Jouni

Actually, you can't even complete the tech tree in 1.0 without deliberately grinding science. If you just fly around the system, taking reasonable measurements, collecting a few surface samples, and sending reasonable reports, you run out of planets before you have completed the tech tree. In my only 1.0.x career mode game so far, I've visited Duna, Ike, Eve, Gilly, Dres, Jool, Tylo, Pol, Bop, and Moho, but I've only unlocked a couple of 550-point nodes. In 0.90, I had already completed the tech tree at this point, because contracts gave you much more science. Unfortunately 1.0 removed almost all contract science, making grinding the only way to complete the tech tree.

It may also depend on the country where you live in. Different countries have different laws on what rights consumers have by default, and what kind of conditions in a license agreement are invalid.

I returned to my 1.0.x career mode game after a long time. The next mission is supposed to be Laythe exploration, but I don't have the parts to build a reasonable plane to land there. I therefore decided to build a replica of my 0.23 Laythe lander. The basic design still works, even though I had to add a third jet engine. The lander completed my standard Laythe simulation (launch on a rocket, deorbit, and return back to LKO) without any problems, so I'll probably use something like this for the mission.

I think somebody accidentally swapped the dry mass and the fuel mass of the LES during development, and nobody has bothered to correct the mistake. With 0.9 tonnes of fuel instead of 0.225 tonnes, the burn time would be around 2 seconds, and the LES would actually do something useful.

KSP would be slower on a supercomputer than on a decent gaming PC. The only thing that really differentiates a supercomputer from a pile of servers is the network between the servers. Supercomputers use Xeons and other standard server CPUs, and like in other servers, it's more efficient to add more cores than to go beyond 3 GHz.

Just imagine a 2.5 m stack with four of these, or a 3.75 m stack with seven of these. No more puny little Mainsails, Twin-Boars, and Mammoths for heavy rockets.

The normal burn time for my liquid boosters is 120-150 seconds, so 90-120 seconds feels suitable for solid boosters.

There are three rocket-related things I'd like to see in the game: A simple way for building interstages and engine clusters. For example, I might want to build a lander around a 2.5 m fuel tank, attach six Spark engines to the bottom, and then place the lander on top of a 2.5 m rocket stack. Large SRBs with meganewtons of thrust and 90-120 second burn times. A low-profile engine that's bigger than the Terrier but smaller than the Poodle. The Aerospike would work otherwise, but it's too late in the tech tree and it's missing the bottom node. I'm not sure whether a larger nuclear engine would be a good thing.

The concept seems workable, but the details are what really matters. They determine what the engine ultimately does: is its performance comparable to existing jet engines, or is its TWR and/or Isp 5x lower. From a gameplay perspective, jet engines work on Laythe, because the entire point of Laythe's existence is to have another Kerbin-like planet elsewhere in the system. Jet engines don't work on other planets with atmospheres, becuse the entire point of those planets is to present difference challenges to the player. If you make jet engines work on those planets, you remove more existing gameplay than add new gameplay. The 'sandbox' part of KSP, where you build Lego rockets and fly them in a physics simulation, is ultimately good. The 'game' part, with science, contracts, funds, ISRU, and recovery is, quite bad. The game mechanics feel arbitrary, they're poorly balanced, they don't work well together, and they're full of loopholes. You can enjoy the career mode, but only if you play it like a pen-and-paper RPG from the 80s. You know that the rules are bad, but they can still provide useful structure to the game, if you deliberately stay within the boundaries where you know that the outcomes are mostly reasonable. The more things you move from the sandbox to the game, the worse game KSP becomes.

Now this is confusing. I thought I was the one arguing from realism, while you were arguing from arbitrary gameplay considerations. In theory, a jet engine burning magnesium with CO2 might be possible. There are still a million obstacles to overcome before we have a working engine, and it's entirely possible that the engine won't be practical. Even if we get a practical engine, its characteristics would be completely different from a jet engine burning kerosene with oxygen. Hence it makes no sense at all to use the same jet engines to simulate both in KSP. A nuclear ramjet is a nuclear jet engine, so it would probably have a lower TWR than nuclear rocket engines. It's a ramjet engine, so its stationary TWR would be 0, and it would be useless at low subsonic speeds. It's an air-cooled nuclear reactor, so the airflow must be high enough or the reactor overheats. It can't go too fast either, or the airflow will heat the reactor instead of cooling it. This is an extremely bad idea. It's the equivalent of giving kerbals rings of ISP +10 (+20 when hovering). It utterly breaks the suspension of disbelief, making KSP just another bad game with a lot of bugs. It would make even less sense in the sandbox, which is the primary game mode for many of us. If that's your experience, your missions are probably too similar to each other. The shape of the nuclear engine is quite awkward, so you can't usually use just one of them, especially in landers. Most of the time, you need at least 6 tonnes of engine mass. If the payload is not large enough or the delta-v requirements are not high enough, there are usually better engines available. LF tanks are physically larger than LFO tanks with the same amount of propellant, making ships less maneuverable and landers less stable. Because other rocket engines use LFO, a mothership with nuclear engines can't support landers and other small ships from its fuel supply.

