HavesteR shares his thoughts on recent KSP2 news

[moeggz]

 

Edited May 11 by moeggz

[PDCWolf]

Love how humble he's remained.

It was also very interested to hear that whilst "T2 kept SQUAD hired to develop updates for KSP1"... that "SQUAD" was literally nobody from the original team.

Edit 1: "Before kerbals went to space they had to learn to fly" -Planes before rockets confirmed canon.

Edited May 11 by PDCWolf

[FleshJeb]

That was an exceptionally good interview.

[RayneCloud]

I do miss that man. Good interview.

[herbal space program]

Thank you for posting that! His POV definitely makes it easier to see how we might have ended up here.

[shelshok]

I think the one question I would have liked to hear an answer to would have been:

"Having the experience of developing KSP 1, and being asked to do so, could you build from scratch a more modern polished copy of the game in a couple years that could be extended with the features KSP 2 promised going forward from that?"

In other words, would it have made more sense to have the original developer and at least some of that original team try again with the experience they had rather than having a new team try to replicate over a decade's worth of work?

[herbal space program]

3 minutes ago, shelshok said:

"Having the experience of developing KSP 1, and being asked to do so, could you build from scratch a more modern polished copy of the game in a couple years that could be extended with the features KSP 2 promised going forward from that?"

I think he actually addressed that  pretty clearly, although not in response to exactly that question.

[Skorj]

Really cool to hear he backstory on the team and the Kerbals.

[Superfluous J]

I still remember when everybody hated him because he didn't want to put dV readouts in KSP.

[K^2]

Good insight, predictably, and I completely agree that starting with a colony game, and then bolting on all of the KSP1 features over time would have been a better start for the KSP2. I think if Early Access was a functional colony builder with barely-working rocket game taped to it, we'd be talking about KSP2 in a more positive way.

 

I have to disagree with some of the points on Unity as an engine. I think HarvesteR might have a bit of a tunnel vision on it. No, I don't think building a custom engine for KSP2 would make anyone's lives better. I mean, it could make for a much better game, but it'd also inflate the budget by a factor of 10, easily. And it's not just because development would take longer - people they had working on KSP2 wouldn't be able to do this. Not in any kind of condescending way, to be clear. You need a team with expertise in engines. It's almost a different profession. You'd have to hire a whole another team of people just to build you an engine for the gameplay engineers to be able to make a game in it. It's a huge amount of work. It can be worth it, but not on KSP2 budget.

But you don't have to go fully custom to have a lot of custom code. Unreal and Godot let you rebuild huge chunks of the engine. Neither of them are built for a game spanning interstellar distances any better than Unity, but you can rip out the parts that don't work and replace them with ones that do. Unity's plugin system is painfully limited in contrast. HarvesteR brought up an example with the origin transposition, and it's all true, but also the KSP1 code had to keep addressing it ad-hoc everywhere it was relevant. And that's bug-prone and scales poorly. Whereas in Unreal I can build a custom root component and have it take care of the coordinate system changes, and now any actor using that component just "knows" where it is. It's a better solution to the same problem with custom code in a canned engine solution. And then there's performance problems that I can dedicate probably an equivalent of a few book chapters. Bottom line, going with Unreal or Godot wouldn't solve the problems magically, but it would expand the toolkit for solving them.

There are other tradeoffs as well, where the toolchain and compiling pains just keep growing with Unity at a faster rate as you work on a larger and larger game. There are a lot of Unity problems that HarvesteR seems to have accepted as just game development problem in general, and they aren't really. A lot of them are much less prominent on large projects in other engines. Of course, there would be negatives to going with Unreal or Godot as well. On the net, I wouldn't call Star Theory going with Unity for KSP2 was a mistake, since the original pitch was for a smaller game. Intercept continuing on that path is a bit more debatable, IMO. But even there, it's not "The mistake that killed the project," as some people serve it. Nothing that Unity would be throwing at Intercept would be unsolvable. But allowing for adjustments in the team composition within similar budget, I am pretty sure KSP2 development would be a little smoother with another engine.

 

One last bit that caught my attention is when HarvesteR talked about Kerbals in other games. It made me feel like he'd totally approve of a Kerbal Kart game, so long as it was wild and reckless kind of kart racing, which I think it basically has to be, and this warms my heart. I really hope that T2/PD consider it at some point.

[Skorj]

Game engines aren't my area of expertise, but from what I've heard, Godot wouldn't not have helped the game's rendering performance problems.  Unity just has a lot more effort put into core optimization (and Unreal even more so).   I haven't heard anyone describe Godot's 3D performance as anything better than "it's acceptable now".  And of course the huge downside of Unreal is C++.  I used C++ professionally for a dozen or so years, and I'd never go back.  Not to revisit the usual language holy wars, so let's just say the cognitive load for writing/reviewing C++ code is just a lot higher than C# for most tasks, because of all the landmines and historical baggage of the former.   When trying to find a complex and subtle bug, the fewer things to consider that might be wrong, the quicker the process.

