BlitzPig_EL Posted March 1, 2015 Posted March 1, 2015 So, what this boils down to, as I see it, is that some of you seriously doubt the data used by the developers to create the flight model for the FW 190 is correct. Some of you are claiming outright bias on the part of the dev team, while some seem to think it's just errors in the data they have. Does that about cover it? So, what data set do you propose they use in lieu of what they have now? Also it seems that some think the FW 190 is good, but that the VVS aircraft are UBER, based not on in game test numbers, because there are none, but based on your personal results online (or off). Yes? I can play that game as well. I was just online flying an F4, which I am no good in BTW, and had two Yak 1s come after me. All I did was put the nose down a bit and open the throttle and I left them standing like their landing gear was still down. So I will say that the Yak 1 is a dog in a straight line, based solely on my recent qualitative experience, and anyone that gets caught by one can only blame their poor decisions as a virtual pilot. See how that works?
reve_etrange Posted March 1, 2015 Posted March 1, 2015 So, what this boils down to, as I see it, is that some of you seriously doubt the data used by the developers to create the flight model for the FW 190 is correct. Some of you are claiming outright bias on the part of the dev team, while some seem to think it's just errors in the data they have. Does that about cover it? So, what data set do you propose they use in lieu of what they have now? I think you are misinterpreting unreasonable's remarks somewhat (though he may correct me). His point about the scientific method is correct, but that doesn't mean he his questioning the motives of the game developers or the accuracy of their data (indeed there may be no other such data). The point is that when you do not (or cannot) divide data into separate training and testing sets, all you can do is better fit your particular data. In this regard, pilot accounts are valuable as a separate data set from the one used to generate model parameters. On the other hand, as unreasonable said, both the historical accounts and quantitative test data are very limited, and are subject to possible sources of bias. Bias here does not mean ill will from those involved in the process from war-era aviation to an awesome modern flight sim. Rather, it is any systematic error including both that from technical sources and the unavoidable subjective decisions made along the way. Any question? Doesn't the deflection of the elevator relative to the aircraft affect the negative lift from the tail? It seems to say so here. Maybe you can explain over PM since it is so off topic. one can only blame their poor decisions as a virtual pilot. I think it's a good sign that pilot skill is the determining factor in the majority of cases!
unreasonable Posted March 1, 2015 Posted March 1, 2015 @reve_etrange - no correction necessary, you express my thoughts better than I do myself, although I was not sure that BlitzPig's words were addressed to me specifically. Just to be clear on my take on bias, as this can get emotive: I certainly was not accusing the developers of deliberate bias, although I know some have done. Everyone is subject to some unconscious bias, but when the devs say they are scrupulous in trying to eliminate this by sticking to the best sources I have no reason to doubt their word. The bias I was thinking about was more in terms of the people originally compiling the data sources, who were subject to pressures that most of us thankfully never face. We just do not know how much effect this might have had on what otherwise might look like objective technical reports. 1
StG2_Manfred Posted March 1, 2015 Posted March 1, 2015 I've put in quite a few hours over the last day or so since the update and patch, 'fanging-about' flying the 190 on 'Normal servers'. While my personal view is that issues such as roll-rate still have not been resolved, I think the 190, as it's currently modeled right at this moment, is as good as it's ever been. I don't know if any little changes have been made since the update but it definitely feels better to me. If you enter a fight from altitude and keep things fast and aggressive you can usually do reasonably well. I note that Yaks can still catch you in a long tail-chase if you don't take steps to terminate the fight at the first opportunity but generally speaking, I've felt that, in the last day or so, I'm the one who's been responsible for my online deaths; not the 190. So, I'm reasonably happy as things stand. I've got the same impression Wulf, but as JtD pointed out in his post a climb performance 20 percent worse as it should be, is nonetheless a issue and a big disadvantage. And also historical incorrect...
303_Kwiatek Posted March 1, 2015 Posted March 1, 2015 If A-3 in BOS got historical accurate climb rate and good controlability at high speed it would be really different plane. Something like in 1942 was?
