K^2

Very mixed feelings. On one hand, yeah, I get that the fact that it flew at all is a success. But I'm also seeing a lot of people trying to over-sell it. Yes, we expected something to go wrong during the flight, but a lot of things went wrong during this flight. Chunks of the platform coming off, fragging 3 of your engines, a bunch of safety equipment, a camera, and a car in the parking lot is not how you expect the engine start to go. And it somehow went worse from there. While I'm sure the damage to the underside of the rocket from liftoff could have led to some of the following problems, it sounds pretty incredible for all of the issues to be related to that. More engines flamed out, some with clear engine-rich exhaust flames. Ok, yeah, I suppose damaged plumbing could have resulted in that. But then we had a separation failure, engine shutoff failure, and was the flip supposed to be part of the booster slow-down procedure post separation? In which case, should that program have ran? Yeah, I get that the ship was lost by that point, but this is supposed to be human-flight rated eventually, and booster doing basically anything but that would be preferable for any sort of an abort procedure. Even if you're going to remote-detonate an uncrewed rocket, keeping the rocket pointing forward would give it better chances to coast clear of population, so this definitely isn't a good failure mode to have. I also see people pointing out that the rocket didn't disintegrate from the flip as a positive. That's another problem. Not with the structure, obviously, but it makes you pay attention to the numbers, and it was over 30km up going less than Mach 2. Two minutes into the flight, Falcon 9 is going Mach 5. The Superheavy was climbing way too slow and never made it close to its Max Q. I think we had an official flameout of what, six engines? It sounds like way more of them were underperforming. Again, could be damage on liftoff, but holly crap, does it show how the failures cascade on this thing. So yeah, by the time it started flipping, it was over 30km up, where the atmosphere is very thin and the rocket wasn't going that fast. Aerodynamic stress would be significantly below what this rocket is designed to survive during the booster's normal return flip, and certainly nothing like what the Starship is designed to survive on reentry. This test actually failed to see if the rocket is sturdy enough to do its job. And yeah, to some degree, all of these are nitpicks. But there are a lot of them, and if SpaceX opened up with, "Yeah, that went poorly, we're going to have to go back and look at what went wrong across the board," I'd be happy to focus on the things that went well. But their response has been, "This was great, better than we expected, and we have another one practically ready to go!" And now I'm concerned about the transparency and what exactly being swept under the rug. Obviously outsider's perspective, pound of salt and all that, and I'd love to hear what people working in the industry have to say about this. Just sharing my first impressions.

A lot of the systems are there too. Obviously, I can't tell what sort of a shape they are all in, and at least some are definitely just stubbed in, but also, a lot of the parts the team can add can run on combinations of existing systems. It's certainly a constraint on what can be added, but it's not a hard blocker to add some new content to keep people playing and providing feedback. I don't think they want to drop everything related to surveying planets, deploying extraction bases, storage, refining, and transport of said resources all in one go. Yeah, having a surface scanner like what we had in KSP before there is a way to extract resources might be almost cosmetic, but it's the sort of thing they can do without dragging in all of the dependencies. They can also add the resources themselves, storage tanks, and some portable versions of refineries for some of them. Again, similar to what we had in KSP to store ore and convert it to fuel. I do not expect a full working system soon, but I do expect components of it to show up relatively early, gradually building up to it. It's a better way to use the team's resources, a better way to test for problems early, and a better way to keep at least a fraction of the community engaged. There are no real downsides, and if something along these lines doesn't happen in the next patch or two, I would actually start worrying a bit about the project. (Which, I'm sure a lot of people would claim is overdue, but so far the patches have arrived on the kind of schedule and with kind of fixes that I have expected, and the receipts are right here on this forum.)

That's not how it works. Different teams work on different kinds of problems. If all that Intercept does is fix problems, a lot of the team are left twiddling their thumbs. From the perspective of getting the most use out of people effectively alpha-testing the game, yes, having roughly the same content from version to version is good. But also, people will start losing interest in the game, and even these who play will fall into the routine, which is exactly the reason that QA isn't catching some of these bugs (besides, likely, being understaffed in a typical industry fashion). So it's very likely that we'll be getting a trickle of new content with future patches. I don't think interstellar is coming until a lot of the things are nailed down, but some new parts for science and resource gathering are extremely likely to land in the next month or two. Even if not all of the associated functionality is available right away.

