scotsman

Registered Users
  • Content count

    6,178
  • Joined

  • Last visited

  • Days Won

    27

Everything posted by scotsman

  1. As I am no longer subbed at the moment and I can't see all that has been written on this subject but I thought I would provide some clarification again on what was generated and provided. Let me set one thing straight first. The unit cost is in -not- driven by the man hours to produce a specific platform. That's just one of many cost elements. How that got started I have no idea. What I handed over was a formula set based on system characteristics that derived the cost number for that platform on an annual basis in then year dollars against a standardized labor force. That cost basis could and did change year to year and that too was in the spreadsheet...which generated the cost of the unit from 1940-1945. Its up to CRS to decide whether to use a single averaged cost or the cost specific to that year. The data is all there...as is the data to test the modeled cost vs. the known actual costs for the platform. That is in fact how the model was refined, progressive iterations lowering the variance vs historical. I ran countless tests to determine statistical significance, and what I generated accounts for the most statistically significant variables...things like weight, gun caliber, etc. Armor is implicit in vehicle weight. When I used specific values for some characteristics they failed statistical significance tests vs cost. Gross vehicle weight on the other hand was highly correlated to cost. If the variable failed to test as significant it was dropped from the data and the equation. The model was derived from ALL infantry weapons, artillery and armor of WWII. It is not a national dataset...and because of that you can and do see variance in some cases between the historic cost and the model cost. This too was tested...and I can assert that in most cases the differences in the two are minor (usually less than 3% variance and frequently right on the money with the documented then year costs) The system can and does cost out things for which no historical information is available. Conversions based on captured chassis are calculated on a new production basis for that year, as CRS can not say how a campaign will flow, and they certainly have no idea of how many vehicles were captured and available for conversion. Any variance is listed in the spreadsheet. I have my own opinions on the 'best' cost number to use... The one thing you NEVER do in this is rely on anything arbitrary or capricious. If you test the old CVC values statistically and find they do not reflect campaign outcome and are not correlated to it...then there is no basis for claiming 'they worked' or affected the outcome. They never worked and had no impact on the outcome of the campaign. That's just a mathematical fact. If however you test the summation of those values as a reflection of total combat value for a nation, with the presence of a significant imbalance in total, and that DOES test as statistically significant to campaign outcome...then you know you have a major TOE/Pool problem. It cost the game players and lead to some pretty bad balance perceptions on the part of the player base. That is the 'was has been'. The only question going forward was what should be... I created the tool but I dont know what CRS will use going forward as there are several cost option paths available within it for TOE...the basis to all being the costs per year are known and the variance against real costs is known. I did finish the aircraft after I left and I forwarded that along with a new model for aircraft damage. I'll close by saying that you --NEVER-- include human effects in determining weapon system performance. The best anyone has ever manage to do is a variance in task performance time based on crew skill level. No-one has ever successfully quantified human effects (or player effects) like morale (or player skill) in such a way as to be repeatable and meaningful. The US Army has its own system for arrived at numbers like that...and I don't have X years to formally educate people in the process...and CRS doesnt have the means to use it in any case. I dont have the time or inclination to educate everyone on how that is eliminated in army analysis... CRS can always weight the costs/cost pool based on player population dynamically or statically,...putting a developer finger on the scale to reflect game realities without touching the historic basis for the other calculations of the individual platforms. Any notion that something like a historic campaign K/D can do that better is nonsense in my opinion...and it certainly would be hard pressed to do it dynamically without risk to the game. The bottom line is the system has to be divorced of some things while reflecting the reality of the game in which it lives. I believe that system does that...but what branch of it the Rats choose to use is ultimately up to them, and I suspect it will remain proprietary to the game and won't be revealed publicly. It took months to generate the data and test it, refining the model to be as lightweight and meaningful as possible, all thewhile passing statistical muster at each iteration.