Wulf Posted March 1, 2015 Author Posted March 1, 2015 I've got the same impression Wulf, but as JtD pointed out in his post a climb performance 20 percent worse as it should be, is nonetheless a issue and a big disadvantage. And also historical incorrect... Check out the runway speed. It will now do 300km/h by the time you reach the end of the runway. I think it did about 280 km/h on Friday. And it seems to retain more energy in climbing fights. Seriously, it seems much less ponderous than it did previously.
Yakdriver Posted March 1, 2015 Posted March 1, 2015 verdict?verdict is "still an ugly piece of Crapola"i still hate her and fear her as much as before.so. there you are.
JtD Posted March 1, 2015 Posted March 1, 2015 (edited) I've done a more detailed climb test now, still combat/climb settings at ~270 km/h indicated air speed with standard loadout. Reference is the same Fw190Aa-3 @ 3850kg. At least at 270 km/h, it climbs worse than it should, by about 1m/s throughout the altitude range. Which is about 6% at low, and near 10% at medium altitudes. It would be interesting to know what the developers think about that - different target performance, and if so, which? Edited March 1, 2015 by JtD 5
Rama Posted March 1, 2015 Posted March 1, 2015 OK since you ask: "But I agree with you it's hard to understand when you don't know what a scientific approach is and what the physic of the measure is." Pretty clear cut in this case, but the same tone runs throughout your post. I allready told you that the wording of this sentence was probably poor, and it may reflect something not intentionnal.... now you may off course refuse to believe me. I don't see anything about the overall tone of the rest of the message that could offend you. Maybe I'm really bad in english, or you're missreading, or want to feel offended? (just hypothesis to try to explain, I don't pretend anyone true, you're the only one that can know the real reason) One of the main things I remember being taught about science of any kind is that you cannot test a hypothesis against the data set that was used to derive the hypothesis. Totally correct If the FM builder was able to generate a hypothesis producing model from one set of data and then test it against other data this would be fine. But as far as I can see this is not what is going on here. The FM builder has a limited data set from which he builds his model. The "testing" that people seem to talking about here is purely to see if the model's outputs are consistent with the data set that was used to produce the model. If the answer is no, this just means that the model will be tweaked until a better fit is achieved. No new empirical knowledge is being produced. No science here. You're making hypothesis about how the FM builder work then draw hypothetical conclusions from them. Actually he's using different data to build the models and to test them. And the aircraft performance test data are not used to build the model, but to check the outcome of the model. Just because something is expressed in quantitative terms does not make in scientific Correct any more than the use of qualitative terms means that something is unscientific. Qualitative terms are used in descriptive sciences (history, etc..) But there's useless in physical models. The criterion is testable predictions. Correct. Testable quantitative prediction. And no, restating logical truth functions as binary digits does not magically transform them into quantities. It's not magic. Binary values are values, and can been combined to form complex quantitative models. BTW, the whole FM is working on a computer, with a code based made exclusivelly of combined binary values and logical truth fonctions....
Rama Posted March 1, 2015 Posted March 1, 2015 In this regard, pilot accounts are valuable as a separate data set from the one used to generate model parameters. Except that, as explained before, the pilot accounts can't be related to a physical performances, and in the very few cases they could be (with a lot of hypothesis), it's with an error margin much larger than the performances quantitative values themthelves. That's why they're useless. On the other hand, as unreasonable said, both the historical accounts and quantitative test data are very limited, and are subject to possible sources of bias. They are not that limited, in particular for WWII, they are lots of data for some planes (and less for others). And yes all the data (in any domain), are subject to error, these error including random elements and bias elements. That's something you have to live with, it's the daily bread of the physic of the measure and of the engineers working on models. The more you want a model to be accurate, and the more quality quantitative data you'll need.... and nothing else will replace them. Doesn't the deflection of the elevator relative to the aircraft affect the negative lift from the tail? It seems to say so here. Maybe you can explain over PM since it is so off topic. Yes it does. As does the variation of AoA of the stabilizer, and yes (to the previous question), both can be combined to increase the pitch authority that you would have if using only the elevator (except if a specific mechanism disallow it). Better don't ask Crump to explain, you'll be lost in no time.... 1
NachtJaeger110 Posted March 1, 2015 Posted March 1, 2015 I've done a more detailed climb test now, still combat/climb settings at ~270 km/h indicated air speed with standard loadout. Reference is the same Fw190Aa-3 @ 3850kg. At least at 270 km/h, it climbs worse than it should, by about 1m/s throughout the altitude range. Which is about 6% at low, and near 10% at medium altitudes. It would be interesting to know what the developers think about that - different target performance, and if so, which? Just a thought.. According to this test the BOS 190 is (in the extreme case) within the 10% tolerance of the german data you used. If we assume a comparable error (10%) in the russian data the devs used, your source and their's could be compatible and respectively measured correctly with the means of the 1940s. There might be just these 10% errors, maybe just in different directions. And this could lead us to the 2% error the devs tuned the 190 according to their data... I know that this is just one possibility. But it shows me that this is still not our smoking gun.