"Everything below sea level is water" is a good optimization. KSP used this and I see no reason for KSP2 not to as well. So if you somehow managed to make it through the surface, it makes perfect sense that you'd end up in a fluid, probably eventually sinking to the center, where your ship would get krakened by the singularity. I do wonder, however, what caused this clip to happen in the first place. Were you time-warping? Switching between the vessels? Or did it just happen completely out of nowhere?

I suspect that there is bias in the results so far. Since MP is delayed, most of the people for whom this was the most important aspect of KSP2 are probably ignoring the early access, and so aren't going to be as active on the forums. So the people for whom the multiplayer is less important are going to be over-represented in this poll. That said, even with this in mind, this is a lot more people who don't care about MP than I expected. To the point where maybe MP should always have been a future expansion goal.

I think you mean it as sarcasm, but since yeah, there are people who make these exact arguments, that's why I started with procedural tubes. They already let you build a ship as if you were building with procedural fuel tanks - just without any fuel in them. Not to mention all the other procedural parts. The wings have been great and the preview of procedural radiators is promising of more procedural parts being added. So the ship has sailed on all of these arguments KSP2 isn't that kind of a game. It clearly expects you to adjust some parts to fit your rocket. And at that point, fuel tanks not being one of these part types is just strange.

I completely fail to understand why they aren't. We have structural tubes that can be adjusted to the desired length. So in structural terms, we already have that. All it needs to do is also be able to hold fuel in proportion to available volume. There seems to be no reason not to have that from perspective of game design, tech, or assets. Everything is in place already, and all we get with the current system is less flexibility with more part clutter. I hope this is just a temporary situation and they're going to replace the current tanks with procedural ones before the game goes final. P.S. That goes for solids as well, yeah. It can make sense to go in fixed increments for either or both of these, rather than a continuous slider, but the lengths of fuel tanks and solid boosters should be adjustable.

For many "grid" games and simulations, yes. A lot of simulations in fluid dynamics and lattice QCD employ clever space-filling curves to keep things that are spatially close together closer together in memory, because even when running on GPU with a cache that's much better optimized for these sort of workloads, the massive datasets that you have to deal with will absolutely destroy you if you don't pay attention to coherency. (I've had cases in simulation where it was worth it to copy a grid transposed for a certain pass, then transpose-copy the data back to get the performance improvement.) So it's worth the not insignificant computational overhead to go for a very complex layout. For games like Factorio, space-filling curves might not even be an option, so no matter how much you optimize, you're at a minimum straddling the stride of your grid which will hurt you on L3. And this can get particularly gnarly if you have to execute updates on the grid in a specific order to adhere to all of the game's rules, and that's often the case in these sorts of games. Even going to something like the aforementioned Cities Skyline, the grid is substantial. Definitely easier to optimize than Factorio's, but I wouldn't expect it to work well with the cache with a naive implementation. In KSP, though? If your engine's component and asset models are coherent, there shouldn't be a way for the developers to screw it up. Again, not without trying to. I'm not saying it like Squad didn't make a mess of it - they most certainly have, but even with the game object model of Unity, the engine should be protecting you from the worst of it. Like, that's the engine's frigin' job. And yes, yes, I know, if individual behavior scripts start allocating left and right, it's very easy to end up with things tripping over each other, but that's why you supposedly outsource the engine development to the professionals. If you're writing the engine, you control the allocation and how the scripts are executed. Pool the memory. Group the execution together. It's textbook stuff. And yes, I'm upset. It seems Unity keeps finding ways to disappoint me. And yes, I know there's a lot KSP and KSP2 teams could have done to make it better, but forgive me for holding dev teams of fewer than 50 people in total to a different standard than a multi-billion corporation whose entire job it is to make a game engine.

That flies in the face of everything I've seen profiling other engines. I don't mean to say that I doubt the results - based on other links this seems to be well established, but no half-way optimized engine should even be allowing that kind of garbage cache performance from its component architecture. That said, Unity keeps doing things I don't expect from a civilized engine, so I probably should hook up profiling tools to both KSP and KSP2 and take a look at what's going on there. With that in mind, yes, it would absolutely be worth testing to see if KSP2 is going to be the same way, but also, if it's still the case with latest Unity, somebody at that company should be reprioritizing what they're investing their engineering effort into.

