Box of Stardust

So I just want to address this in one of my own posts, primarily so I have my own reference regarding it, as well as my own suggestions on how we might be able to help streamline the submission-review process. Because the ideas in this are quite good. It seems to me that some of the most submitted classes to are the jumbos and supersonics. Not by a huge margin, but the queue is still larger for those two compared to the others. Supersonics especially, but this is, after all, Kerbal Space Program; I suppose that's to be expected. In terms of "problematic" designs, those also seem to be generally spread around, but I expect will be more present in the two previously mentioned larger categories. It's a good night for distractions and rambling, so here's my take on restrictions based on "company progression": Firstly, maybe just disconnect from any initial pricing-related restrictions. Those could come afterwards, and it's already tacked onto the aircraft submission parameters anyways, and should also factor into a company's "first impressions" in tandem with the actual aircraft's performance and quality (because someone could just come along with a relatively expensive turboprop but it's quite good and fair for the price). In terms of submissions, a "company" should start with light, small, or maybe medium aircraft (possible as per Airbus). This puts them in a set of "primary" review queues (per class, as we had before), which judges should aim to address and review in order of submission (or such). The idea behind this being that these smaller/more manageable aircraft are fairly easier to put a review out for, and in my opinion, somewhat more fun of an experience to fly around, thereby making it less of an ordeal for judges to review. It then therefore allows people to get their foot in the door a little faster if they can prove that their planes won't be something future judges will have to slog through. As "company" prominence raises, they can expand their "market lineup" to the larger or more expensive/lucrative aircraft (supersonics, jumbos, even just general large airliners), and their submission allowances reflect this. Maintaining such reputation upholds their submission privileges (in a formal and meta sense). This should ensure that these classes of aircraft that can be somewhat more of an ordeal to review (heavier to handle and whatnot, finnicky flight characteristics, etc.) are generally reserved to "companies" that are known to make reliably flying aircraft that won't be too much of a bother to undertake as a review and would provide an interesting, productive experience. Making a poor impression imposes a sort of submission penalty that makes future aircraft less of a priority to review until they prove themselves able to build reliable aircraft, keeping the combination of "not so productive aircraft of a tedious class to review" to a minimum. That said, anyone is still free to submit whatever they'd like; they'll just be put in an according set of "secondary" review queues. If someone wants to just build a supersonic or whatever, they still can. But they should be notified that they will be put in a secondary sort of queue that judges can select when they want to review it; it's not a priority or anything. So at least, the person is warned that they're not exactly following the bounds of the challenge, but their effort is still respected and, if presented attractively enough, may still perhaps get a judge to take interest and review it. Heck, maybe we could still allow people to submit full-range catalogues, but only place the "starter class" aircraft in the primary queues, and leave the rest in the secondary queues until further reviews deem a possible upgrade for those aircraft to the primary queues. If it goes somewhat according to plan, we'll have a weirdly big backlog of turboprops, small airliners, and some medium liners, and then it should taper off as it gets into the bigger, more heftier aircraft. We could even have judges assigned to certain tiers/classes of aircraft, which could help onboard people looking to start judging. Like, "Hey, you can start helping us judge by going through some of the turboprops!". Could also act as a bit of quality control for reviews, and ensure that the "bigger" projects go to judges of higher tier that are known to be pretty good about posting interesting reviews. Restrictions may need adjusting based on how popular general medium and simply large aircraft are, or even smaller business class-like jets. Though I guess it would be pretty simple to divide up the aircraft classes by "starter company tier" and "experienced company tier". I like how KEA inspires people to just build, and I don't want to take that creativity-inspiring aspect away from the thread. But we also need rules in place to ensure that the thread doesn't die off like it did, and so there needs to be a way to encourage people to actually play the challenge in a productive manner. This is just reiterating the entire discussion that's been going on right now: the queue and review process needs to be manageable such that judges don't burn out quickly and also be easier for new judges to get in so that the challenge can continue along in a healthy manner. (Or we can also take the route other various challenges have done that, by virtue of the challenge, have taken the core community that had formed and create a Discord server...)