Crump Posted March 1, 2015 Posted March 1, 2015 (edited) Yes it does. As does the variation of AoA of the stabilizer, and yes (to the previous question), both can be combined to increase the pitch authority that you would have if using only the elevator (except if a specific mechanism disallow it). Better don't ask Crump to explain, you'll be lost in no time... It does not increase the Clmax of the wing.... It will not increase the turn performance of the airplane. End of story. Edited March 1, 2015 by Crump
JtD Posted March 1, 2015 Posted March 1, 2015 (edited) Just a thought.. According to this test the BOS 190 is (in the extreme case) within the 10% tolerance of the german data you used. If we assume a comparable error (10%) in the russian data the devs used, your source and their's could be compatible and respectively measured correctly with the means of the 1940s. There might be just these 10% errors, maybe just in different directions. And this could lead us to the 2% error the devs tuned the 190 according to their data... I know that this is just one possibility. But it shows me that this is still not our smoking gun. Not meant to be a smoking gun, it's indicative of the trend. Yes, real life performance did vary, depending on test authority and individual aircraft performance. But over the years I've read maybe 100k pages of mostly original documents on the Fw190 and I do think that the figures I used are fairly representative for the model, with data in second gear ranging from about 11m/s to about 13m/s. If Russian tests disagree, they are probably wrong. I know several figures from Russia, but all I know make sense and fall into the picture. For instance the one attached was tested at a higher weight (3989kg instead of 3850kg) and reached 11.2m/s average in second gear. Which would again be around 12m/s if corrected for weight. Anyway, several times now I've requested the developers to share their sources, at least name them, so we would know if the problem is with the FM or with the target performance. It would also help to get some proper tools for FM testing, because as you can see from the individual data points on my chart, that there's a lot of variance. Right now, as a trend, it looks as if it is underperforming. It's hard to quantify, but I do get other planes to exceed historical climb performance as far as I know it, so as a trend, it is pretty safe. Another m/s or so would not make the FM less accurate, though I can't say if it needs another 0.8 or 1.5. Edited March 1, 2015 by JtD
NachtJaeger110 Posted March 1, 2015 Posted March 1, 2015 (edited) Okay JtD, I'm convinced you're right. We now should focus on how to summarize this as self supporting argument for the devs. you should especially name the "100k pages of mostly original documents on the Fw190". But I assume you have already sent dozens of such arguments to Zak or the others via PM? Edited March 1, 2015 by NachtJaeger110
GOAT-ACEOFACES Posted March 1, 2015 Posted March 1, 2015 @ACEOFACES thank you for a considered and thorough look at my point of view. Not a problem at all! In that I totally respect your point of view and option. The points of difference seem to come down to two main issues. 1) While not ruling out qualitative information in theory, you claim that in practice quantitative information is more reliable for a variety of reasons and easier to test. Agreed.. As noted, I think qualitative data is great for flying qualities.. Also as noted, I think qualitative data could be used for 'other' aspects of FM validation if (<-- for the reading challenged here, note the use of the word if) you were able to do what has never been done, but everyone says is doable. That being taking all the pilot accounts from all the WWII after actions reports, letters, books, etc, than and come up with some sort of criteria for what gets used and what does not get used, than compile a data base from the data you used, to ultimately come up with some 'values' that can be used for in-game testing. But your not done yet! You now have to adjust your data base to take into account all the WWII after actions reports, letters, books, etc that were never written by all the pilots that died in combat and thus never got a chance to write a WWII after actions report, letter, book, etc. And while your trying to come up with a way to do that, you should realise that the data base you currently is a skewed statistical data base that favours