I honestly haven't even considered the neutrinos being ionizing and whether that can contribute to a radiation hazard. Thanks for running the numbers! So I guess, we have to add "non ionizing" to the list of requirements to this magical engine. The reason I didn't put much thought into the neutrino energy is that decay of any sort will produce so much waste heat that there is no plausible way to manage it. You have to be incredibly near 100% efficiency on whatever means of neutrino production you go with to make this practical. Given that, I don't know if there is any reason to think that 0.5MeV neutrinos are going to be any easier to produce than 0.1eV neutrinos. And if you can dial down the neutrino energy, you can get to the point where it's non-ionizing. Since they are not entirely massless (theoretically, but well-established), you do expect a roll-off, but our upper bound on mass is below 0.13eV, so even with room-temperature neutrinos (a little under 0.03eV), you'll still have at least about a third of maximum thrust. Likely a lot more. For non-ionizing, less than about 1W/kg will be absorbed easily enough short term. Human body is capable of producing more than that internally. But this does bring up an interesting point. I didn't bother computing the energy density at which a low energy neutrino beam will still absolutely cook you, and based on your numbers, it's not as absurd as I might have thought. Which brings us to potentially the best possible use for an overpowered neutrino beam, achieving the goal that has been eluding humanity for millennia. A piece of technology that should be in every house, and now, maybe, we can hope that one day it will. We can finally have an oven that heats your food evenly throughout its entire volume.

Yeah, and there's certainly no one-size-fits all - you really have to profile and see what you're hitting. But as a general rule, a single thread might be L1 or L2 limited. If a single thread is L3 limited, either something's horribly wrong with your CPU architecture or you are doing something especially awful with your algorithm. A modern L3 is designed to feed 8+ threads. If you can saturate that with a single thread without intentionally trying to, any multithreaded task is going to struggle. In case of X3D, you're only really getting benefits if you blew out your L1 and L2, and the X3D gives you an L3 hit instead of a miss. And X3D isn't going to help you accessing any memory you've never touched before. It only helps if you've had other memory access walk all over the relevant cache lines. That happens all the time if you have 20 threads doing their own thing, but if you have mostly the one thread doing the work, it tends to be pretty unlikely. Based on everything we've seen of KSP2's performance, I'd be shocked if KSP2 has a lot of L3 misses that X3D would make any difference to. I'd be happy to be proven wrong with a benchmark, but everything we know already heavily favors single-thread optimizations in the architecture, which means faster cores and more L1 and L2, and that's not at all the sort of things that X3D helps you with.

I doubt X3D will help with KSP2. It's a cache improvement, and I don't think that's the bottleneck for the game. KSP2 is still main thread bound, and you generally only start getting into L3 cache bottlenecks once you're making good use of all the threads.

And with magnets, magnetism itself is a warm-up exercise. Because that naturally flows into, "Oh, so when a current flows, there is a magnetic field.? - "Yes." - "So in this magnetized chunk of iron, there are current flows?" - "Well, no, the magnetic field in iron is produced by point objects." And then you have to talk about the spin. And intrinsic degrees of freedom that lead to it. And why that particular hammerspace happens to induce magnetism in the real space, which in itself a wonderful topic. And then how all of this conspires to give an entire atom a magnetic moment, because of course Pauli exclusion is involved. And when that explanation is finally wrapped up, that is usually followed up with, "Well, don't a lot of atomic nuclei have a magnetic moment? Why isn't nearly everything a magnet?" And this is where you break down and cry, because now statistical mechanics is added to the mix. We have quantum mechanics, relativity, and statistical mechanics contributing. The trifecta of all that is modern physics. Welcome to the nightmare, enjoy your stay, don't forget to visit the gift shop on the way out. We have fridge magnets that just say, "MAGIC!"