  2. They aren't. The budget for each TOE and weapon pool is equal without connection to any physical plant or factory. That could be changed - that was the original design of the game - but it was never fully realized or implemented. The labor forces and labor rates are standardized, as is the production learning curve. None of that is constrained by the presence or lack of physical plant. In that sense it's no different than the current TOE concept. What IS different is that every weapon of every nation in WWII was part of the cost estimation database, and the resulting cost model. It's not a guess on a dev's part as to the value of any unit vs any other unit. It's hard economic data. Could it be grown to a full economic model? Yes but that's not the intended use of the tool. The data is there to include factory floor space calculations. It was designed to allow weapons and systems trades within a equal budget among the TOE types of the combatants....and get those TOEs to more closely resemble the weapons mixes employed in each year. The national budget(s) should reflect player population and can be scaled up or down increasing the size of the spawn pools without altering the underlying data. It allows for the easy introduction of new equipment (say panther or a new aircraft) without having to redo everything and rebalance it. It could be used to make the spawn pools dynamic based on population rather than static...but that's not an intended use when it was built. Id love to do a full economic model with attackable supplies..even if it was only the tactical supply to begin with. I have the various consumption rates, supply volumes etc, for each TOE. Higher on the list though (my own personal list) would be audited fuel burn rates for everything. That's how you stop mechanized forces or attack with them.. fuel is the biggest single tactical constraint with ammunition much lower in terms of logistics demand, (unless we are talking non-mechanized forces which can live under a lower fuel burn) if we had a better fuel model and attackable resources (POL ppo?) then fuel state begins to matter. You can't attack without fuel even if you have an acceptable equipment state. in summary I'd think it better use of time to focus on gameplay and a better tactical environment than constructing the full economic model.
  3. There was >>no defect or reliability penalty imposed anywhere<<. The data is there but there is no imposition of reliability or production quality on anyone. I'm not skipping anything that I am aware of. The cost basis changes every year gents and is based on the number being produced and the learning curve. Again using a US example....that 60K 1941 P-40 was 44k by 1944. That's the historic example number. Not sure of the point you are trying to make but if you're going make any accusation with regards to side bias you'll have to have another crack at it, as what you suggested isn't the case. All sides have equal budget and there is no instance of quality or reliability influence imposed anywhere. What I said was that you could, on either side, alter the prices based on data present but not included in the price model. In short changing the price to reflect that stuff. That was not done. What I also said is that public domain pricing may not reflect the real cost of the aircraft in a combat ready condition. I gave several examples of that...and why you have to be cautious about quoting prices without knowing the underlying data. As for which cost is used that's up CRS, it can cost average or use the specific annual price. Not my call. Again - the ultimate composition of the TOE is up to CRS...the historic production data for each aircraft data is in there along with the price data from the model AND the documented historical price. Those two are not the same for the reasons already laid out. Everyone has to be costed on an equal basis of learning curve and labor force... What I would care to do is point out that quality and reliability is an issue for both sides, but not part of the game's assumptions.
  4. All part of making a smart TOE...but you still need to do the basis in trades. The Infantry TOEs for example were not static...they change with time as new weapons come in and old weapons are retired.
  5. my pleasure....and you have it exactly right
  6. The aircraft live in a different setup... Again its COST - not man hours. Where this man hour thing is coming from I have no idea. I know the man hours and even the factory floor space required to complete each element of the aircraft. I do think you are likely barking up the wrong tree entirely on this but I'll try to give you an example of why I think that. I'm going to use some open source numbers here and --not-- specifics out of the tool as thats proprietary. You have the F4 man hours wrong by the way...the German contract performance data for the f4 in 1941 is somewhere north of 7,000 hours. Yes - I went into that degree of detail...I looked at individual contracts and the performance for that contract for everyone. The next thing to remember on aircraft is that the cost basis between the powers is VASTLY different. The allies added quite a bit of cost in testing and in other ways (additional equipment) to the end cost of their aircraft. The germans did not. I did not penalize the germans in costing for the number of aircraft rejected as not serviceable on completion. I assume you wouldn't want that done as that's a quality measure and we largely ignore that in the game. That number is there though and its significant. It could be used as a price penalty as those aircraft were junked and never flown. An F4 in 1941 had a 'documented' cost of 56k RM (not a combat ready cost)...with the cost changing over its life cycle and production learning curve in that year. Note that this early on in the war the slave labor issue really isn't a factor yet..the workforce is still primarily German with some foreign reinforcement. A Spitfire Mark V (6479 built out of a total of over 20K) did take about 13,000 man hours to build but that's just one element of its cost chain. A hurricane at that time ran about 10K hours with lower material costs, and earlier marks even less than that. The British contract costs don't represent combat ready aircraft either as they are missing additional items. Cost of flight tests, radios, armament etc and other items are added in. That British contract number is NOT a complete cost number and that's frequently an issue with quoting public domain cost numbers without understanding what's in them. The average cost