Binary research => delayed research

[gavenkoa]

Binary research saves you 1B every turn otherwise you'd lost due to rounging error. First 100T it is significant as comparible with cost of an early tech.

Delayed research is saving of money while you are awating for reseach multiplier: a library, an academy, etc.

For the latter let's estimate the benefit. Let profit is P, expanses are X, multiplier is M and it arrives in T turns.

If we save all gold (0% research) our treasury will accumulate T×(P - X). Lets spend accumulated gold with 100% research slider: T×(P - X) - T2×X = 0. We have T2 = T×(P - X)/X turns to finish the task collecting: T2×M×P beakers. Total time is T + T2 = T + T×(P - X)/X = T + T×P/X - T×X/X = T×P/X.

If we instead kept research slider with the treasure deficit at zero our totsl research progress would be: T×(P - X) + T2×M×(P - X).

Delayed research should produce better volume, let's find the difference:

T2×M×P - (T×(P - X) + T2×M×(P - X))
= T2×M×P - T×(P - X) - T2×M×(P - X)
= T2×M×P - T×P + T×X - T2×M×P + T2×M×X
= T2×M×X - T×P + T×X
= T×(P - X)/X×M×X - T×P + T×X
= T×(M×P - M×X - T×P + T×X)
= T×((M-1)×P - (M-1)×X)
= T×(M-1)×(P-X)

So the extra is Delta × Surplus. If you income is 25gpt, expanses are 5gpt and you finish library in 5T you can benefit is extra 5×25%×(25-5) = 25g in 5T×25/5 = 25T (you save 5T and spend next 20T).

 

[drewisfat]

I'll throw in two other considerations.
Binary research doesn't always cost you that penny in rounding. Sometimes you lose it at say 90% but gain it back at 50%. If finishing the tech a turn earlier matters, I recommend playing with the slider and checking.
Delaying research also allows more time for AIs to finish researching the tech, granting you a further bonus.

 

[gavenkoa]

Sometimes you lose it at say 90% but gain it back at 50%

Right, if (profit - expanse) is dividable by 2 it works for 50%.

If finishing the tech a turn earlier matters, I recommend playing with the slider

That is especially true for racing tech (Music, Liberalism, Economics, Physics) - the price of losing is high.

Getting Pottery, Mstycism, Writing, Sailing, Currency enable more profit or greater expansion 1T earlier.

Also if you racing with AI - you could sell for bits if yuo take tech earlier. This requires guessing what AI is doing (if researching something too long that's one of 2-3 costly tech).

Delaying research also allows more time for AIs to finish researching the tech, granting you a further bonus.

Giving time opens trade options if you sure AI is busy with distinct tech. Otherwise this closes trade options... if you compete for the same tech.

 

[Peteyboy]

Delaying research also allows more time for AIs to finish researching the tech, granting you a further bonus.

I've seen a couple of "Bonuses" like this mentioned. (I.E Meeting other civs adds a bonus, I didn't realise the AI knowing a tech also added a bonus.)

Are there any other such bonuses that mean teching is quicker? Would be useful to know.

I have to confess I don't use 0% / low % enough to then benefit bonuses, but it can work very well if over stretched.

 

[sampsa]

Meeting AIs doesn't per se add a bonus, they need to know the tech. So for example if you go AH first, meeting AIs gives you nothing. If you start with agri and play on say immortal, every AI you meet has agri, so you get a bonus.

You should definitely use 0%/100% slider nearly always.

 

[floydmcw]

One issue with delayed research is that an AI may shake you down for some of that cash.

 

[gavenkoa]

One issue with delayed research is that an AI may shake you down for some of that cash.

I expect that happens early first 100T so you get -1 to relation and then AI has more means to attack you.

You buy 10T peace but it also an instrument to speedup war start where you avoid penalties from AI friends.

 

[AcaMetis]

Pretty sure that if you don't have money to demand an AI will just demand a tech or resource instead. Mind you whether that's better or worse depends, I guess.

One other minor bonus of binary research if you play with Random Events on is that it can give you a stockpile of money with which you can buy off a lot of negative events. Might not be worth it, but it at least can't hurt to have the option.

 

[Nick723]

I think the main draw back is obviously delayed techs. This can be one turn if just due to min/maxing, but can actually but much more.

The most obvious culprit in my experience is user error (I forget to turn the slider back on)!

Also however you have to plan for fail gold, capture gold and gold from trades when deciding how long to delay.

I would say as a whole that binary research has a fairly minor impact and should be ignored even on deity unless you are a careful person (which I am not) or happy to reload when you forget.

 

[elmurcis]

As person with "memory leak" (I might "skip turn" for workboat for 20 turns just because I know I will find it 40 turns later still sitting there long after border pop) - can confirm, need to reload just because I collected too much gold is frustating so "constant" slider position tends to be more used.
Like during heavy "REX" phase I tend to keep research at 20% as for me it is best indicator (+/- gold situation) how healthy economy is (green = keep settling, red = keep settling but in red)

 

[sampsa]

You can set a reminder (Alt+M) or set a note somewhere on the map (Alt+S) or just not be too hasty when ending turns. Play slower -> play better. Playing better allows you to start playing faster little by little, after most things have become automatic for you (like checking slider position every turn before clicking end turn). I wouldn't describe myself as a careful person, but I close to never forget to have the slider in exactly the position that I want.

Minor impact... well, depends on what exactly is meant by minor.

 

[Nick723]

I would say minor in that I doubt a game has ever been won or lost over the use or non-use of binary research.

