Shadowzone's findings on KSP2 history

[almagnus1]

That confirms a lot of what I suspected happened...  and also why I think KSP2 is dead at this point in time.  We might get bug fixes, but I doubt Take2 is going to provide the resources to fulfill it's vision and I don't think they're going to attempt a No Man's Sky with KSP2.

Here's to hoping Take2 applies what was learned (the hard way) with KSP2 and does a KSP3 right... but I doubt we'll see that within a decade, if ever.

[Vanamonde]

Overlapping threads have been merged. 

[chefsbrian]

3 hours ago, RayneCloud said:

Again, I question LGG's full view of the code base in the first place and yes I will bite that bullet and die on that hill.

I'm inclined to agree with LGG's opinion, both from my career developer experience, and my hobby gamedev experience. The problem isn't so much any single piece of code in the base, or any individual trend in the base, but the overarching nature of how its been assembled, based on what Shadowzone is presenting.

The starting point for KSP2 is apparently a relatively undocumented rip of the KSP1 source code, with no access to the original developers to explain why certain things went the way they did, which is incredibly important when it comes to just upgrading that codebase and its associated dependencies to newer versions of Unity. Furthermore, the goal at the time was effectively to polish it up, make it handle some newer shinies and features, and call it a day. That screams a light touch in most areas to make the deep investment places work, and with the limited timescale and budget, odds are the joins between the two are duct taped together. This isn't explicitly 'wrong' or 'bad' so much as it is the nature of the beast.

Then, the great Yoinkening happened, and after the dust settled, they had a handful of original mid and junior developers, and patched the talent with whoever else they could get in on relatively constrained payscales. At the same time, the proposed scope of the project ballooned. Suddenly, this minor upgrade with a few new pieces bodged into place until "It worked well enough" was expected to deliver on a significantly larger scope of work, including aspects it wasn't even fundamentally suited for anymore, mainly multiplayer. I cannot stress enough how significant the sudden expectation of multiplayer from the engine was to the codebase - True, no questions asked multiplayer is pretty difficult from the ground up in the best of scenarios, and a heavily physics and simulation driven game is already one of the worst scenarios possible. Trying to retrofit that into the build would require an extremely talented, driven team with a lot of time to do it, and a huge familiarity with the existing codebase. Not to demean anyone from the studio, but they didn't have any of that - Nor would it be realistic to expect that, in all honesty.

Despite now having a project scope that was about as close to incompatible with the existing code design as you can have while still having the game boot up at all, the mandate was thrown that No, you will use this, and no it won't be heavily refactored or majorly rebuilt from the ground up. Combined with the lacking senior leadership (Not any individual on this aspect, the general mix of tech and creative and implementation and QA and the whole nine yards) this sort of stew is all but guaranteed to generate even more spaghetti and quick patch code, and the rapidly approaching deadlines and project bloat mean that spaghetti duct tape is probably poorly documented at best, and undocumented at worst. This lines up with some of the extreme difficult on bugfixing as well - Bugs are a pain in the butt at the best of times, but with an undocumented codebase they're practically nightmarish.

Now, suddenly and with barely any warning, the entire development team is gone, every piece of knowledge they had with them is out the door too. Stuffs half finished, incomplete, sitting in various staging and testing branches. Trying to collate all relevant information from 70 employees isn't too hard, trying to collate all relevant information in any sort of organized useable fashion is a complete impossibility, a lot of stuff is probably lost in various documents, now cancelled meeting notes and invites, and generally a lot of incredibly important knowledge about various decisions is likely gone to the wind - if you're lucky, the final decision itself is documented somewhere, but good luck explaining "why" for most of them.

So you have an old codebase fit for a completely different scope of project, that has had not one, not two, but three cataclysmic events thrown at it, all while consistently underresourced or unreasonably constrained from achieving what they're trying to do with it, and then its been ripped outta its last set of safe hands right in the middle of what is likely some major surgery, and want to throw it to what has to be the third explicit, if not fourth in practice team that'll try and do the same with all the same mistakes and baggage? You're just gonna get the same result.

Assuming Take2 wants to deliver on the KSP2 roadmap as advertised on steam right now, then I am in agreement with LGG that the KSP2 codebase as it exists right now is a massive sunk cost fallacy, and needs to be sent out to die, unless you want to give it literally years of rehabilitation before its fit for purpose, and its still gonna struggle with the scars where its been stretched too far.

[dprostock]

Silent Hunter III

Silent Hunter IV  Wolves of the Pacific ( aka: Bugs of the Pacific )

Silent Hunter V Battle of the Atlantic ( aka: What's F....!? ) 

Der Untergang

 

[miklkit]

Watched ShadowZone's video and read this thread.  The lesson learned is that one cannot build a house on a foundation of sand.  It sure was a pretty house, but it has all crumbled into wreckage.  At this point it is best to start over with a strong foundation first, or just walk away. 

[Strawberry]

As some of the regulars on the forums might have noticed, I havent been nearly as involved in the community for several months now. I've mostly moved on from ksp2 just because other stuff began catching my focus, and only the whole studio closure debacle has made me pay attention again. Im surprised my "Star theory ksp2 was intended to basically just be a heavily modded ksp1" theory was correct. Honestly I mostly just feel bad for the people who worked for the game, especially the public facing ones, and I wish they all find greener pastures.