of the spitfire across all marks was about 38k in then year dollars (again minus necessary equipment)...which naturally includes all the later marks which were far bigger and heavier aircraft. The estonian export order for spits came in at about 50K per aircraft, to include some spares. If you looked at documented US costs a P-40 in that year (1941) it might run around 60k then year dollars combat ready. The point is that the tool reflects the aircraft cost properly at its time of completion to the best level of detail I could manage based on the assumptions I have already given everyone. If one nation had 'better' performing aircraft at that time is immaterial to the tool. That's a TOE balance issue within the total TOE cost constraints. The tool has all the aircraft from other nations in there too. The aircraft are what they are...when they are introduced in game play is a TOE is up to CRS....all the tool does is the costing to allow for trades within an enforced economic bottom line....though the historic production numbers are also there. Now as to why I think you have it wrong - if you are looking at it from the German perspective. The cost of the spit is what it is, but as you well know the spit isn't the only fighter in the front line in 1941. You should be FAR more worried about the cost of the Hurricane vs the 109. There is a reason the Brits focused on them and its cost. It's a half generation aircraft and is significantly cheaper than the spit is or the 109. You are too narrowly focused on 1 vs 1 and not on the force balance. The tool reflects changes in armament as that changes the aircraft gross weight and other features. Lastly - the aircraft side has the same underpinning in assumptions as ground does (a common labor force with standardized learning curve and a cost model built on data for all aircraft in WWII from all nations) The price out of the tool is not the same as these historic cost marks I gave you. I'm not going into details there as thats proprietary to CRS. Suffice it to say that the point you are raising there is largely mitigated by the assumptions I've given you. Generally speaking I was very happy with the tool projected costs as the database it was constructed from was LARGE sample space. Lots of data means very good detail on production cost in the assumed environment specified. One additional final personal comment...I spent a lot of time on the phone with RAF and made them drag out I don't know how many contracts and contract performance documents. The tool has data in that regard that has never been made available to the public or even published to the best of my knowledge. No stone was left unturned....Its complete overkill for the intended purpose actually, and has far more detail and work than is really necessary. That, unfortunately, is a personal failing I have.
  7. I think I provisioned it adequately to enforce balance within pools of weapons by type (small arms, armor etc) . I can't go into the specifics there as that relates to its use by CRS. There were several included options there. Anything within a TOE is possible including enforced 'balance' on the infantry side. The cost of small arms is paltry compared to the cost of the heavy weapons so I really don't foresee any issues on the infantry balance side. The system lives within the current game so its the responsibility of those designing the TOE to do so properly. This simply allows the trades to be made...bolt actions for semis or SMGs etc, with each having the proper cost year by year, and the weapons living within specific weapons pools by type. That basically -enforces- trades within pool categories...there is an infantry pool, an armor pool etc. Trading within the respective pool in a TOE is easy..trading across the pools can be done but I really don't see the need or necessity to do that. An early german semi-auto can easily be trade vs a k-98 on a cost basis in a given year. The guys know the game construct and how those trades might affect things like capping etc. I included the basic TOE templates for each side as a function of year. The guys know what should be in a TOE... It's a hammer...I can't guarantee that someone won't hit their thumb with it...I know I wouldn't....some common sense has to be used with it. I trust bmbm to be able to do that if he is working with the tool....and the fact there are several pools should insure that you won't end up with anything nutty... Going forward CRS could even trial new TOE formats in intermission. That way the time isn't wasted...and everyone could get a feel for a trade within a pool. Like it or not? The point is to make it fun for everyone while living within the history - each side having an equal chance to win. What it will let CRS do is add the new equipment as its built transparently while enforcing a uniform bottom equity among the forces. Also the system is the same for everyone for the first time in the game's history. There is nothing arbitrary and nothing left to estimation or guesswork or opinion. Its well documented. I would expect it to continue to evolve over time... if CRS likes the results they get with it
  8. It isn't - you -never- include undefinable, arbitrary or capricious variables in sound system design - nor do you assume levels of player expertise, tactical situation etc. I'm not sure how many times I have to repeat that. 'combat effectiveness' of any weapon system is driven by many human effects...morale, crew quality, tactical proficiency, etc, that are difficult or impossible to quantify..not to mention tactical scenario, weather, etc. Thats only a partial list. Crew quality can be reflected by a time to complete an action with a time distribution but the others are all problematical. The system won't stop you from, for example, including the Porsche Elephants in a TOE, though we all know their tactical shortcomings. The cost is there if the desire to include them is there. It doesn't protect a player or side from making mistakes. It simply provides the frame for any WWII weapon system to be included in a TOE. As far as I know CRS still controls the TOE makeup. It would be crazy to impose human effects assumptions in a TOE on the player base. A good player will be a good player, and a bad one will be bad. If your attack plan doesn't provide for mutual support and your head is handed to you that's your business as a player and a side. If your question is how cost is apportioned in a specific TOE, there are several paths there and I'm not going to guess which CRS might opt for. There are many ways to use any tool.
