Aircraft R&D

[n01487477]

We are not discussing about being generic on some game subject, like R&D (that's fine for me), we are saying that the statements about R&D in the game manual are totally wrong if your testing are correct.
That's a bit different IMHO.
Expanding and repairing a 300 R&D factory is a huge investment in resources in game terms for the japanese, it would be nice to know - officially - if this is a waste of time because it's the same of a 30 R&D factory.

Bye

Invernomuto, I completely accept & agree with what you are saying, I am not part of the development team, but agree with you that is would be nice if it were cleared up esp. if the manual is so wrong on this (which I haven't read about this yet)... I never truly believe the manual, having come from a sceptical position on witp... I am just giving what I think to be true from me witp testing, if the dev's want to correct me on this then please do so...

Damian - does tracker give us any ability to determine how many "development points" have been accrued? I'm sure the number is tagged in the database somehow. I have run a couple of months test in AE where I set up 3 factories, each of size 800 and each working on one late war Japanese plane type (like 10/45 and later) and set the scenario start date to end '44. I ran two months and saw no advancement of any plane by R&D and thus concluded that plane R&D in AE was useless. Of course, if a size 800 factory is no more useful than a size 30 factory then this would make sense that I saw no advancement. Note I would have generated approximately 4,800 dev points in these two months (or 48 "tries" at 100) so I suppose if the advancement chance is less than 2% per "try" for planes about 1 year out then I may not have run the test long enough.

While Tracker,in its current state, doesn't show the development points, I can see them if I really searched for it & trackers philosophy was to show data not in game (which it isn't). I note that staff, (according to Marcus), does show an estimated value, I have yet to test if he is showing accurate data, or just a calculated field...
Anyway, as to your test, it is inconclusive either way (but I applaud the effort), cause in witp, I found no difference with time remaining, except for repairs (which is another really interesting discussion)

Maybe it is time to sharpen the pen & paper and do some testing again [;)] Even if I'm not on the Dev team, and am really a nobody who just likes tinkering, and extrapolating theories [8D]

[erstad]

ORIGINAL: vaned74

Damian - does tracker give us any ability to determine how many "development points" have been accrued?  I'm sure the number is tagged in the database somehow.  I have run a couple of months test in AE where I set up 3 factories, each of size 800 and each working on one late war Japanese plane type (like 10/45 and later) and set the scenario start date to end '44.  I ran two months and saw no advancement of any plane by R&D and thus concluded that plane R&D in AE was useless.  Of course, if a size 800 factory is no more useful than a size 30 factory then this would make sense that I saw no advancement.  Note I would have generated approximately 4,800 dev points in these two months (or 48 "tries" at 100) so I suppose if the advancement chance is less than 2% per "try" for planes about 1 year out then I may not have run the test long enough.

The one post on classic Witp that I recall that was authoritative or semi-authoritative was that 100 points gave a probability to advance, and the probability was based on how far out on the future the standard date was. The further out, the less likely. You may want to try doing the same experiment with an arrival date of 5/42 or something.

[vaned74]

Hi erstad - understood.  That's why I started the test scenario in like Dec, 44 so as to be less than 1 year from any planes arrival I had set the factories up to research.  I have yet to see a single advancement by a plane in R&D in any situation.

Interestingly, engines don't seem to work the same way.  In my current PBEM, in July,42 I have had multiple engine advancements.  Eg.

The NE turbojet advanced one month despite only having one size-8 factory working on it at the time, same thing was true for the Toka rocket.  The Aichi Ha-60 advanced 4 months in the 7 months we have played - had about 140 production rate for that one in R&D.  I have had another engine (one of the 1943 ones) advance a month as well.

[Rob Brennan UK]

ORIGINAL: vaned74

Hi erstad - understood.  That's why I started the test scenario in like Dec, 44 so as to be less than 1 year from any planes arrival I had set the factories up to research.  I have yet to see a single advancement by a plane in R&D in any situation.

Interestingly, engines don't seem to work the same way.  In my current PBEM, in July,42 I have had multiple engine advancements.  Eg.

The NE turbojet advanced one month despite only having one size-8 factory working on it at the time, same thing was true for the Toka rocket.  The Aichi Ha-60 advanced 4 months in the 7 months we have played - had about 140 production rate for that one in R&D.  I have had another engine (one of the 1943 ones) advance a month as well.

This is really odd then. looks like common sense won't work in this theatre [;)]. As to testing etc. Personally i don't have the time or a necessary understanding of the editor to make any conclusions of my own.

But Kudos to anyone who does , and re-reading my earlier post it was a bit confrontational in its tone so apologies if i offended.