the victors of the battle. In that most if not all of the data used in your statistical data base was written by the pilots that survived. Sort of like history, it is written by the victors! And we all know how biased that can be! So, in summary Is it doable? Yes.. Has anyone ever done it? Nope.. Will anyone ever do it? I am not going to hold my breath waiting for it! I understand this point of view in general terms, my greater skepticism stems from my exposure to economics and finance, where such a belief can lead to rapid bankruptcy. Well, that is because your trying to predict 'future' events based on 'historical' events.. That is not what we are doing here when we use WWII quantitative test data to validate a FM. The particular case that keeps cropping up - the Fw190 in an E. Front setting - seems to have far more of the problems of limited data sources, possible bias and lack of opportunities to test. Hence qualitative information that might be redundant if the quantitative data set were richer may be useful. Maybe You asked me to provide an example: I refer you back to the supercharger gear controversy. We had quantitative graphs aplenty, but not in the form that could answer the question of what caused the gear to change. Some other graphs would have solved the problem but we did not have them. Fortunately we did not need them anyway, because we could look at descriptions of the mechanism and see how it worked. I have to disagree with this example of being a qualitative one, it is a quantitative situation contained some disagreements on the basic definitions. Sort of like the P-38J rate of climb example I gave you, where the data was there, Oleg and I were just disagreeing on what the data represented. Only after a few emails and examples was I able to convince Oleg he was wrong and I was right. 2) More on the testing: as I said to Rama above, I cannot see what scientific testing is being done here, since no new data appears to be generated. A model is being fitted to a data set, then tested for internal consistency, fine. But what new predictions are being produced? How can they be tested? Where is the science? Of course you want your model and results not to violate any laws of physics etc. You kind of lost me there.. So, let me respond to what I think your saying, than correct me if I was off base.. With regards to testing and you not seeing 'new data'. I don't expect 1C/777 to provide us the results (new data) from their in-house testing.. It would be nice! But sadly what with the forum members we have, it would most likely cause more problems than it solved. Heck I would be happy if 1C/777 just provided the data that they ultimately used! Sort of like HITECH creations does by providing the top-speed-per-altitude and rate-of-climb charts they used. Note, not the originals sources! But a summary of the data they used. Than we could at least see what 1C/777 target for the FM was! Which would make allot of the FM debates here go away. For example, JtD did some very nice ROC testing of the Fw190, and he concluded the ROC is about 20% lower than it should be.. Than it should be.. Retaliative to what? As you noted, there are conflicting quantitative sources out there! So, of those, which one did 1C/777 pick? Maybe 1C/777 picked one, or averaged a few of them to obtain the ROC that we currently have. In English, there may not be a BUG in the Fw190 FM, it may have the ROC that 1C/777 intended for it to have. And now it is just a mater of the forum members to provide data to 1C/777 to convince them that the ROC 1C/777 picked is not a good choice. Sort of like I connived Oleg to increase the ROC of the P-38J in IL2. Or it could be a bug in the FM.. Hence my claim that Fw190 FM building is more of an exercise of historical reconstruction than a scientific program. As noted above and before, there is defiantly some detective work that goes on when looking at quantitative data! But as noted before, at least with well documented data you can typically figure out what was going on that made it different! A good example is the P-38J example I provided on the ROC time to climb values. Where Oleg thought the time was min:sec instead of fractional min time. Something that is nearly impossible to do with qualitative data! Finally, you may need to count organs to tell if the hot looking "girl" is indeed as advertised: I can tell at a glance. Quantify that! You have never been to Las Vegas have you?