I learned the hard way that anything magnet is always magic. You might think you can explain magnets. Then you try. And then you fail, as the explanation devolves into a number of tangential topics. And you try again, trying to stay on target this time, and you fail some more. And then you start wondering if you understand magnets. At which point, it might be easy to fall in a despair. But then you watch Richard Feynman be asked about magnets, and instead of explaining it, he goes into several minutes of diversion on whether you can know why something is at all on the fundamental level, finally giving hand-waving non-explanation to magnets. And then you just settle on, "Magnets are magic." Of course, there is science to them. And of course, any given question about them can be rigorously and mathematically answered, but then there is nuance upon nuance, relativistic and quantum effects, and even some quantum relativistic effects, and you're scraping the very fundamentals of what is cosmos, and you still have gaps in what you're covering about the magnets. All the meanwhile, Gandalf, the smug so-and-so that he is, stands there smirking at you, smoking his pipe, because he figured out long ago that you just call it magic and move on. Magnets are magic. Hall effect is most certainly magic. Hall effect in superconductors? Look, I'll just put it this way, the best quantum computers we got are just a chainmail of tiny superconducting magnets connected to each other via Hall effect. They can't do general computation very well, but they're fantastic for quantum annealing, so people are running neural nets on these. People jest about the thinking sand inside computers. These frigid ceramic donuts are going to out-Hall-effect our brains at this rate. Not only is it magic, it's the highest form of magic. I will be taking no more questions on the matter.

They'll orient neatly in a strong magnetic field. And this would absolutely work. Get a bunch of baryonic matter that beta-decays - neutrons being probably the most energy-efficient, but something like tritium would be easier to store, shove it in a strong magnetic field, enclose it to prevent any leaks, and it will produce thrust by emitting more neutrinos in one direction than the others. It's also going to have an absolutely abysmal efficiency, since only a tiny amount of decay energy is going into neutrino production, and they aren't emitted in a single line - merely preferentially kicked out more in one direction than the others. Neutron decay releases a few hundred keV of energy, of which a few eV go to neutrinos. So we're looking at something like 10-3 efficiency here, and then the recoil momentum bias is less than 10% (I don't recall the exact number, and this will pain me to look up, as this is starting to get a bit obscure, but lets just take 10% as an upper bound). So for every 1N of thrust, you'd be releasing something like 300GW+ of power as electron recoil, which in this case is just guaranteed to be heat. Which on one hand is really cool. Yeah, 300GW is energy production of a medium sized country, but I can imagine some sort of a huge colony ship producing that much power for various uses, and 1N of thrust will probably be measurable even at that scale. So with the right type of a reactors you can magnetize it and get a little bit of thrust out of it. On the other, I wouldn't call it practical, because you're dumping waste heat to space anyways, and if you paint your radiator panels black on only one side, you'll be literally getting a lot more thrust from asymmetric IR radiation than the neutrino flux. So that's why I brought up the efficiency in the initial post. To make this kind of propulsion practical over photon drive, it has to be really, really efficient, and I just don't see a practical way of achieving that.

Well, I can see one use case for a neutrino thruster. Imagine you had an efficient mass-to-energy conversion that you wanted to use for a drive. Photon drive is a natural choice, but suppose you want to go to a hover over a colony's landing platform. Well, a 100kN thruster you'd want for a modest shuttle is a 30TW beam. You can cover that landing pad with your best reflective material, and it will still melt like a stick of butter in front of a blow torch. That's not ideal. And forget about using it in the atmosphere. Air might not burn exactly, but with tens of terawatts of input, even with the tiny absorption you're likely to get, it's still going to be a plasma tornado. You just can't use a drive like this anywhere outside of deep space. If you could generate a neutrino beam instead, even at these ludicrous energies, it's going to be safe to walk under. For all intents and purposes, that shuttle's going to hover in the air like it's on grav plates, with no noise, no dust, no moving parts at all. Just 10 tons of pure thrust pushing against the engines that can be fully concealed within the body of the craft. The cross-section is just that small for weak interactions. Other than the fact that you still need an energy source to fuel all of that somehow, such a drive might as well be reactionless. Of course, all of that said, that's also precisely why it's so hard to imagine something like this working. The weakness of interaction is what makes production of neutrinos so difficult. And for something like this, you need to be ridiculously close to 100% efficiency, or you're just going to die on thermals. You might be able to dissipate, what, a few MW of heat from a landing shuttle without causing problems? So you need to be better than 99.9999% efficient. Something like Mossbauer Effect might give you a recoil into the lattice at the right sort of efficiency. But now we're trying to invent a Mossbauer neutrino laser (naser?) that also has power in terawatts instead of milliwatts, which is what you usually get with these kinds of techniques even with optical devices? Yeah, I don't know what sort of quantum wizardry might even approach these sorts of magicks. A graviton drive might be more plausible. Same benefits, different particle. And at least, we have an idea how to couple something very strongly to a gravity field, because gravitational field isn't strictly linear. All you need is a lattice of tiny black holes and you're in business. And the fact that this seems plausible in comparison should speak volumes about how miraculous the neutrino drive would have to be. But, you know, would be neat if we could have it.