[pompack]

Not a dev, but I ran a number of trials on WitP some years ago. I cannot find the data anymore, so this is from memory

Created sandbox and ran about 15 trials (including some real games). Test was aimed at investigating Tojo and Tony R&D only.

Boosted R&D level to about 200 for both Tojo and Tony
1) In NO trial did either produce more than two months early
2) Three times one or the other produced two months early
3) In only one case did neither produce early
4) The rest (nominally 11) one or both produced one month or less early (Several (three I think) advanced in the middle of the last month so they were early but not a full month early)
Sorry, cannot remember exactly how many failed to advance even one month


The conclusion I came to was it was not worth it to get everything repaired more than five or six months early; it did not do any harm, but R&D more than six months in advance did not seem to affect anything.

[n01487477]

I believe you Pompack, but only that your testing most likely failed to take into account what I'm saying about 300(0) & 30(0) ...

Hell I may be failing to take into account stuff too ...

3 day turns (for sanity) one base on my sandbox ... make your own conclusions [;)]




Rob Brennan, no probs mate [8D]

The real problem with R&D is getting those factories repaired, but that is for another time and debate, and why should I reveal my secrets to you guys [;)]

BTW. Tracker 1.1 has just been released

[Djordje]

Looks to me like working as designed for large factories, and that problem comes from lots of small factories researching same aircraft. Considering that in the real game player won't have so many research factories available to him (for a single plane type), and that he would have to spend some time and supplies repairing them to full capacity before they start researching I can understand why this slipped below the devs' radar.

What worries me is that dedicated player can sacrifice lots of semi-useless later war plane types and focus on good models from 1945 and make them appear in early 1943...

Nice catch and great work with the testing.

[n01487477]

Considering that in the real game player won't have so many research factories available to him (for a single plane type), and that he would have to spend some time and supplies repairing them to full capacity before they start researching I can understand why this slipped below the devs' radar.

What worries me is that dedicated player can sacrifice lots of semi-useless later war plane types and focus on good models from 1945 and make them appear in early 1943...

For the reasons outlined in your first sentence, the second sentence is not so much of a worry anyway ... The time to repair factories of models a long time away is really prohibitive, but in a modding sense and even from what people have said here about engine R&D, it is something to be aware of. I did some testing on this (repairing R&D factories) a while ago, and that was very interesting too ...

I play Nemo121's EA (for witp against him), a great mod, with lots of options of building in many places, this flexibility has certainly allowed me to do some things, others can but dream of re: R&D and production, 6 engined planes, more CV's than historically possible, troops [;)] ... back to the point Damian ... what I mean to say is that this system is fine for historical AE, but what happens with a modded version giving this type of flexibility ?

I'll refrain from other issues, and desires to open up more to the editor [;)]

[pad152]

n01487477

I though you couldn't advance a plane more then a couple of months (true/not true)?

In your test with the A6M2 Zen Baku, how many factories did you have on R&D?

[pad152]

ORIGINAL: pad152

n01487477

I though you couldn't advance a plane more then a couple of months (true/not true)?

In your test with the A6M2 Zen Baku, how many factories did you have on R&D?

What about engines, do engines follow the airframe availability?

[pompack]

[quote]ORIGINAL: pad152

n01487477

I though you couldn't advance a plane more then a couple of months (true/not true)?

In your test with the A6M2 Zen Baku, how many factories did you have on R&D

/quote]

Note that in his test, he had 20x30 factories completely repaired on 7dec41 and he advanced it one month in the first week and four months in three weeks. The key is multiple small factories totally repaired. At that rate it should be available by about the middle of May42.

[vaned74]

As Pompack said, the key is the number of factories repaired.  A few key numbers come to play to make Damian's test results very hard to actually bring into play in a real game (which I suspect is what the designers intended).

(1)  repairing factories for which the target plane is way in the future is very difficult.  eg - I set up a test starting in Dec, 1941 with 17 factories each with 20 damaged planes for Ki-84-1r. At the end of 30 days, only 11 points of production had been repaired.  This is a repair rate of 0.6 factories/day for 340 factories being worked on - ie the repair rate for a plane that far out is 0.0017 per day thus meaning a size 30 factory would take about 17,000 days to fully repair. 

(2)  if any part of a factory is damaged, it doesn't contribute to R&D as best I can tell.  This exacerbates the repair rate in #1 above

(3)  when you expand a factory, it doubles - thus, while you may be able to get a bunch of functioning 2,3,4 plane size factories doing R&D relatively quickly - if you expand them again to get more R&D you are shutting off your current R&D (per #2 above) for a long time to repair your doubling in size.