GOAT-ACEOFACES Posted March 1, 2015 Posted March 1, 2015 Better don't ask Crump to explain, you'll be lost in no time.... +1
JtD Posted March 1, 2015 Posted March 1, 2015 (edited) Okay JtD, I'm convinced you're right. We now should focus on how to summarize this as self supporting argument for the devs. you should especially name the "100k pages of mostly original documents on the Fw190". But I assume you have already sent dozens of such arguments to Zak or the others via PM? If I name 100k of pages, should I reference each page separately? It's of course nowhere near all performance related, it's mostly technical descriptions, but it helps to build a good general understanding. This topic marks my biggest effort in BoS FM analysis since Han's episode on the forums in October last year where he essentially told us we're all stupid, so: No, no PM's sent. I'm just doing this out of interest in the subject. I'll leave the PM business to those who are doing this because they want the game to be changed. Edited March 1, 2015 by JtD 1
GOAT-ACEOFACES Posted March 1, 2015 Posted March 1, 2015 I've done a more detailed climb test now, still combat/climb settings at ~270 km/h indicated air speed with standard loadout. Reference is the same Fw190Aa-3 @ 3850kg. At least at 270 km/h, it climbs worse than it should, by about 1m/s throughout the altitude range. Which is about 6% at low, and near 10% at medium altitudes.Now that is some fine testing right there! For all you want to be test pilots, take note here!
Rama Posted March 1, 2015 Posted March 1, 2015 It does not increase the Clmax of the wing.... It will not increase the turn performance of the airplane. End of story. At high speed in a turn, even at high inclination, the stall AoA will not be reached with the max pitch authority given by the elevator only. Increasing the stabilized negative AoA will increase this authority (until the wing AoA does reach the stall AoA). ... so yes, it does increase the turn performance of the airplane at high speed... End of the story. PS: I don't ask you to understand.... 1
=LD=Stache Posted March 1, 2015 Posted March 1, 2015 My impressions about the current Fw 190 : Too little elevator authority , it locks up at speeds around 700 kph (a dora could reach that in level flight ). The fw 190 was known for it's highspeed performance, at high speed it could outmanouvre spitfires and the like by using it's awesome roll and good overall controll , not in BoS atm ,where you do not have enough elevator authority to keep the nose down as speed builds up in a dive (this plane was known to be great at BnZ ) so the main mode of attack is pretty screwed up atm, so you have to keep speed at below 700 watch making you easy prey for all russian fighters wich seem to perform superbly in dives and climbes and can keep up with a diving 190 with noproblems. In short the 190 is dumbed down in it's vertical maneuvering ,while the russian planes seem to do better then i remember ,the horizontal was always in favor of the russian planes and somehow they managed to make the snapstall in the 190 even worse (more like a A8 version) so the horizontal performance is also worse then before . atm your bettter of with any of the 109s
GOAT-ACEOFACES Posted March 1, 2015 Posted March 1, 2015 At high speed in a turn, even at high inclination, the stall AoA will not be reached with the max pitch authority given by the elevator only. Increasing the stabilized negative AoA will increase this authority (until the wing AoA does reach the stall AoA). ... so yes, it does increase the turn performance of the airplane at high speed... Interesting Thanks for the explanation! It's nice to get feedback from real pilots! End of the story. PS: I don't ask you to understand.... Some advice.. Don't bother fighting fire (Crumpisms) with fire (Cumpisms).. I learned long ago, never to wrestle with a pig. You get dirty, and besides, the pig likes it.. I know I know, I don't always follow my own advice, but it is still good advice!
Crump Posted March 1, 2015 Posted March 1, 2015 At high speed in a turn, even at high inclination, the stall AoA will not be reached with the max pitch authority given by the elevator only. Increasing the stabilized negative AoA will increase this authority (until the wing AoA does reach the stall AoA). ... so yes, it does increase the turn performance of the airplane at high speed... End of the story. PS: I don't ask you to understand.... Think about this statement. Moment about the CG = Coefficient of Moment X Reference Area X Dynamic Pressure Assuming subsonic incomprehensible flow, the faster one goes in an airplane, the more effective the controls. You are confusing trim forces the pilot feels with what the control surface is doing. Once more, turn performance is purely a function of angle of bank and velocity. Adjusting the trim system tells the aircraft to remain at the speed the pilot has already told it by the control surface deflection. It does not provide "extra deflection". You are in good company with this error in thinking btw. Focke Wulf had to explain this very same issue to Luftwaffe pilots on at least one occasion.
