Binary Research question

[Paperbag_Hobo]

I did the reading on binary research a while back, and since then I've pretty much only had my slider at 0% or 100%. Exceptions are if I need a tech within a certain number of turns and can afford it at specific rate, or in the early game before modifiers become a factor.

I notice in most of the games I see on the forums people tend to run the slider normally. Am I missing something?

 

[Abegweit]

The reading on binary research is outdated because Firaxis finally noticed that modern CPUs come with floating points units and started using proper math. There still are some advantages to BR. There's a small improvement in research rate, especially at the beginning. Also you can change your mind about what tech to research while gathering the beakers instead of when you're halfway through

However, it's no longer the huge gain that it was when CIV first came out. This being said, I think most folks don't use it out of laziness more than anything else. It's a pain and I often don't bother, especially if neither of the conditions mentioned apply anymore.

 

[Paperbag_Hobo]

Yeah the foresight you get is another decent benefit I forgot to mention. I did not however, know about them fixing this problem. Thanks for the update, I'll probably still use it mainly for the aformentioned reasons, but I won't feel so smug about it anymore

 

[Cam_H]

The simple explanation outlined by Grey Fox outlines how I understand to best leverage it;

The best ways to use Binary research is to save up gold while your research multipliers are low and then increase it to max once you got the research multipliers in place. For example the turn you get Writing, build libraries and turn the slider up again once you either have libraries or an Acadamy.

Once you're at Education, same deal. Or while building Oxford.

Other reason is to get a tech quick in one research cycle to trade it. This is to be sure to have something to trade. If you are teching it at a low slider someone might get it before you do and then much of your research was a waste (unless you have something to trade for the rest of the tech).

 

[lymond]

Yep, that is my understanding as well. I find I use it a lot early game after writing - using Aesths and/or Alpha as a placeholder - to fund more expansion. Like getting to six cities quickly. The timing is usually right at that point.

 

[Abegweit]

That's certainly an excellent reason as well. And the opposite applies too. If you are building markets, you should have your research rate as high as possible to take advantage of the wealth multiplier when they come on line. Takes more planning for this, of course.

However, PH was referring to this article in which the author explains how to take advantage of Firaxis' peculiar obsession with integer arithmetic. In a lesser extent, Zombie's point is still valid today. Firaxis now adds up all the science in floating point but still insists on rounding down the result. Same thing with wealth. Therefore, on average, you lose 1/2 for each turn your research and 1/2 for each turn you generate wealth. Binary Research reduces the losses here too - especially, as I said, early in the game. Saving 1/2 commerce per turn is pretty significant if you are only making 10 or 20.

 

[Paperbag_Hobo]

Yes the rounding rules were really why I began using it in the first place - the rest are just bonuses

 

[Aoxomoxoa]

I have fallen back to the 'laziness' side of the binary research. Mainly because it drove me crazy to forget to change back to 100% by a couple turns or more.

Does anybody know what the effect that has? That is your running 0 or maybe 10% and you know you'll need xxx gold to finish whatever tech at 100% and blow past a turn or two without switching back. How does this change the efficiency gain of hitting all your markers on the binary path?

I know you'll get your tech later, but I'm wondering about the gain by doing it perfectly and how many turns missed cause the gains to become losses.

 

[Aoxomoxoa]

The same reason I gave up 'prechopping' forests...

 

[Kid R]

BUG needs a new notification message, e.g. "Engineering can now be researched at 100% science (7 turns)". Even better it could move the slider for us too.

 

[kossin]

There are still rounding issues, albeit minor ones. They are most noticeable in the first 50 turns as you can lose 1~2 commerce from partial sliders.
example: Total commerce = 21
50% research = 21*.5 = 10.5 -> rounded down to 10
50% gold = 21*.5 = 10.5 -> rounded down to 10
10+10 = 20
1 mysterious commerce vanished...

The other reasons for it have already been highlighted.

 

[coanda]

I have fallen back to the 'laziness' side of the binary research. Mainly because it drove me crazy to forget to change back to 100% by a couple turns or more.

Does anybody know what the effect that has? That is your running 0 or maybe 10% and you know you'll need xxx gold to finish whatever tech at 100% and blow past a turn or two without switching back. How does this change the efficiency gain of hitting all your markers on the binary path?

I know you'll get your tech later, but I'm wondering about the gain by doing it perfectly and how many turns missed cause the gains to become losses.

If I understand correctly, the situation you're discussing is that you're running Binary Research. You earn enough gold at 0% to research something directly at 100% research all the way... then you don't bump up your research for another turn or two.

How much difference it will make depends on what tech it is and why you're researching it. Delaying Paper for 2 turns due to laziness, for example, is not going to hurt you - you'll make up those two lost turns on researching Education, so the only effect is that you can map-trade 2 turns later and you're a little less likely to get University of Sankore. It costs you basically nothing (unless you really needed those map trades... ), and during the two turns you waited you might well have finished another library or two - which means it would actually gain you beakers. That is, it doesn't hurt, it helps. Delaying Currency for 2 turns, on the other hand, can be a lot more painful. It's almost certainly going to cost you at least 15 gold between direct and secondary effects; it's certainly not unreasonable that a mistake like that would cost you 50 gold by the time all is said and done.