I considered bringing up the idea of having "rounds" for entries and classes, but ultimately decided that, even with a lenient period of 2+ weeks, I have enough deadlines in my life and don't want to deal with something like that. As well, I liked the open approach to KEA that anyone could design anything for whatever class at whatever time, making the design process rather enjoyable to approach, as there wasn't any pressing need to really work on a design. You want to design a medium-sized jet? Go for it, it's not like there was a specific entry period for that class of aircraft. Each aircraft entered could potentially have its own set of advantages and disadvantages, and it was up to the judges to decide on how an aircraft could best be used. I like designing craft with an objective in front of me; I feel like I rarely build aircraft for the sake of designing one. I don't think I'd feel motivated to design my entire lineup of aircraft if calls for specific aircraft classes came and went. And the entire design process was certainly educational through experimentation, and I'd rather not reduce that effect by making the entry process more restrictive in terms of requiring a completely focused goal. In this scenario, I think it's more beneficial to encourage free experimentation instead of directing people to a specific design goal that they may meet better if they had gained experience in building other projects. That's just me though. There was also talk about refining the aircraft classes and requirements though, but I think that discussion can be left for last. That said, I think a case can be made for judges coming together and selecting an aircraft as "best fit for role" (e.g., current top pick for long-haul medium capacity, current top pick for isolated airfields, etc.) and awarding the winning builder in some manner. Though, selecting as such is a bit odd with rolling submissions with open submissions for all classes. I do agree that a way to filter out poorly thought-out designs would be nice, but perhaps it might be more productive to encourage well-thought-out designs (not that that seemed to be a huge problem anyways). I guess, my point is, the nature of this challenge is such that we're probably going to get the lesser designs one way or another, and it's just something we'll have to work through. Besides, they do provide interesting material to write the more entertaining reviews. The judging system we were working on, if you recall. It was going along pretty swell, actually; we had most of the important scoring categories down (from aircraft range, control authority, ease of flying, comfort of ride, etc.). It's just that there was pushback for the idea of going through with it because the current KEA was still going at the time, so it would have been somewhat disruptive to suddenly have this value-based scoring system pop up. I think that the scoring sheet we were working on provides sufficient enough uniformity in actual feedback for performance in meeting the challenge's goals, maybe with the addition of easily accessible metrics such as size dimensions for some additional help in easily parsed value-based differentiation between aircraft.

Given how much the challenge has evolved even during the time it was running (inclusion of spreadsheets for organizing, testing methods, etc.) it does sound like a complete refreshed thread (if/when it happens) is good just so everyone can get on the same page of what to do, because the KEA processes only really hit its stride once we starting judging enough planes, thus finding and solving kinks in the process, and figuring out what we all really want out of the challenge. Things that would be best listed out in the main post, instead of having these kinds of things lost deep within the thread somewhere where it will never be found by anyone new that might want to judge (or even enter a plane). Things like, "if judging, use Pilot Assistant". Or "don't use Aircraft Autopilot, it messes up control surfaces between installs with and without it". These sorts of things. I personally think KEA's original mod list is sufficient for the challenge, and better retains the spirit of the whole thing. Splitting up between "lightly modded" and "heavily modded" where "heavily modded" just adds extraneous parts or near-functionally identical parts just seems like straying from the strengths of the KEA challenge. I like customization and variety, but I also think KEA is better with its tighter constraints.