If the same jet engine worked on every planet with atmosphere, we could just reduce the clutter and bloat by removing the redundant planets. Choosing the right engine for the task is the single most important part of KSP gameplay. Each engine has its own strengths and weaknesses, and the choice you make often affects everything else in the mission. For example, jet engines are small and powerful, but they only work in dense oxygen-rich atmospheres, and you have to carefully manage the flight profile to keep accelerating. Electric propellers would work anywhere, but if you wanted to use them on Duna, you would have to build a slow lightweight plane with huge wings. Nuclear ramjets would also work anywhere, but they would be too massive for Duna, and overheating would be a serious problem at high speeds.

What about a rescue mission from a retrograde orbit beyond Eeloo, while recovering all the stages used to get there? That's the rocket equivalent of your Laythe mission: a simple task made difficult by self-imposed rules. A more reasonable Laythe mission would be based on the standard launcher/mothership/lander approach, even if the launcher happens to be a spaceplane.

It should be 'sexacore', if anybody cared about consistency. After all, we have quad-cores instead of tetracores, so we should also have sexacores instead of hexacores. Mixing Greek and Latin is a cardinal sin.

It's always better to have at least 2x more memory than a single application requires, unless the system is devoted to a single purpose. 16 GB is enough, if the desktop is just a gaming console, but otherwise I'd go for 32 GB to avoid having to close and reopen applications all the time.

KSP is a game about waiting. Docking large ships just requires a bit more waiting than usual. I've had good results with one RCS thruster block for every 15-25 tonnes of mass. With half of the blocks firing at once, the acceleration is around 0.02-0.03 m/s^2, and you get decent changes in velocity with 5-10-second burns.

In practice, Laythe landers need more delta-v in the rocket stage than Kerbin SSTOs going to a low orbit. Reaching the mothership may require another 500-1000 m/s from a low orbit, while the margins should also be higher, when you're far away from home.

That's the standard procedure for testing Laythe landers before the mission.

It depends on the planet, whether I circularize or not. Around Duna, nobody cares, because the delta-v requirements are so low anyway. On a Tylo mission, I usually circularize the mothership, because it's cheaper and easier to have some extra delta-v in the mothership than in the lander. Around Moho, the situation is the opposite. I never circularize there, because the delta-v requirements for the transfers are quite close to what can be achieved with reusable transfer stages, while landing there is cheap.

Laziness is my most common type of failure. I just can't be bothered to do things properly, so I make guesses, take risks, and hope that it will work. If it doesn't, I reload and try again, perhaps taking the situation a bit more seriously.

Tylo is essentially a fast-paced Mun landing. You need a bigger lander, more powerful engines, and some piloting skill, but otherwise it's quite forgiving. Eve is just as nasty as it has always been. The hardest part is landing in one piece, with something capable of returning back to orbit. 1.0 reduced the size of Eve landers significantly, but compensated it by making the landing harder due to reentry heating and proper aerodynamics.

The TWR of the basic jet engine is about right. It's comparable to the engines used in modern airliners, as well as to the Pegasus engine used in the Harrier. Rotating nozzles could be handy, but more powerful jet engines would just be too ridiculous. VTOL jets don't work too well in the real world. They always have to sacrifice range and/or payload capacity, because the TWR of jet engines is so low.

The Twin-Boar is probably the most cost-effective engine for launching large payloads. This rocket delivers the fuel tank (and a few other parts) to a 120 km orbit for 107,484 (including the payload) without resorting to silly things, such as refueling or recovery.

There are two aspects to lander TWR. If you only care about efficiency, initial TWR 1.5 is already good enough, while 2.0 is very good. If you fly manually and want to make your life easier, the absolute value of g·(TWR-1) is what you should really care about. During the critical phase of the landing, you're probably flying almost vertically, and the absolute acceleration is what makes you succeed or fail. In low-speed situations, 3-4 m/s is already pretty good, which corresponds to TWR 2.85 to 3.45 for a Mun landing. In high-speed situations, you'll probably want to double that. In a Tylo touchdown, 6-8 m/s corresponds to TWR 1.75 to 2.0.

That makes more sense. It seems that your baseline memory usage is a bit higher and your problem threshold a bit lower, which is why the memory leaks crash the game for you but not for me. I think memory leaks have been identified in three situations, which may all be related: Just playing the game. Memory usage seems to increase slowly, when you rocket is sitting on the launchpad. This is pretty minor, and I'm not sure whether it really is a memory leak. Scene changes, such as repeatedly launching a rocket and reverting back to the VAB. This is the fastest way to leak memory. Asset caching. KSP loads all assets on startup, but it seems to cache a second copy of the assets you have actually encountered during the game. If you keep launching and reverting the same rocket, the memory usage doesn't really increase after the first two attempts. If you use different parts in the rocket, the memory usage just keeps increasing. Similar things happen, if you fly for long distances close to the surface. Right now, the only way to avoid crashes is reducing the baseline memory usage. Relatively small differences in the baseline can mark the difference between frequent crashes and a stable game. This is why I believe that a 64-bit version would solve the problems. Until then, Master Tao's workarounds are your best hope. (And forget even trying to use part mods.)

Is your memory usage really that low? I've never seen such low figures, unless I lower the graphics details to an absurdly low level. What is your exact memory usage in the Kerbal X test I described?