That might actually be easier, because here you're fine with it being pulsed. If you take a gyro, spin it very, very fast, and then start rotating its cage in a perpendicular direction, you will generate a small amount of artificial gravity along the surface of the gyro. This artificial gravity will rotate along with the cage, meaning it will only point where you want it for a brief moment, and will point in the exact opposite direction at some other point in time (just like before, average gravity being zero makes it a lot easier...), but if you time your propellant in pulses, you can get thrust out of it. Will it be a lot? With anything a physical gyro can survive, no. Actually, I don't think it'd even be measurable with anything we can practically build right now. But we might be able to find some sort of an effect that lets us bottle an enormous quantity of angular momentum in something without it physically spinning. Not only would it make for amazing reaction wheels, but we'd be able to do something like this with it. I still don't think it'd be useful for actual rocket propulsion, as with that level of tech, we ought to have better options, but just the concept of being able to briefly pulse gravity for various needs might be useful in some other applications, and that might actually be achievable without us discovering fundamentally new physics. Just some exotic materials that may or may not be possible.

Well, therein lies the complexity of it all. I've mentioned magnetic fields above, primarily because people are familiar with magnets a bit more than electrostatic effects, but it is, indeed, the electrostatic forces that are an almost perfect analog for gravity. Even with gravity as immense as the Sun's, the difference from electrostatic forces is effects like precession of the orbit of Mercury, which is tiny all things considered. You really have to get to neutron star gravities before it really becomes important. So with that in mind, you can pretty much apply basic electrodynamics to gravity. And just like with electric fields, there are rules you have to follow. Relevant one here is the one about the field divergence. I won't go into the math, but the way you picture this is by drawing the field lines. Any time a field line stops or starts, you have non-zero divergence. So you can't have the field lines just start at the gravity plate. That would require the plate to be very massive. How massive? Well, the Earth needs all the mass that's located directly under your feet to generate 1g of gravity, so you'd need that much mass. I hope you agree that's not remotely practical. Because of how much mass we need for even the tiny amount of field divergence, to build a practical "gravity plate" it would have to be divergence-free. This is where we go back to magnets analogy, because magnetic field also has zero divergence. You achieve this by looping all the field lines back on themselves. If you look at any picture of the magnetic fields around a magnet, you'll see the field lines pass through the magnet, then expand out at either pole, and then wrap around outside of the magnet creating a sort of a field vortex around that magnet. We'd have to do something similar. We'd want the plate to attract things above it, repel them from below, but then around the edges, the situation would be reversed. It'd be a much weaker field, rapidly dropping as you move further away (as an inverse cube) acting in the opposite direction, applying a bit of gravity in the upwards direction. The net gravity across a slice passing through the plate would be zero, with positive 1g near the plate, and negative, but much weaker gravity, far away from the plate, dropping to zero as you get far enough out. There are a whole lot of additional complications here. Like, getting the gravity field to turn like this requires it to have a non-zero vorticity. Or curl, mathematically speaking. There's a simple way to generate that without bending space-time, but that requires accelerating a lot of mass at very high rate, so again, not practical in the least. Whether this can be a stable system at all is unknown. There is a famous result from mean field theory that the gravity allows for a torsion term. Einstein set it to zero in his field equations, either because he missed them or because he saw no evidence of them. To this day, we have no idea if they are physical. There is no known space-time torsion "charge". If we had one and could bend it into a loop shape without causing problems, you would, theoretically, have exactly what you want, a gravity plate with properties described above. It might even require no upkeep. It'd just exist, like a permanent magnet. You could install it on a ship and fly around with it. You'd just have to be careful about two ships with such plates passing close, because they'd behave exactly like you'd expect a pair of giant magnets to. But this goes of into such deep hypotheticals that it's pointless to speculate further. We have zero evidence that this is a physical phenomenon and people have looked. So we might be out of luck with this, and there is no known alternative.