I mean, I was under the impression that it would be possible to create a general-usage kOS script that would just run a test that would work for any aircraft, not need a script specifically for an aircraft; simply set a speed, altitude, then it once it achieves those, auto-trims (because I guess I have this notion that a good program will be able to achieve a proper trim better and faster than a human tapping the trim controls). But if Pilot Assistant already works sufficiently well enough for most aircraft, then that seems fine then. That said, I was also thinking that FAR is able to do the simulation graphs and that a stock analogue would work equally as well along with feeding it some other values. Kerbal Wind Tunnel looks quite promising in that regard. But there's still the extra caveat of the actual aerodynamic layout of the plane including thrust offset, trim settings, and such, which I'm not sure it replicates. Though one could argue this is the point where "margin for error" comes in, and any angle of incidence seems like KWT does account for. I think I'd still prefer an automated, but actually experimental, test process though. Working things off of graph readings just... seems a bit against the spirit of the whole thing? Maybe that's just me though. and since the topic of mods came up briefly, I really like KEA's highly condensed mod list, and dear lord, I loathe mod bloat, but maybe perhaps consider expanding the mod list to maybe include some other smaller but also useful mods out there for aircraft-building? Not sure which; maybe procedural wings and maybe parts, just to give people a little more choice (though I personally would probably continue just using the basics). Tweakscale (including AP+ patch) covers most other bases that's really required for building commercial planes. I do rather like the challenge of forcing people to work within tighter constraints in terms of part availability and using Tweakscale to its fullest potential instead, as I think it's great "training" for learning the fundamentals of KSP aero and even some things about real aero (a fact I maintain as the KEA challenge's true best value); a focus on making parts work, rather than having this huge parts bin of various stats all over the place that everyone can just pick and choose whatever and muddles learning how KSP's aero system works. Mostly just mentioning this as a note that we could come back to, instead of me forgetting about the topic.

That still leaves judging with the relatively arduous and boring task of verifying range to some extent. A standardized process that's straightforward and that everyone can agree on is still needed. I think the way fuel/range calculation might need to be looked at. Gah, what a complex issue. I do really like having more precise test methods, since this is still a KSP challenge where entrants should be encouraged to try and score high in certain metrics if they're submitting with the intention to do so, but finding a good verification process for that that everyone can agree to is really a puzzler. Maybe a look into Pilot Assistant would be good; I've never used it, but if @NightshineRecorralis says it's as good as it is, then maybe that's all that's needed to make judging a better experience. If not... maybe a kOS script that automates the entire trimming process and allows input for altitude and speed? idk, something that allows me to turn my brain off while verifying an entry's performance for what is ostensibly the "primary" challenge of the whole idea of this design challenge.

Would've maybe been nice to know of the option, but I think the entire process (starting from even the entrant's side) can still be streamlined further. Somehow.

I'm a little bothered my highly unconventional "seaplane" never got reviewed. Anyways, at its core, the KEA thread has always been attractive for its relative simplicity as a design competition. There's some basic requirements to be met, but the challenge, really, is otherwise pretty open-ended. I think that shouldn't change (though I, personally, would participate in an add-on challenge that would go further). If I were to think back to the most tedious part of judging, it's really the part getting cruise conditions down, since some of us are really reaching for that maximum performance metric. I think measuring aircraft range is a part that can be smoothed over, which will help make judging slightly more palatable to perform. I'm not really sure how that would be done, but it does just seem that manually finding experimentally discovered values is just the worst part of it all. You're just staring at numbers and trying to maintain decimals while you tap the trim on the plane, and god forbid you take too long that the fuel drain has kicked in and you're not sure if you're really getting the right numbers anymore. I also like searching for optimal performance, but geez, this was such a tedious part of judging. I'm pretty sure I gave up because of this. There needs to be some better standardized and potentially automated replacement for that process. Maybe some other mods to fill that gap, at least on the judge's side, but also probably changing up how cruise conditions are defined and calculated. (A mod/calculator that can straight-up simulate values of flight conditions and aircraft attributes, i.e., lift/drag, fuel burn, etc., would be fantastic.) Every other part of the judging process (the optional ones as well, though an argument can be made for making them mandatory) is more free-form and actually focused on flying/observing the plane yourself, and I generally enjoyed those portions.

The split off part, really. That seems... in some ways, advantageous.

I have to say, I'm a bit confused at the sequence of events and how the IA-42s always start off more split apart. That said, I think my primary flaw was assuming that altitude is a good thing when dealing with low power aircraft, but seemed to forget my knowledge of BDA dictates that the AI really defines how battles play out, and that trying to gain altitude is a really, really dumb idea with BDA.

