While in general I think it does actually make sense in certain circumstances for design quality on a single component to go backwards, this should always be due to some logical reason, and this reason should affect probabilities, e.g.:
- lack of knowledge of underlying science (IMO should be related to how many designs currently use this technology, not just whether it's been researched and/or level of tech (I believe currently, all that's taken into account in this regard is whether we have researched a specific binary tech, and in some cases the level of advancement for non-linear techs).
- lack of experience with similar components for that model (e.g. if we've previously fielded a bunch of tanks with equivalent-spec engines, then this engine should always be an improvement).
- as above, but for similar components on other model types (e.g. first time we're using a double diesel for a heavy tank, but we already field it in medium tanks so we should know what we're doing).
- for the overall model design %, even if designing a new model, previous models of that type should contribute somehow; however it sort of makes sense for this to actually be worse overall, provided this is due to factors such as one of the above (e.g. first time attempting to build a weapon of a certain calibre; first time attempting to use a specific research (i.e. it's one thing to have researched the principles of Gauss weapons (for example), quite another to use them in a tank where previously we'd only used more traditional weaponry); we have a requirement to build a certain model with certain mandatory features, resulting in a compromise on lesser features*.
*Potential tangential addendum: Not wanting to derail the gist of this thread, so this is somewhat tangential (but I'm not sure it needs/deserves a separate thread either, and if implemented it would definitely contribute to resolving the OP's issue): it would be cool if:
- weight wasn't just a "set" weight, but was a weight range (or alternatively, the weight is the "base" weight which could be adjusted based on design %, similar to how firepower and armour works?)
- we could choose to specify which aspects of the design are more important. So in addition to specifying the type of engine/armour thickness/weapon, we could specify we want to focus on "fuel economy", or "movement speed" (or even better: engine power and/or overall weight), survivability (HP), attack capability (even better if this were against a specific target, e.g. "focus on anti-inf").
If the above were done, then it would be possible to, for example, order your designers to come up with a tank that while it uses the exact same component "types" as a number of previous models, and uses the same technology, we'd like to specifically build a tank for the anti-tank role, but it must have a survivability at least equal to previous models, and we'd also like it to most definitely be faster than the previous version since we'd like to have our tanks keep up with our motorised infantry for once. At the end they might end up with a tank that's marginally better at AT than the previous model, has around the same (or marginally better) survivability, but our engineers have done well with the engine and it's much improved, so much faster tank at same weight/engine, with the compromise being that it uses fuel at an insane rate now, and out anti-inf capability is down a fair bit (e.g. maybe to save weight where they could, they removed all mounted MGs or something? I know technically we don't specify MGs on tanks (though IMO we probably should as this affects anti-inf a fair bit I would have thought?)). Also, while technically it's armour isn't any worse, perhaps its survivability against infantry is actually worse somehow due to other weight-saving considerations? Or perhaps the ammo consumption is now off the charts because they came up with a "new" version of an existing model/calibre of weapon with improved AT capability without weighing any extra (e.g. maybe it can fire slightly faster now and that's the only difference?).
< Message edited by TheSquid -- 10/17/2020 5:57:53 AM >