I'd just take a very different approach to this kind of game.  For a game with colonies, I'd want to effortlessly support thousands of objects in stable orbits.  There's a way to do that with fairly simple and straightforward code, by using a solid understanding of real physics rather than doing the work with game engine physics.  But this forum is probably not the right place to discuss that.

 

[Cryptobux]

Hey @Skorj, I reckon this forum is exactly the place for (generalised) discussion on the right way to develop a KSP-like game. We're all nerds here.

[Skorj]

Well, I'll summarize by saying that given a "patched conics" universe and stable orbits (not under boost, not crossing SOIs), you don't need to track the location of an orbiting object frame-by-frame to know where it will be at any given time in the future.  This is a "well-solved problem" in physics.  Sure, you have to approximate using numerical methods, since the equation for the objects position as a function of time has no closed form solution, but you can approximate to the accuracy of a double-precision float very quickly, and FP rounding errors won't accumulate.

Stable orbits is what we'll care most about, since that's every commsat, scansat, orbital colony, and so on.  So from the start you want a solution that scales to many thousands (or millions) of them, so that it's a non-concern performance-wise in normal gameplay. 

And of course part-welding is a huge performance win for rockets under boost/atmosphere.  Kithack nailed that IMO: model joint stress, but as far as game engine physics goes a rocket is 1 big part.  A lesson that HarvesteR would surely have shared, had he been asked.

[cocoscacao]

5 hours ago, K^2 said:

Kerbal Kart game, so long as it was wild and reckless kind of kart racing

Where you build custom karts/vehicles? That would be similar, but awesome nonetheless. Taking power/mass/aerodynamics into account...

[shelshok]

13 hours ago, herbal space program said:

I think he actually addressed that  pretty clearly, although not in response to exactly that question.

Maybe I missed it, but I meant more specifically him and his own team rather than a team that just inherited the project.

[TMS]

13 hours ago, Superfluous J said:

I still remember when everybody hated him because he didn't want to put dV readouts in KSP.

LOL... as do I.  Pepperidge farm remembers.  And the gui skeumorphism.  And the the noodle nodes.  In his own words, he had to force the humour and kitsch.  But he was committed to the bit, and he did at least publicly (accidentally?) commit squad to free expansions for original purchasers. 

His awkwardness at the first KerbalKon was charming in comparison to the PR-heavy slickness of today.  It all seems so long ago now.

[herbal space program]

4 hours ago, shelshok said:

Maybe I missed it, but I meant more specifically him and his own team rather than a team that just inherited the project.

Towards the end, he got asked if he would want to try to create a KSP1 successor, and if so, what would he have done differently.  His first answer was no, he did not want to do that, because the way he described the process of birthing KSP1 you got the idea it nearly killed him, and he did not want to go through such an ordeal again.  After saying  that, he remarked that if he were them, one thing he would definitely never have done is to start by trying to create a feature-complete upgrade of the original game, because that is a very high bar to reach without actually creating any kind of  really new content that would represent a novel sort of hook. He then said that what he would have done in their shoes (and actually pitched unsuccessfully to Squad) would have been to create an aviation-based prequel rather than a sequel. If constrained to make a sequel however, he would have started by building out the colony mechanics without actual spaceflight simulation, and would only then have tried to incorporate all the KSP1 flight simulation elements.  Not quite sure how he would have managed that exactly, but I think he did make an excellent point that implies they set themselves up to fail with the approach they actually took. His philosophical statement about a good new gameplay hook being the first and most important foundational element, from which everything else ought to grow organically, also showed IMO that he apparently understands marketing  games to folks like us better than all the suits at TT put together. Lastly, he tacitly admitted that the source code he and his team put together for KSP1 was basically the product of a years long, pedal-to-the-metal hackathon, and fell so far short of industry standards for portability that it would likely have been of little use to even his own team, much less a new one. So anyway, I think he did touch upon your question in various ways, although perhaps not quite explicitly. 

[PDCWolf]

7 hours ago, herbal space program said:

he apparently understands marketing  games to folks like us better than all the suits at TT put together. Lastly, he tacitly admitted that the source code he and his team put together for KSP1 was basically the product of a years long, pedal-to-the-metal hackathon, and fell so far short of industry standards for portability that it would likely have been of little use to even his own team, much less a new one. So anyway, I think he did touch upon your question in various ways, although perhaps not quite explicitly. 

Whilst I'd like the planes-only prequel, there is value in recreating KSP1 but bigger and better. In fact, I'd say a lot of folks that are still here or that bought the game at any point wanted exactly that: better KSP1.

Then you realize they went with Unity again, same middleware, same mistakes, absolutely failed to expand on anything or iterate on any system, actually gimped other systems, set themselves for failure by making other systems horribly bad. In short, instead of revisiting the base game with more modern and better systems, they made a game that suffers the same foundational issues and doesn't really bring anything new to the table to justify them.

Of course, this doesn't contradict that there's lots of space for Kerbal-based spinoffs that aren't lego rocket + spaceflight sim.

[KincaidFrankMF]

10 minutes ago, PDCWolf said:

Whilst I'd like the planes-only prequel, there is value in recreating KSP1 but bigger and better. In fact, I'd say a lot of folks that are still here or that bought the game at any point wanted exactly that: better KSP1.