  9. Plausable as the time period under consideration is pretty limited, 6 years....and the underlying technologies are approximately the same across all nations...though national production means were not. As bmbm has already explained - by setting national budgets equal for each type of TOE you eliminate that as a consideration. You can also equalize total budget if you want to have different numbers of brigades divided among the combatants. Aircraft are a bit more complicated, as you do have a technical shift with the jets etc. Surprisingly those early jet airframes continued to follow established cost patterns as their construction techniques vs prop aircraft didn't change much.
  10. Excel for putting all the data together.. multiple tools to hunt for the best model fits. As I got towards the end I sometimes ran the data set in multiple tools to confirm I had the best solution. I had a go to but I'll keep that the myself - Generally speaking most of the big names can do the job... it comes down to familiarity with the tool. Once you get into neural nets and other AI type solutions it gets a bit more hairy as your results can frequently be determined by how well you know how to employ it. I looked hard there but in the end settled on a more conventional solution. A neural net is something of a black box (or can be depending on the tool used) and I wanted an open solution that the other guys could use.
  11. I agree - no harm in exposing the major cost inputs imho....but not my call
  12. Laugh
  13. Hey Aismov... Well as a stats guy you know there are many ways to skin the cat statistically....but as for determining 'the best' model...the approach I used was to examine all the vehicle or weapon characteristics to determine what correlated vs cost and did a T test for significance. If it tested as significant to the cost the variable stayed for consideration - the next task is to build a workable model that worked for all equipment across all nations. If you can't model it for all systems then it's not much help. Thats easy to do for a small number of variables and increasingly hard to do as the number of variables increased. I fed the raw data and known costs to a statistical program that ran some 130,000 permutations of different types of regression (linear and non-linear) as well neural net and other approaches to arrive at the 'best' model for the known data...as measured by R2 for the total model. That varied a lot during the course of the work. Frequently what tested best on one iteration had a reduced R2 as new data continued to be introduced. In the end it became obvious that out of a large candidate list of variables only a couple were truly significant in driving the costs...and even if the others were included and resulted in little model improvement and R2 - the model itself became increasingly difficult or cumbersome. As always you want the best of all worlds. When all was said and done the R2 across all nations was about 93-94%(often higher in specific areas) for armor, infantry and artillery weapons (<hint) ...probably the best we can hope for lacking some historical data and keeping in mind that it was done for a standardized labor force and learning curve (number of systems produced definitely is significant to cost). If it had come back at 99% or something I would have doubted my own work and figured I was too closely coupled to the data and not nationally independent so to speak. We all know the dangers of excessively high R2 that you must guard against. The modeled vs known costs is right there in the sheets so there variance is known for every system made in WWII. Given the differences in history, national basis, etc, it was surprising to see how consistent the pricing was as a function of those variables. Things that required lots of machining costed more...as they did in real life. Stamped metal small arms were much cheaper, as they should be. Low volume weapons were more expensive than high volume weapons and so on. A great example of that was some of the early german semi-auto rifles. Very very low production numbers and high cost both historically and coming out of the model. I wont go into all the little things I learned by doing it...but what I can say is that its sound mathematically, unbiased, and scalable in a static or dynamic application. Theoretically, that data can be linked to player population and use whatever game only variables you might want to include without altering the underlying data. (hence no need for arbitrary and capricious stuff like K/D - the data behind those K/D changed with game history and there are a host of other statistical issues that make that a total non-starter) You have to think of your data basis first if you are going to derive a robust system. I'm ok with them publishing the numbers by nation/system but I doubt it will happen. Perhaps a reduced dataset for specific in game items tgo illustrate? Not up to me...I believe in transparency. As for RHA vs FHa etc...no...the model didn't end up reflecting armor type...in the case of FHA that's really just an additional process in construction and materials per say. The end price of the vehicle didn't seem affected by that. (yes I did screen for it but I dropped it from consideration early on based on testing) Hope that helps answer the question(s)
  14. One final comment....the test of success IS NOT a historical 1:1 campaign win ratio - it's whether you have a transparent system that increased the player base and gives both sides opportunity to win. Transparent, fair, enjoyable game play is how you 'win' as dev.