Yeah, that's kind of the problem. We can put some bounds on how much or how little it can be, but without having a specific method in mind, you can't really say what it will be. Linear acceleration will add to the kinetic energy. Do we want that? Is it a waste? Can it be avoided to save on power requirements? All depends. We clearly can with rotation. Can you "rotate" in place without moving? There might be degrees of freedom we can tap into, but that gets uncomfortably close to the attempt of reinventing string theory. *big shrug*

It's just like energizing a magnet. A magnetic field has energy. You have to put that energy in somehow. Any objects within the field may be magnetized by the field, and that will add to the energy requirement, as they now gained potential energy relative to the magnet. That's the minimum you have to add. From there, it all depends on how efficient your setup is. A solenoid will drain power. A permanent magnet won't. A superconductor may require cooling, which is a power drain that had nothing to do with how much field it's generating. Artificial gravity is much the same way. @darthgently is spot on with centrifugal gravity. That's real gravity and it plays by the same rules as any other methid would. We just know how to make it actually work. The more mass you have, the more energy it will take to get the wheel going. You're putting that potential energy in. If ships depart or arrive you might have to adjust that. Other than that, if the wheel is frictionless, it just keeps going. If there is a source of friction, you'll have to supply some power to maintain the spin. How much energy you'll actually spend? Depends on how efficient your setup is. If you can literally attach an electric motor, it will be very efficient. If it's rockets, not so much. Other means of generating gravity will work by the same rules, but we have no idea how much overhead there will be for something practical. We know that it can be a lot. If you have to provide energy equivalent to mass energy sufficient to generate that gravity, it will be more energy than the civilization produced thus far. By a lot. We know this brute force method works, but it's not practical precisely due to energy requirement. Anything else will be between these extremes. P.S. Completely forgot to add that acceleration is gravity. There is no distinction. That's part of why we can use the centrifuge as a benchmark. But linear acceleration also works. 1g gravity aboard a torchship is a real gravity. How much energy is that taking? A frigin' lot, but ot's energy you need to spend either way. So that's one more example.

Are you... Are you Kraken-driving this thing to hover, so that the engine can look like it's a burner?

Lighter-than-air buoyancy is even easier, actually. With water, you have to worry about a surface, and that's the tricky part - you have to estimate how much of a part is actually submerged. It's very simple for some colliders, like spheres, and gets progressively harder for others. Even for a cube, if it's tilted, computing the precise volume displaced at any given position is surprisingly complicated. Of course, there are very simple estimates, like you can just take the top and bottom extents and see where the water surface falls relative to the two. If it's between, take a percentage of the collider's volume. It's not quite right, but it's good enough for a lot of games, and I wouldn't be surprised if KSP1 did something like that. For lighter-than-air, there is no real surface to worry about. Yeah, there is a cutoff for atmosphere in KSP1/2, but by that point, the atmosphere is so thin as to be irrelevant. So you can always use exactly the same formula for displacement lift while in atmosphere: gravity * volume * atmospheric density. All of these are available to the components in the game, so it'd be very easy to write a module that handles this. This would be very simple for Intercept to add if they decide it's something they want in the game, and it will be very easy for modders to add if Intercept doesn't. The only barrier right now is lack of the SDK, and even that might not be a blocker already.

I know people figured out how to add modded parts already. It'd be trivial enough to make a part that has a negative weight, but, of course, that would float straight to the edge of SoI, which is not exactly great. You could also make an "engine" part with an infinite ISP and a custom altitude profile that allows it to only go up to certain pressure altitudes, but then you could angle that and get an infinite thrust, which is also not ideal. Proper atmospheric buoyancy would require a custom module, I think, and I don't know if anyone got that far with modding the game. It's certainly possible, but I don't know if anyone has done the work yet. I'm also not sure if that level of modding is in agreement with forum rules, given that SDK isn't out yet, so that might have to stay on the outskirts of Discord until official support is added. But I'm definitely looking forward to some sort of lighter-than-air builds in KSP2. It seems like it would very much fit the style of the game.

A shout-out to MinerBat on YouTube (I don't know if they have an account here) for building a full return mission to Jool. It uses a Kraken drive for the final bit of flight with rotors to gain altitude enough for the Kraken drive to work, but even with that drastic reduction in weight of the final stage, this is a very ambitious build, and it's a good watch.

That clearly works from the energy perspective, but I instantly have a lot of "How?" questions. Like, how do you inject oxygen into that (also run it through NTR to get the gas up to speed?), how do you design an afterburner that works at these temperatures and flow speeds, and how do you build a nozzle that survives this kind of an abuse? Do you happen to have a specific design in mind, perhaps, that you can point to? None of these problems seem unsolvable, just really hard, especially in concert, and if somebody already put in the work to try and figure it out, I'd be interested to see what they came up with.