BlitzPig_EL Posted March 1, 2015 Posted March 1, 2015 (edited) I was not sure that BlitzPig's words were addressed to me specifically. They were not, it was a general observation of years of reading FM posts. I think we can all agree that there are those on both the Allied and Axis "sides" of things that come at virtual aircraft FM discussions with an agenda, be it politically based, or just a desire to have the Uber machine to become an online ace point whore. We can all learn from civil discussions. I know I do, as I am not an engineer, just a guy that has to clean up their messes at the pointy end of the customer service spear. Like most I just want to see the aircraft in a sim modeled accurately, and let the chips fall where they may when it comes to virtual combat. I will interject one more thing that I believe bears on the discussion. Historical outcomes. I think that this is what gets most fans of a particular aircraft or "side" in a twist. We all read about how this ace or that plane dominated, and yet we don't see that in our sim world, so we blame the developers and the FMs. The thing is that no sim or user made mission or scenario can come close in any way to duplicating the circumstances in play at the time of the real events that are being portrayed. For example, those of us here, by and large, have more "combat" hours than any pilot in any air war in the history of military aviation. No matter the time period being modeled, we come at the problems we are given with the experience of for some, decades of "combat" flying. Do you not think that this won't skew the results we see in a sim? How could we possibly make say, the next scenario we will get from 777, the Battle of Moscow, realistic and with an outcome that matches the books? Lets assume we still have 32 slot servers for the sake of argument and further assume we have even numbers on both sides. OK, the VVS will have to limit it's 16 players to ONE pilot with more than one year of virtual combat experience, the other 15 will have to be n00bs with little to no training, and furthermore, the use of voice comms and text chatter will be banned. For the Luftwaffe you will have 8 very experienced online aces, and 8 experienced combat wing men that will funnel all the n00bs to the aces for the coupe de gras. Of course the Luftwaffe pilots will have full use of Team Speak and text chatter. Also for the first day of the campaign, two thirds of the VVS will have to stay on the ground and wait to be bombed or straffed. See what I'm getting at? Real world outcomes cannot ever be a yardstick to judge flight models. We have to understand that if we ever want to move forward in a constructive way when analyzing virtual aircraft performance. Edited March 1, 2015 by BlitzPig_EL
Rama Posted March 1, 2015 Posted March 1, 2015 You are confusing trim forces the pilot feels with what the control surface is doing. No I'm not. The force the pilot can apply to pull the stick is part of the max pitch authority. No pilot is superman. Now if you ment that you could do the same with trim control surface on the elevator, yes you could. But it's much more efficient with changing the stabilizer AoA, and not only to reduce the drag. It also allows to get the same moment with a lower elevator deflection, thus reducing the deformation of the govern and the potential aerodynamic effects induced by this deformation. 1
Crump Posted March 1, 2015 Posted March 1, 2015 No I'm not. The force the pilot can apply to pull the stick is part of the max pitch authority. No pilot is superman. Now if you ment that you could do the same with trim control surface on the elevator, yes you could. But it's much more efficient with changing the stabilizer AoA, and not only to reduce the drag. It also allows to get the same moment with a lower elevator deflection, thus reducing the deformation of the govern and the potential aerodynamic effects induced by this deformation. Now you are saying something completely different from: At high speed in a turn, even at high inclination, the stall AoA will not be reached with the max pitch authority given by the elevator only. Yes a trim system is designed to relieve the control forces. That has no effect on the ability of the elevator to reach stall AoA at any speed. In fact Rama... If you take pilot's lessons your instructor should at least demonstrate the danger of overuse of the trim in slow flight as part of learning about stalls.
Rama Posted March 1, 2015 Posted March 1, 2015 Now you are saying something completely different from: No, I'm not. No elevator move by itself. It's either a engine commanded by the pilot, or more commonly in WWII by a pilot in a plane.... "So max pitch authority given by the elevator only" implicitelly include "when actionned by the pilot". Considering, like you seem to do, the theoretical effect of an elevator who could magically reach max deflection or close at high speed is totally useless in a discussion about trim....In fact, if it could in real WWII planes, without any effort of the pilot, no trims would have been needed, and by consequence, no discussion about trim would happen in any forum. So, sorry not explicitelly writting all the obvious... I forgot I was talking to you. If you take pilot's lessons your instructor should at least demonstrate the danger of overuse of the trim in slow flight as part of learning about stalls. I'm a pilot, are you? Doesn't matter anyway, in any case you can keep for yourself this kind of "pseudo-lessons". 2
Leaf Posted March 1, 2015 Posted March 1, 2015 (edited) Rama, I think your tone can sometimes come off a bit rude... Just saying. Moderators can and should take part in discussions, but in a way that sets an example to everyone else. I mean, this So, sorry not explicitelly writting all the obvious... I forgot I was talking to you. was really not necessary. Especially since Crump is [afaik] an engineer and not just some random FM critic. Edited March 1, 2015 by 19te.Leaf 2
BlitzPigVoidcracker Posted March 1, 2015 Posted March 1, 2015 JimmyJiro touched me in a naughty place back in 02 Did it make you a better virtual pilot?