Yeah, thanks FAR, because moving my rudder down and making my vertical stabilizer shorter should worsen my L/D and Cl. FAR is great, but it still ain't perfect...

Yeah, I suspect the single engine fighters, given that they have much less power, are less prone to GLOC-ing themselves in most cases outside of a high(er) speed dive.

So far, the true winner of this competition seems to be GLOC. I suddenly don't have high hopes for my own plane.

I'm quite rather disappointed you didn't, I wanted to see the XF5U fight.

I actually thought it was just going to be a Spitfire when we saw it in the group picture, since you couldn't see it from the side. 2. It actually originally had 160 points, but then he revised it before final submissions, I guess, and missed something when he was adding up points or something.

You mentioned earlier about your plane's lack of a rudder. So, a plane can fly without a rudder, but it won't have any fine control in the yaw axis, which is kind of important when you need the plane to point exactly in a certain direction (to, you know, shoot and hit another moving target). That's probably the big thing here.

Curiously enough, I did at least now realize that BDA uses the GLOC code as the same one for its EMP effect. The vertical space usage, I suppose, is partially because BDA was originally envisioned around modern tech levels, or such, where power (relatively speaking) isn't an issue. Though, even then, energy conservation still does help. As well, it comes from a particularly stupid part of the routine that I've always hated, the Extending routine (BDA's Extend routine, not SI's additional one... which has its own issues), which forces the AI to attempt to achieve its default altitude setting (hence, why I often claim that the low ground is the advantage in BDA, unless you're playing with ludicrous amounts of TWR and an obscene amount of guided missiles). Its evasion routines are severely lacking at this point in time, when many players are generally accustomed to how BDA works, and it's showing huge flaws when it can't play to an aircraft's strengths, or even just not doing stupid moves like making a half-turn for evading, only to stop, then turn away, resulting in only letting the enemy get closer and nothing else. Dundun92 was working on some code that had a dive-angle-and-velocity-adjusted min alt pull up routine. Airbrake usage, actually, I've found, has been pretty decent, at least in normal BDA 1.3, certainly seemed like an improvement over BDA 1.2.4; not sure about SI's version, as I haven't experimented much with it. For top speed, or speed parameters in general, I think there needs to be a few more settings (as in, a top speed for "powered" flight, and a "do not exceed" speed for dives). Dundun92 was also working on something like this. And yes, target priority needs a rework. Also, at least here in BAD-T with slower aircraft speeds and relatively shallower maneuvering, the AI really, really needs to be able to figure out how to break away from a gun pass or something, because reversals are some of the stupidly deadliest things right now, and they're not even satisfying to watch the way BDA does it.

Methinks GLOC should have the pilot release the stick or something, but I guess getting locked inputs is semi-reasonable too?... (That constant GLOC in the turn circle from the H-4 though... it had no chance.) Also, uh, this may be getting ahead of ourselves, but should we have max gun range limits? IDK how "WWII" this intends to be, but I'm pretty sure planes were not shooting at each other from 2.5km away. But the main point here is that in BDA, that causes the AI to do really stupid things, which just seems... out of the spirit of the whole thing. Not to mention that the AI really, really needs better evasion routines or something. The AI can do something really stupid things a lot of the times, especially in regards to giving up any advantage it could have gained, or at least giving up the opportunity to escape its current disadvantageous position.

The 23mm are much better investments than the SparkVAKs ShVAKs at least, for only 3 points more per installation. For lighter fighters, a better investment. Although going along the lines of WT aircraft gun memes here, the MG151 sometimes feels like it hits with the level of WT's MG151 minengeschoß rounds, so those are also pretty decent.