Now the repair rate may increase as you get closer to scheduled availability of the plane - I think it does based on my game experience in July, 1942 in my PBEM.

So JFBs don't get your hopes way up to get that Ki-84-1r frank in 1942...or 1943...or likely not even 1944.  I think this system is probably working as intended, early to tell though.

[pad152]

Setting up a test with 600 (30 x20) factories is just bogus and doesn't prove anything because he never completed the test to see if the plane and engine really became available.

Things I still don't understand.

1. If total R&D < 100 for an single aircraft type does it really help advance it?

2. Do new engines advance too and/or magically become available when the airframe becomes available?

[vaned74]

answers to your questions are:

(1)&nbsp; yes - it does, the points accumulate until 100 is reached

(2)&nbsp; the engines don't magically become available as a result of airframe availability; wasn't this way in WitP either.&nbsp; so the japanese advancement through R&D is further complicated in AE due to the multiplicity of engine types, once again probably working as intended - keeps the player from stockpiling a bunch of engines in the first part of war that somehow work to provide for both the A6M2 and the Ki-84 Frank.

[pompack]

ORIGINAL: vaned74

Now the repair rate may increase as you get closer to scheduled availability of the plane - I think it does based on my game experience in July, 1942 in my PBEM.

I am not so sure about that anymore. n01487477 (aka Damian) has caused me to re-examine what my old WitP tests really demonstrated. I am not convinced that the Summer of 42 kick in Tony/Tojo availability wasn't really caused by the fact that it took me that long to repair all of the facories that I had converted to R&D. At this point, without data, it is hard for me to determine. And I am certainly not going to set up WitP again just to repeat the tests [:D]. Shortly a lot of people will have AE data to see if a lot of small factories (for rapid repair) will produce more immediate production advances (and at this point I am conviced that they will)

[vaned74]

Checked that too - on smaller factories.&nbsp; Ran a test earlier today - set up 20 small factories of size 2 each (all damaged).&nbsp; Factories were set for Ki-84-1b (avail 45/03) and game date was 1941.&nbsp; Ran 30 days - repair rate was still about 0.0017 per day - 2 plane production points repaired.

Decided to change the game date to April 44 (less than 1 year from original plane availability date).&nbsp; Ran 30 days - same repair result.

Really tends to mean that airplane research for Japan will be a very slow process indeed if partially damaged factories don't contribue R&D.&nbsp; Ok by me - if Allied production is fixed then Japanese plane R&D shouldn't really merit much benefit unless very significant resources are dedicated to it (like pushing almost all your factories to R&D thus denuding yourself of needed production while awaiting your R&D result).

[n01487477]

Once again from tests with witp, NOT AE, I thought it was determined that repair rates do actually increase the closer you get to its production date...
Taken from the witp-wiki:
For aircrafts in research probability of repairing of factory element depends on number of days till aircraft enters production and total factory size (sum of ready and disabled elements): If RANDOM(0 .. NumberOfDaysTillInProduction) > SizeOfFactory then factory element will not be repaired that day.

Having tested this independently some time ago, I have to say that at face value while larger factories do repair faster, they also have a lot more repairs until they are completed. Smaller, have less to repair, but take longer ... anyway, I found that there was a convergence point in time where both small and big factories became fully repaired... pretty good math WAD.[;)]

The upshot is like Vaned,Pompack and others have said, there are no free rides here due to this requirement of having fully repaired R&D. R&D is a gamble, (no not dice rolls), just the reduction of time V the ability to fully repair enough factories, which then need time to produce 100 points. It is a nice conundrum [8|]

Pad152:
Setting up a test with 600 (30 x20) factories is just bogus and doesn't prove anything because he never completed the test to see if the plane and engine really became available.

Oh, they become available, you can believe the in-game screen availability (well if the engine is also available -- we can see this even with witp-CHS) And Thanks again Pad for bringing my shortcomings in testing to the table, I set up the tests as a proof of more than 3 months advance ... Why 600 ? cause I don't want to waste my time running months of testing on something I already know to be true. Believe me, don't ... different strokes for different folks AYE ?

[Czert3]

These runs proved one think - that fully repaired mega-factories are waste of time for RaD, and bunch of small factories are usefull.
But I think we still need respond from devs WHY game work in this manner. I cant see any reason why 1x600 speed-up points are not effective as total 600 points, but in 20x30.

[Mynok]

Well, if the 600 factory doesn't do ANY R&D until it fully repairs, then it is obvious why the smaller ones work better: they get fully repaired faster.

[Czert3]

ORIGINAL: n01487477

Mynok, this is for you, if you missed this. This was sand-box test with all factories fully repaired (and producing speed-up points) 1x600 and 20x30.