When HarvesteR suggested starting with a colony sim first, my first thought was: he'd be burned at the stake by the community for even suggesting it. But if it was pitched as a spin-off rather than a sequel, as "Kerbal Colonies", that could work. An achievable goal, easily monetised in the short term. And then expand backwards from there, making further colonisation the point of all the rocket stuff, until eventually you end up with KSP2 without actually promising that from the start. It could have worked.

16 minutes ago, PDCWolf said:

absolutely failed to expand on anything or iterate on any system

I think that's a little harsh, but only a little. The new science system in particular really improved the gameplay, for me, by providing an actual reason to go places. But then they hid it behind a science UI that literally couldn't have been worse if they tried, and I'm not sure most people got far enough into the game to appreciate it. It's certainly not enough to justify the vast pricetag, or even the original KSP1 price.

I think that's what ultimately killed it. Far, far too ambitious, so that everything ended up half finished and nothing was really in a state fit for the public, until someone made the fatal decision to push it into EA at nearly full price. Whoever made that decision ultimately killed the project. After that they were stuck making the game for a community that felt actively ripped off, and nothing short of amazing was ever going to be enough to convince large numbers of people to buy it. FS! was ok but a long way from amazing - and it didn't provide anything for marketing to work with because a science loop isn't new.

Maybe another year would have brought that wow factor. You can certainly chart a trajectory that goes abysmal->not bad->awesome. Sadly the actual trajectory turned out to be distinctly sub-orbital...

[K^2]

36 minutes ago, PDCWolf said:

absolutely failed to expand on anything or iterate on any system

Visuals are much improved, and while a lot of this could have been (and has) hacked in with mods, the framework that was created for KSP2 is a much better starting point. There are a lot of nifty things both the team and modders would have been able to do with planets and pushing visuals further.

Also, while colony building hasn't shipped, the framework that was being put in was quite general by the looks of it. Maybe modders will do something with it even in the current state. Though, I'm not sure a lot of people will bother with KSP2 now.

So unfortunately, yeah, most of these things would only really have mattered with 1.0, making it kind of a wasted effort in the current state. Currently, KSP1 can provide more of these features at quality with mods, so I'm sure that's where the community will hang around.

[PDCWolf]

1 minute ago, K^2 said:

Visuals are much improved, and while a lot of this could have been (and has) hacked in with mods, the framework that was created for KSP2 is a much better starting point. There are a lot of nifty things both the team and modders would have been able to do with planets and pushing visuals further.

Also, while colony building hasn't shipped, the framework that was being put in was quite general by the looks of it. Maybe modders will do something with it even in the current state. Though, I'm not sure a lot of people will bother with KSP2 now.

So unfortunately, yeah, most of these things would only really have mattered with 1.0, making it kind of a wasted effort in the current state. Currently, KSP1 can provide more of these features at quality with mods, so I'm sure that's where the community will hang around.

Visuals are... subjective at best: the overblown toy-cartoon style was one of the things I was hoping to mod out right away. Colonies could have an as amazing framework as they wanted, wouldn't really be useful when built on top of the bad multithreading and save/off-loaded-simulation system. Of course, MT was one of the things promised to get better someday(tm), along with HDRP, the PQS and so on, but we gotta judge what we have, because judging what they promised or showed on trailers... that hasn't gone so well.

[K^2]

53 minutes ago, PDCWolf said:

Visuals are... subjective at best

I'm talking tech. PBR is built into KSP2 materials from the start, so you could build good, realistic materials on top of it. People have done that with KSP, but it's hacked in at best, causing visual mods to trip over each other.

[moeggz]

1 hour ago, PDCWolf said:

Whilst I'd like the planes-only prequel, there is value in recreating KSP1 but bigger and better

Yup. The prequel sounds fun, I would’ve bought it. But colonies, interstellar, larger part counts, more resources, and “near” future tech that allows for brachistochrone transfers in late game is really the game I was looking for. 
 

[Bej Kerman]

47 minutes ago, moeggz said:

Yup. The prequel sounds fun, I would’ve bought it. But colonies, interstellar, larger part counts, more resources, and “near” future tech that allows for brachistochrone transfers in late game is really the game I was looking for. 
 

This is why I think doing colonies first is the complete wrong approach. People wanted KSP with colonies to extend the game, not a colony management game with a bit of KSP. Doing colonies first would also divert time away from optimising large part counts.

[herbal space program]

2 hours ago, KincaidFrankMF said:

The new science system in particular really improved the gameplay, for me, by providing an actual reason to go places.

I started clamoring for more of that in KSP1 probably six years ago already, and I was really happy to see some significant moves in that direction with KSP2. Driving around the complex, challenging terrain around the Duna monument and Stargazer Point in a rover, and then enjoying the artful rendering of the monuments while ruminating on the meaning of the cryptic clues they held was some of the most fun I had in my ~500 hours of  playing KSP2. A  lot more of the same, with the sequential mission goals eventually creating a trail to some required new tech for colonies/interstellar would probably have kept me happily playing the game for hundreds of hours more.