Bearcat Posted March 1, 2015 Posted March 1, 2015 JimmyJiro touched me in a naughty place back in 02 Uhhh oh .. There goes the neighborhood.. Hey!! Whusssup dog!
Crump Posted March 1, 2015 Posted March 1, 2015 was really not necessary. Especially since Crump is [afaik] an engineer and not just some random FM critic. Thank you. I only post trying to help you guys out.
CUJO_1970 Posted March 2, 2015 Posted March 2, 2015 Handle the FW190 like you are disarming a nuclear bomb that could explode any second and you should do just fine.
unreasonable Posted March 2, 2015 Posted March 2, 2015 I allready told you that the wording of this sentence was probably poor, and it may reflect something not intentionnal.... now you may off course refuse to believe me. You're making hypothesis about how the FM builder work then draw hypothetical conclusions from them. Actually he's using different data to build the models and to test them. And the aircraft performance test data are not used to build the model, but to check the outcome of the model. It's not magic. Binary values are values, and can been combined to form complex quantitative models. BTW, the whole FM is working on a computer, with a code based made exclusivelly of combined binary values and logical truth fonctions.... Fair enough on the first point: if you can accept that I do in fact know a fair bit about how science works, I am prepared to believe that your tone was a result of your writing style in a foreign language, not a deliberate attempt to disparage. The second point I think is highly questionable. Unless you are defining the model only as the general equations prior to any plane specific values being inserted. These equations obviously reflect some simplified idea of the many physical processes at work. Then some empirical values for plane 1 are inserted: things like weight, size and shape of various elements, depending on the complexity of the model. So I agree at this stage the preferred historic performance data may still not an input. (Except perhaps thrust in various engine states? Without actually seeing the FM it impossible to know if any of the performance figures are not in fact inputs). The key is the next stage: the model is run and various performance outputs are compared with the historic data for plane 1. At this stage it might be possible to think of the process as a scientific hypothesis/test/falsification process. But suppose the results of the run do not correspond to the preferred historic data set? Do the modelers conclude that their equations are wrong? I think not. They go and "tweak" the fudge factor variable they have put next to every computation until they get the outputs that correspond to the previously selected preferred historic data set. Then lets us suppose they move to the next plane. They repeat the process of inputting relevant non-performance empirical values and run the model again using the same "fudge factors" as for the plane 1. They then examine the performance outputs and compare them to the preferred historical data set for plane 2. Suppose no fit. Is this taken as a refutation of the model? Obviously not, the "fudge factors" are just adjusted for plane 2. I do not see this process as a scientific exercise of the generation and testing of a refutable hypothesis at all, since there seem to be no possible outcomes that lead to the model being rejected after the process of fitting has been completed. I am not sure what point your last comment is supposed to make: the issue is not how a PC works, but whether it is possible for qualitative statements to contain useful information. The fact that a PC needs information in a particular form is irrelevant.
StG2_Manfred Posted March 2, 2015 Posted March 2, 2015 The same like discussion was going on about the oscillation of the planes where some here defended the behaviour till the very end. Now it is corrected by the devs and you could ask who was right and who was wrong...