  15. True - but you know very well my reason for departing had -nothing- to do with being able to 'take heat'. I've taken far more heat in other careers than anything the game could ever dish out. I believe in the vision...always have...but I also believe that we (the royal 'we' here) have had long enough to effect change and achieve the original vision...to at least some degree. Of course it's resource constrained...it will always be. The player base, however, has been incredibly patient in that regard...the resource pool has increased. I am unaware of a single instance in which the information you were given with regards to armor or ballistics was ever in error. I will admit to making some mistakes in units and math when pushed to deliver quick calculations...lol..human nature. I'm not perfect - no one is. Very glad the system has built in checks to preclude that. That same system needs to be applied uniformly to include leadership. I think what I'm aligned on here is the bigger picture. The chance of getting to a good end game in something less than four lifetimes. I don't have that kind of time any more...many people do not. The team should accept that and act accordingly (the royal we again) The team is far better off than it was when I first came on board. Of that there is no doubt. I have no doubt the potential is there...but the clock is ticking and other titles abound. We all know that. The question is how to best get to an end game that is truthful, honest, and without bias....what our original box promised the player base...well that and suceeding financially. No news in any of this...performance and accountability are a two way street.
  16. Any organization has to be able to take heat - even more importantly it has to be able to be honest with itself. Anything less and it ultimately will fail. where you draw the lines is up to each individual...it's ok to have decisions go against you. It comes with business. More frustrating are decisions that go against without any reason or reasonable data to back them. Even more frustrating than that (to me personally) is shooting from the hip without understanding what the impacts of said decisions will be. If any of these become a pattern..then you have to be honest with yourself about the likely success of the endeavor.
  17. The G is in the can - up to crs to decide when they drop it.
  18. Very high on my list but it will take more than me...
  19. It's HEAT has 110mm penetration at 30 degrees - it can kill anything - but I'll check it one more time...
  20. Well by using your own thread...I'm sure you read this: http://forums.thecmp.org/archive/index.php/t-166792.html Canfield's assertion on AP usage for the M-1 does not track with other wartime documents. I wouldn't argue that it was used on occasion and that the airborne folks and others actually preferred to have a basic load of it. For line infantry units? No...big difference between wanting and getting. Some might get it for a specific circumstance or situation, or have some tucked away for an emergency. (I did! ) Its simply economics. Hard core ammo is more costly than FMJ ammo....andbelieve me the bean counters ALWAYS win...frequently even in war time. I served in USAREUR HQ and had access to basic load information going back to WWII and also had info on what stock was available ammo wise at given points in time. Have to agree with the other comments in the thread above that the assertion it was used by rifleman across the board is an urban myth. I'm very open to adding AP back in for the K98 (where documentation exists for AP during the battle of France), the M-1, and other weapons at some point, once we have user selectable basic load capability....but until then...and as we can only have ONE standard basic load for a weapon to represent service during the entire time frame...I'm content to leave it parked as is.
  21. And that folks in a nutshell is the curse of any developer - two veteran players with opposite views of the same issue. On a more positive note we made a test change on the human modeling and the initial results look very positive.
  22. Chasing our frag issue gents... definitely related to the Infantry model...not munitions performance. Some fixes already being tried.
  23. What little I can say so far is this....when we had few large fragments traveling at ordnance velocities...the energy transfer to the trooper model was massive and overwhelming....now that the munitions are where they are supposed to be accuracy wise theworld is different...we have things like .7 gram frags traveling at 800 m/s (mortar frag there) and those frags seem to be interacting with the trooper target differently. Hits are being recorded, but the frags arent transferring energy to the trooper target in the manner is which they should. Generally speaking, 80 joules worth of frags to an individual will create a disabling wound and anywhere from 110-200 would stand a very good chance of dropping them. Lots of testing going on now...but I'm guessing our issue is with the way the frags are bleeding velocity on impact and doing their energy transfer. stay tuned.
  24. Chasing some issues right now...again thinking is that its the data for the human body thats the issue...not the munitions...stay tuned
  25. If you were clipping into terrain and sheltered from frag calcs that's why you didn't die... intentional clipping isn't kosher by the way... If you were intentionally clipped into terrain then ( a) that's not kosher play and (b) you aren't subject to proper frag