding a maximum is concerned; but it does not include a minimum. Getting advances too quickly is just as bad as getting them too slowly.

Basically, both of those issues are what need to be corrected; how we get there isn't as big a deal as far as I am concerned.

[ April 18, 2004, 01:15 PM: Message edited by: Reepicheep ]

[pzgndr]

Basically, I'm not suggesting a complete overhaul- more of a 'shock absorber' addition to cut out the most egregious (and unrealistic) random results.

Once upon a time I suggested having a more bell-shaped normal distribution for research. There's some math involved with making it happen. Statistically, though, the averages work out about the same with what we have now. It's just the distribution of results that's more random with the SC system, sometimes very quick and sometimes very slow. Over many different games, you're going to see a normal distribution.

Remember that R&D is both research AND development. A lot of the research for various tanks and subs and aircraft was already done in 1939. Engineering and production decisions took longer. Then you have training and doctrine implementation, etc. So how can you argue exactly when something went from L2 to L3 historically? You can't. It's all an abstraction in the game. All it's doing is representing relative improvements over time, with an element of randomness for replayability.

If you look too closely at such things, you're going to be frustrated. Step back away from the trees and look at the big forest.

[Reepicheep]

Remember that R&D is both research AND development. A lot of the research for various tanks and subs and aircraft was already done in 1939. Engineering and production decisions took longer. Then you have training and doctrine implementation, etc. So how can you argue exactly when something went from L2 to L3 historically?

Those are good points. But clearly with only five increases over the war each one represents major (and probably cumulative) advances- doesn't matter what they actually are.

Now if research was already completed (as you said), the time required to actually implement it should not be indefinite. Also, given that there was so much involved in actually implementing an increase, rapid-fire increases (at this scale) should not be possible.

As I said, my primary concern is the problem of research taking far too long to complete, or research completing far too quickly. Given the grand scale, wild swings as we see now shouldn't be possible, regardless of whether or not it averages out over time. And you don't need to change the current system's random engine at all to correct it- simply add a minimum-time-required limit and a maximum-time-required limit. The results will still average out about the same over time- but will eliminate the really crazy, unrealistic stuff. I don't see any reason why that wouldn't be better. :confused:

[ April 18, 2004, 01:16 PM: Message edited by: Reepicheep ]

[KDG]

Those are good points. But clearly with only five increases over the war each one represents major (and probably cumulative) advances- doesn't matter what they actually are.

This time we have an advantage, though. Here's why.

1) Advances aren't immediate. You need to upgrade each individual unit, at a cost, thus some delay even after an advancement.

2) We can control how many advancements there are. We can have from 0 to X(I'm not sure what the max is using the editor) for each technology.

3) More research options. We have twice as many technologies. If we believe a tech is too powerful(or not powerful enough) we can edit the tech.

4) The chance to change the cost for each research option. If everyone is using one tech, and its a game killer, we can double the cost to invest as well. Or if no one uses a tech, we just cut the cost to 100, everyone will invest 1 or 2 chits then.

I agree with you that adding a min/max is beneficial(mostly to our sanity), but it sounds like this is a difficult thing for them to program.