reve_etrange Posted March 2, 2015 Posted March 2, 2015 (edited) Unless you are defining the model only as the general equations prior to any plane specific values being inserted. These equations obviously reflect some simplified idea of the many physical processes at work. Then some empirical values for plane 1 are inserted: things like weight, size and shape of various elements, depending on the complexity of the model. So I agree at this stage the preferred historic performance data may still not an input. I believe that's how it works - since it is a physics-based model. An example input could be engine power delivered to the propeller. The key is the next stage: the model is run and various performance outputs are compared with the historic data for plane 1. At this stage it might be possible to think of the process as a scientific hypothesis/test/falsification process. But suppose the results of the run do not correspond to the preferred historic data set? Do the modelers conclude that their equations are wrong? I think not. They go and "tweak" the fudge factor variable they have put next to every computation until they get the outputs that correspond to the previously selected preferred historic data set. I do not see this process as a scientific exercise of the generation and testing of a refutable hypothesis at all, since there seem to be no possible outcomes that lead to the model being rejected after the process of fitting has been completed. I'm not sure you're giving the supposed approach enough credit. Any unphysical fudge factors could be equivalently included as changes to the input parameters for the aircraft in question. If unreasonable inputs are required (e.g. engine power must be increased by more than a few hundred Watts) to match target data, the modelers could very well conclude their equations are wrong, in the sense of being oversimplified. At that point, assumptions can be relaxed, additional series expansion terms added, etc., and the process repeated. No doubt many such iterations occurred during the game development. Although we are left having to decide when to reject input parameters, target data points, or physics simplifications, we aren't at all forced to overfit the target data. The same like discussion was going on about the oscillation of the planes where some here defended the behaviour till the very end. Now it is corrected by the devs and you could ask who was right and who was wrong... Of course, we don't know how the oscillation damping was altered. Perhaps each aircraft has its short-period yaw oscillation damping coefficient explicitly defined as a function of angle of attack, and these functions were increased. Perhaps this oscillation emerges naturally from the physics simulation, and it was physical constants or aircraft design parameters that were tweaked instead. Edited March 2, 2015 by reve_etrange
Rama Posted March 2, 2015 Posted March 2, 2015 I do not see this process as a scientific exercise of the generation and testing of a refutable hypothesis at all, since there seem to be no possible outcomes that lead to the model being rejected after the process of fitting has been completed. Again, you're making hypothesis about how the models are developped and tuned, then draw conclusions from your hypothesis. I understand that since you don't really know about how Anton Petrovich is developping and tuning the FM, you'll stay doubtfull. So be it, I Can't help more. I am not sure what point your last comment is supposed to make The point is that a yes/no test on a precise and detailed proposition, which error margin is known (or at least overestimated), is a quantitative data.
unreasonable Posted March 2, 2015 Posted March 2, 2015 Again, you're making hypothesis about how the models are developped and tuned, then draw conclusions from your hypothesis. I understand that since you don't really know about how Anton Petrovich is developping and tuning the FM, you'll stay doubtfull. So be it, I Can't help more. The point is that a yes/no test on a precise and detailed proposition, which error margin is known (or at least overestimated), is a quantitative data. Ah yes, the secret knowledge ploy. You show when a hypothesis can be derived from the FM and tested experimentally against real world observations not previously known to the FM designer, leading to some changes in the FM that are not "fudge", and I will agree that this is science. In the mean time you are right that I have to stick to the interpretation that seems more probable. On your second point, I attach http://regentsprep.org/regents/math/algebra/AD1/qualquant.htm This is not so much for you Rama, but for anyone else interested. Quantitative and qualitative statements are qualitatively different: the difference is that qualitative statements are not quantitative, this is part of what qualitative means! What is true is that you can assign quantitative prior probabilities to the truth of any statement, perhaps this is what you are meaning. But this is just an expression of uncertainty about what you will find in future checks. The statement is either true or false, you just do not yet know which. It is also true that you can count them, hence you can do quantitative analysis of qualitative statements if you wish: but again this does not change their nature.
NachtJaeger110 Posted March 2, 2015 Posted March 2, 2015 If I name 100k of pages, should I reference each page separately? It's of course nowhere near all performance related, it's mostly technical descriptions, but it helps to build a good general understanding. This topic marks my biggest effort in BoS FM analysis since Han's episode on the forums in October last year where he essentially told us we're all stupid, so: No, no PM's sent. I'm just doing this out of interest in the subject. I'll leave the PM business to those who are doing this because they want the game to be changed. Ok then I'm a little afraid that this great stuff might get lost in this big dicussion without reaching the right devs... shouldn't anybody who had contact with them previously send it, maybe to Han and Zak? They promised that they would look at it if is sent the right form, which they specified. 1
Recommended Posts