Of course I take your point that it’s better play, and striving for the best play is kind of the point of this forum. I just feel that a newcomer to this forum may sometimes feel that all they need to make the jump from noble -> immortal is to run binary research (I exaggerate for effect), whereas in fact it’s a pretty minor thing that could be safely ignored completely.

 

[Mr_Trotsky]

In a recent game I had the choice of teching engineering or guilds. I needed engineering a lot more (cannons) and nobody had either when I started banking gold. By the time I was ready to start teching, engineering was available in trade and only one AI had guilds. I teched guilds and traded for engineering. More importantly guilds was the only tech I had that I could use to bribe a 12 city AI off an 8 city AI that was about to become a 7 city AI and cap. I then used knights and trebs to cap the 8 city AI who then teched to rifling for me while I teched steel and won the game. I’m not sure I’d have lost if I’d just started teching engineering at 30% (or whatever crippling low level I could manage at break even) research but it would have been much more difficult. Therefore, I do think it can be game breaking. Probably not very often but sometimes.

I do agree that it’s probably not a huge point on lower levels though. It really comes into its own when you’re teching to desperately stay in the trading game. I also agree it’s more complicated than it initially seems with fail gold etc needing to be factored in and that it’s easy to just forget to start teching again. I do that a lot. Especially early on when you may only need to drop the slider for a turn or two to bank enough gold to go for the wheel or similar.

 

[Fippy]

Binary might be minor, but waiting for the right moment to start researching isn't with tech trading on

 

[ArchGhost]

Another little tidbit outside of min-maxing: if you use binary research you spend through gold stockpiles faster (you have a large amount of gold for fewer turns), giving the AIs less of a chance to demand it as tribute as it has to be over a certain of amount of gold before they ask for it. I tend to try to keep my treasury low on maps with lots of players partially out of this (it's annoying!) and partially because of RTS habits of avoiding large resource banks

 

[floydmcw]

Another little tidbit outside of min-maxing: if you use binary research you spend through gold stockpiles faster (you have a large amount of gold for fewer turns), giving the AIs less of a chance to demand it as tribute as it has to be over a certain of amount of gold before they ask for it. I tend to try to keep my treasury low on maps with lots of players partially out of this (it's annoying!) and partially because of RTS habits of avoiding large resource banks

Isn't it the opposite? If you start with a small amount of gold and research say at 50% which is breakeven for gold, you will always have that much gold.

Binary research, you stockpile gold till you have enough to run 100%, then you burn through it. You will average more gold this way. You will sometimes have more gold than the steady-state approach, and will never have less.

 

[SittinDown]

delayed also offers flexibility. you can change what you tech, because gold isn't applied to a specific tech yet.

 

[civac]

Assume you have 18 base commerce each turn, pay 9 gpt maintenance and you research a tech with 120 beaker cost and one arrow bonus.

Researching 5 turns at 0% and 5 turns at 100% nets 5 * 1 + 5 * 22 beakers for a total of 115 beakers, tech is not finished.
Researching 10 turns at 50% nets you 10 * 12 beakers for a total of 120 beakers, tech is finished.

What does this tell us? Nothing. This situation is so rare it can be safely ignored. Just throwing it out there to confuse people for fun and profit.

 

[Peteyboy]

Assume you have 18 base commerce each turn, pay 9 gpt maintenance and you research a tech with 120 beaker cost and one arrow bonus.

Researching 5 turns at 0% and 5 turns at 100% nets 5 * 1 + 5 * 22 beakers for a total of 115 beakers, tech is not finished.
Researching 10 turns at 50% nets you 10 * 12 beakers for a total of 120 beakers, tech is finished.

What does this tell us? Nothing. This situation is so rare it can be safely ignored. Just throwing it out there to confuse people for fun and profit.

I think that misses the primary point of why Binary research works.

If you take pure numbers and no other factors, there is relatively little in it (I haven't proof checked the numbers, but from experience, there is little in it purely based on a non-real example.)

What it doesn't account for is the following:

During those 10 turns building research boosting buildings, meaning the benefits are worth proportionally more. Especially valuable with the academy (I'm not a good binary user, but even I will use it when I know a GS for a planned academy is coming.)
Similarly growth in the 5 turns at 0% of the cities (which across a reasonable empire is likely) adds the same effect
The AI techs something that you plan to tech (and can't trade) - Then you get a discount on top of your extra research
If you are have a tech good for trading (I.E Aesthetics is a common one) - Delaying allows you to adjust depending on what the AI techs and you trade for (No wasting research on something you can trade for.)
Meeting AI (especially on maps that AREN'T Pangaea, lake etc) will reduce the cost of further techs once at 100%.
I imagine there are a few others (like utilisation of specialists too.)

Like many of the things mentioned by some of the pro players here, on their own they aren't "game breaking" - but they all add up, and good utilisation of things like binary research, alongside these other elements can make the difference at the highest level, between winning and losing a deity game, whilst I'm not there myself, these have filtered through to me becoming an immortal player. I've seen write-ups and often Deity games will be very close at the end, so to say it can be ignored is possibly a little naive,

BUT (I feel like I'm repeating a point from another post re fail gold) it's not essential, and if you aren't fussed then it's each individuals choice. Using the optimal strategies takes time, which some won't want to use. But I do take Sampsa's point, that if you slow down and use these regularly they become common practice. Probably without these little advantages you're probably restricted to Immortal, maybe Emperor level (In many wayd where I was 2 months ago, my standard play was reasonable, but I played fairly quickly without enough management and sometimes it would show, and I wouldn't win where better players would.

 

[gavenkoa]

Meeting AI (especially on maps that AREN'T Pangaea, lake etc) will reduce the cost of further techs once at 100%

What are the math behind this?