Truly optimized binary research would actually leave the slider at 0% until ready to beeline to a tech that would actually be of use (whether for direct use or trade purposes), then put it up to 100%. So, for example, if you're isolated and have plenty of health (no need for harbors), you might leave the slider at 0% until you have enough gold to go straight through Compass -> Machinery -> Optics with the slider staying at 100% the whole way. Of course, this sort of long-term binary research would require being able to predict how much gold you'll need for that, which is a little tricky... but in theory it's better play.

 

[AutomatedTeller]

There is a bonus for being a tech laggard in Civ, too - which is that say you want compass, and it takes 10 turns at 70%. you do 3 turns of 0%, then 7 and 100% to get it in 10... but someone else researches it on turn 6, so you end up with a bigger beaker overflow for your next tech.

personally, i leave the slider where it is, or try to run 100% and subsidize myself with trades, great merchants and merchant specialists

 

[UWHabs]

Occasionally I do binary research when I'm hovering around 0 gold, and the slider at 30% would put me in the red, but at 20% I'd be up a lot. So my slider is effectively at "25%". Then it can be easier to just shut it off for a few turns and swap it back on later.

*to add: also, keep in mind, it can actually be beneficial to not do this. So if you have a commerce city building a bank, you might want to keep your slider at 50% the whole time so you don't waste as many turns with a bank.

 

[DMOC]

I notice in most of the games I see on the forums people tend to run the slider normally. Am I missing something?

I research with binary (almost) all the time.

As mentioned before, I wait until research multipliers are built, then crank up research.

I have gotten "lazy" managing research so I just make my research 0 or 100% almost all the time.

 

[Praeclarus]

Can someone explain how this works?

 

[coanda]

Can someone explain how this works?

http://www.civfanatics.com/civ4/strategy/micromanagement.php

Note that, as people mentioned, it's a little out-of-date. All research across your empire is now summed up in floating point numbers before being rounded off.
Example: Suppose you have two cities, one with 9 commerce and the other with 7 commerce, and no infrastructure. At 60% research, city 1 generates 5.4 beakers and 3.6 wealth, and city 2 generates 4.2 beakers and 2.8 wealth. When the article linked above was written, that meant that city 1 would contribute 5 beakers and 3 wealth, and city 2 contribute 4 beakers and 2 wealth - for a total of 9 beakers and 5 wealth. Over a period of 10 turns, you would generate 90 beakers and 50 wealth. On the other hand, running at 100% research for 6 turns and 0% research for 4 turns, you would generate 96 beakers and 64 wealth - a 6 beaker and 14 wealth gain. This is because the game rounded down each city's production before adding them all together. In the most current version of the game, that is no longer the case - at 60% research, each city generates the same amount, but they are all added together before being rounded down. So instead of a total of 9 beakers and 5 wealth at 60% research, you'll get a total of 9 beakers and 6 wealth. You still lose 6 beakers and 4 wealth over a 10-turn period in that example vs. binary research, but you can at most lose 1 wealth and 1 beaker per turn, no matter how large your empire - so once you start generating more than 15-20 commerce, the savings become negligible.

 

[PieceOfMind]

There's a good chance that if you don't come up with the reasons yourself to use binary research it's not going to benefit you much anyway.

In the PBEM games I play, every turn I'm thinking about where the slider should be.

As a quick example, suppose setting the slider to 100% science means you would get the next tech in 4 turns, but you don't have enough gold to support that rate for 4 turns and at break-even you would get the tech in only 6 turns. Further suppose you have a library about to complete in 2 turns in a city with high commerce and/or beakers. In this case, it would probably be a good idea to set research to 0 for 2 turns until the library completes, then set it to 100% again after.

Even that example is hard to explain, and other factors would complicate things even further. The point is, binary research is basically extreme micromanagement that is not even going to do much for the players who know how to use it.

It's quite understandable IMO that people like DMOC who are used to using it, do it almost more out of laziness now than anything else.

Another reason to use it is that in MP you could mask how close you are to getting the next tech. It doesn't take much espionage for a rival to see your research, so if the info that player is getting is that you will have Civil Service in 600 turns lol, all he can work out is that you are at least intending to get that tech but not necessarily know how soon you can get it. Next turn he might discover you'll have it in 3 turns!

Even more cunning is that you could invest that free 1 beaker per turn into a completely irrelevant tech so that your rival can't know what tech you are intending to research. Once you have enough gold to be able to research the tech you want at 100% science the whole way, you switch only at that time and so give your opponent as little time as possible to know what tech you are researching.

 

[Kid R]

I guess it would make most sense then to set research to "no tech" and bank the free 1PT for later.

Or better still, when you do turn up the slider, still leave research on "no tech" until the point when you really do have to commit to a specific tech to avoid getting some cheap one like Divine Right automatically.

 

[coanda]

Ok, now that's something new for me. How do you research "no tech"?