It's hard not to look German when you plant the unmistakable Jumo 213 at the front of a conventional fighter. Like, what else is going to look like other than a Fw 190D / Ta 152? At least the Double Wasp (B40 in the pack) can pass off as something else a lot more easily. Originally my design was a lot more pseudo-Dora, but I decided I could do something nice with the wings, since the wings were initially a lazy attempt at making a sort of elliptical wing that ended up sorta... uh... probably like a non-gull wing Corsair wing. Really, my original design was a weird hybrid of a D-9 and a P-47N, made from what was originally supposed to be a generic plane? After I figured out what I could turn the plane into based on what I already had without messing around too much with the aero, I settled on adjusting the wing shape and some other things to make a pseudo-Sea Fury, because I'm sure at one point this experience will be just as depressing as flying it in War Thunder against the Germans and their fantastic BS machines. Also it flies marginally worse than said original design, but screw maximum performance, I wanted the Sea Fury. And it still took me an entire day to fix its aero back to the original levels (well, as close to) despite trying to make as minimal changes possible. Thanks, FAR. Because somehow adjusting the height of my vertical stabilizer and rudder affects the L/D and Cl. My plan for the plane was to out-pilot competitor planes and get lucky otherwise. Maybe it'll work, but I'm starting to wonder if lowering the G-limit on the Pilot AI was a mistake, because it's now more likely to linger at high-Gs than just pull hard and bleed speed and then Pilot is unable to sustain such accelerations. Those are tip fuel tanks, by the way.

That one stuck out to me too. It's like an extra-funky BV 141. https://en.wikipedia.org/wiki/Blohm_%26_Voss_BV_141 Also, I really thought the 23mm would be more popular.

The AI never seems to be *not* aggressive in roll, but too high steer ki seems to cause it to under-roll, then increase the roll input too much and too quickly, causing it to overshoot its intended roll target, and then it just sort of keeps waggling around with the roll like it's confused. My own entry is running 6.0/0.27/2.5, which I found to be a good compromise between controllability, aggression, and stability. Still not happy with it though, if I had more time I'd try more things.

@dundun92 Pilot AI tuning theory has changed a bit actually, due to minor changes in BDA 1.3 behavior. Higher damping values are now more acceptable (though I still don't recommend putting it too high, since it might still limit pitch aggressiveness or stability). Damping is just... a lot more nuanced now. I find that damping also has more impact on gun accuracy now, since it's back to being the "stability" sort of setting. While 2 and below is still good for maneuverability (and indeed, almost necessary in stock KSP combat for extremely maneuverable aircraft), 2-3 damping is a good middle ground for permitting the AI to be maneuverable, while also tightening up the plane's flying more. I've done some tests with some higher values, like 4-5, and planes can also fly just fine now, but I feel like it kills the plane's aggression in turns. You either have to compensate with higher steer factor, which can affect other things, or leave the factor lower and accept the lowered pitch aggression. This *could* be advantageous though, since it can *kind of* act like a "dynamic" speed limiter (which, sadly, I didn't have time to test for my BAD-T entry). Steer Ki is also far more noticeable in roll control. It seems to manifest itself like damping, except for roll. Too high, and the craft just seems to roll "too stiff", which causes problems of its own (like "struggling" to "find" the correct roll input). This one is also tricky because you also want it set higher to enhance gun accuracy. I haven't fully figured out what to do with this, but for now, I've focused it more on roll control than the gun accuracy aspect. Steer Ki seems to want to rest between 0.2 and 0.3 for most conventional setups. Seems to be working well still. 0.5 steer ki seems to cause issues with proper roll control, at least for a number of craft I've tuned recently. (Though, interestingly enough, the PEGASys-K still seems to be fine with 0.5 steer ki, so idk. Things are confusing again.)

Were craft downloaded yet? I had last-minute adjustments I made and reuploaded to the KerbalX page this morning.

I wanted more proper replica-ish in my plane, so I took a look at my plane and tried to match up various characteristics to real examples. Thought the wing on my initial design was good, but sloppy and could use a bit more work in the aesthetics department. The elevator/tail configuration was of the rudder-between elevators type, unlike a 190-Dora, and I didn't want to change up the aero too much. So I turned the plane into a Sea Fury. The "little bubble canopy on top of big body" look seems to fit it anyways. Now to come up with a good name... Also, is Airplane Plus okay? Kinda want a smaller tailwheel.

I just noticed the Advanced AI settings were hidden. welp, time to go adjust those too Craft aren't downloaded until tomorrow, right?