Q about how game chooses Blind Research?

[Andz83]

interesting discussion you brought up here, fellows! go on!

[MariOne]

Thank you for the bumping.

I did provide my Pactmate Aredhran with a spreadsheet neatly arranging all those tech preference weigths exctracted from alpha.txt, is all that time gone by? ... wow

So, apparently the answer was simple: the game seems to not respect your field choice according to the alpha.txt weight, just because of withheld options exactly like in normal research...

MtG, did you find the 3-levels rule somewhere, or did you determine it from your own tests?
I'm ready to believe you without questioning, I'd just like to know how exhaustive have been those tests.
If you know something about debugging for instance, you know that if you don't observe the occurrence of an event, this doesn't mean that it will "never" occur (unless the input space is finite, and such that you can be SURE of having explored ALL the possible input values, which is unlikely to be our case...)

BTW, the 3-levels rule can't explain alone why some available techs are withheld from the picks the game offers you in normal research.
I was indeed doubting whether the picks history can have an influence too, which would make it much more difficult to investigate.
For instance, many times I observed that when a newly available tech is immediately offered to my pick, and I refuse to pick it, the third time around it will be withheld for one turn. Sometimes a tech can be withheld for more than one turn.

This is also a lovely characteristics in pbem tech trades, when you leapfrog research! That is, when another player offers you in advance the tech you're about to discover, in exchange for the one you'll immediately get when redirecting research. If for instance you don't feel ready to share AirPower with him (maybe he's outproducing you), you just pick a lamer, inoffensive tech to complete that trade, and lie to the trader telling him that the nicer tech was withheld from you... (of course you too are renouncing to immediate AirPower that way, but maybe you hope to discover it earlier than your occasional trade partner anyway, and outside the frame of a leapfrog trade...)

[Chowlett]

BUMP! - for Scipio

[theohall]

For the record, the strat guide doesn't cover how techs are selected. It just covers the formula for the Rate at which Techs are discovered. ie - how much it costs for each tech.

If anyone wants the formula, I'll take the time to post it even though it isn't really related to the original Q.

[DavidByron]

I haven't looked at the file for those numbers.... I do know that you can indeed gain a tech that is more than 3 higher than other techs you don't yet have because I've done it to transcend faster. I skipped Unified Field Theory (level 7) and got to Threshold of Transcendence (level 15).

Interesting to see some new posts added months afterwards......

[DavidByron]

The order in the alpha.txt file seems to be C,D,B,E. Numbers range 0-8. Perhaps there was a change with a patch...... this is SMACX.

[Chowlett]

I think this warrants not just a bump, but a veritable

-=Ressurect=-

[Adalbertus]

Maybe there is even one of the experts on this topic who summarises this topic for the FAQ thread. Maybe also the other formulae and things of research done.

[Blackavar]

I do believe that the game can be steered away from certain critical technologies when playing blind research.

Playing a game as Lal doing double blind, I came early in to a war with the believers. When I captured one of their bases.. steal technology gave me a choice of all *three* of the resource restriction-lifting techs. How could the believers have had them and not I?

[johndmuller]

I'm currently in a blind research game and have spent a bit of effort lately in trying to steer toward Industrial Auto. I've noticed that the program appears reluctant to switch categories in mid research, even if you feed it the tech(s) it must have been working on. In the case in point, I was apparently researching "conquer" (despite having selected build/explore/discover) and in an effort to get it to switch to build, switched to build only and fed it 2 or 3 conquer techs, including all the conquer techs it could have been researching when it first went to conquer. I realize that it is assumed that changing categories doesn't take effect until the next research choice, but it has nevertheless continued to stay at conquer even when having gotten the conquer techs, so it must now be researching a tech for which it did not even have the prerequisites when it first went conquer. I have noticed similar behavior in the AI Spartans (who presumably do have conquer selected) when trading conquer techs with them - they also have stayed with conquer. The tentative theory here is that it only switches categories when you acquire the tech under research and there are no more techs of that category available given your prerequisites. Of course, coincidence is entirely possible as well, but it would seem to be in the face of any reasonable probabilities given my research priorities.

As to the 1 year out of 3 cycle of withheld research choices in directed research, I have been aware of this theory for some time now and have intermittently noted the tech choices (and accounted for additional techs from trades/thefts/pods) and have not noticed any contradictory evidence to the theory. In particular I have not seen any deviation from the 2 years available out of 3 pattern; nor have I noticed techs being withheld twice in a row. As I understand that theory, each tech belongs to 1 of 3 groups and each time a tech is acquired, the "not" group rotates to the next one - when a choice is offered, techs from the current 'not' group will not be offered. I believe that an add-on to that theory posits a "joker" tech which at any given moment will be offered irrespective of the current "not" group; I believe the add-on, but I haven't examined enough tech choices to say one way or the other. The criteria for the selection of joker tech is not so clear, there have been several possibilities offered - for example, the highest rated explore tech (presumably Cent Ecol at the beginning), "something" depending on your faction, and "something" depending on your initial free tech, and probably others.

I couldn't say for sure whether or not either of these ideas are ultimately correct, but playing as if they were, seems to yield the expected results, so they must be close to true if not completely so.