This whole story is filled with what could have beens, what if the studio was different, what if they didn't do the studio swap, what if intercept had X more months, what if middle management was better. Part of me wants to believe that this game could've been like no mans sky if things were just slightly different. A public facing head with starry eyes overpromises in spite of technical wisdom, wide whiplash when reality hits, then they internally acknowledge their mistakes and eventually more then make up for them. Intercept did in many ways improve, their organization improved, they were no longer working with a skeleton crew, etc. In spite of all of that though, it wasn't enough. At the end of the day as far as I can tell, the mandates that made this project such a mess still for the most part remain. Even if the ksp2 team internally acknowledge/acknowledged their mistakes, whatever higher up's called the shots that made it play out like this very likely won't. They were the ones who answered "it might have been" with "it couldn't have been". 

Unless if something big comes out, which I doubt, this will likely be one of the last comments I make on these forums. I'll reply to any replies if I feel like it and stuff like that, however in the past 5? 6? months I've only felt it worthwhile to comment on one thing and you're reading that comment right now. At this point there's not much left to be had, even if there's a studio move, which is increasingly unlikely, nothing will fundamentally change. I wish you all hope in finding greener pastures elsewhere. 

EDIT: Instinctively I went to open up the forums and see if I got a reply, you may notice this particular edit is two minutes after I commented this and to that I reply shush. I've never bookmarked the forums, instead I just press f and click on the first link for the ksp2 forums. This link is for some reason a thread I made nearly 2 years ago speculating about the original scope of ksp2 and I never really put much thought into this because why would I. I guess I got my answer somewhat? I still have no clue on some of the specifics but at this point speculation is as productive as anything else, though honestly that may have always been the case.

Edited May 25 by Strawberry

[Skorj]

Just catching up after having been away for a while.  Amazing video.  I do think Nate deserves real blame, because of his role as creative director.  When you're in charge, it's on you.  But as far as his direct actions, other than wobbly rockets I'm not sure what to blame him for; the font maybe.  Wobbly rockets were bad though.

I've been the lead on a project to fix an unsalvagable code base inherited from people you can't talk to.  While we were eventually victorious, I have to say firmly never do that, unless there's really no other way.  Our excuse was that we knew that some of the worst, most bizarre bits of code represented behavior that had been negotiated with customers, but in hindsight it may have been better to risk losing those customers than the expense of continuing with that code base.

It still seems to me that the deeper problem with the coding side was the lack of best practices.  There were bugs in core functionality they shipped with that never should have been allowed into the daily build, let alone production.  The worse your code bade, the more important this sort of thing is. 

[AstroEvada]

My biggest question is

 

WHY [Snip] DID TAKE TWO THINK IT WAS EVEN A REMOTELY [Snip] GOOD IDEA TO BE SO TIGHT-LIPPED AND SECRETIVE ABOUT DEVELOPMENT DURING EARLY ACCESS, A PERIOD WHERE YOU'RE SUPPOSED TO BE THE COMPLETE OPPOSITE OF SECRETIVE?! WHAT KIND OF GENUINE MORONS ARE TAKE TWO'S EXECUTIVES?!?!?!?!? "THE NEXT MINECRAFT" MY AFT END.

Honestly, my heart goes out to Nate and the developers for KSP2, they were being lead by executives (read: clowns) who have absolutely no clue how to do literally anything. I can tell these devs cared for KSP2 and KSP as a whole, I wish these devs well and hopefully they join a team that is lead by someone who actually knows what the hell they are doing. (I use large text for this part because it's easier for me to read)

Edited May 25 by Vanamonde
Redacted by a moderator

[KincaidFrankMF]

Awesome video, thank you so much @ShadowZone for revealing at least some of what TT should be telling us and clearly never will.

So the devs spent all these years doing the equivalent of taking a car apart on the garage floor and then failing to put it back together again while TT blindfolded them and kicked them repeatedly in the balls?

Wow. This has to be the textbook case of What Not to Do in game development  I had so much sympathy when I thought they were trying to redo it all from scratch and just found it far more difficult a project than they anticipated. This is just sad.

One thing I disagree with is that Nate's goals for the project were overly ambitious. Maybe so for a from-scratch development, but starting with an already working copy of KSP1 and building from there? Everything they were trying to do except multiplayer has already been successfully achieved by the modding community. No one forced them to do open heart surgery on the code in an attempt to update it to the latest version of Unity - sure, that would have been nice, but it wasn't necessary. Honestly, the roadmap is downright humble for a reskin of KSP1.

[Temporal Wolf]

I think people forget that when T2 bought KSP, literally the first thing they did was update the TOS and mandate the insertion of RedShell, all the while the PR folks were saying "this is just boilerplate, none of the changes meaningfully impact KSP, etc."

T2 is the cancer. I said it then and I'll say it again.

Let's also not forget that T2 forced the EA launch by the end of FY23, and apparently decided to close down the studio when they failed to deliver the 1.0 launch by FY24.

This has T2's fingerprints all over it. Doesn't surprise me at all to find out they were causing problems with the actual development as well.

Edited May 25 by Temporal Wolf

[aeroeng14]

8 hours ago, J.Random said:

There is something that I don't actually understand from this video: salaries. It seems like @ShadowZone (or whoever he was speaking to) tries to argue - at the same time - that KSP2 developers were all juniors with no experience AND at the same time that they should have been paid above the market. idk, maybe Seattle is some sort of CA corner in WA, but 150k/yr sounds like A LOT for a junior coder, especially considering that it looks like it's after the tax (later on, there are estimations of 200k/employee).

It does seem like a lot, yet compared to other big hitting software companies in the area it's already on the low end. Why the studio was put in Seattle is beyond me already inflating salaries due to CoL besides naturally competing in the tech space. I'm traditional fortune 500 engineering and we have a big presence in the Seattle area. We can't keep software engineers at any level to save our skin because we get outbid on compensation handily. Not just straight salary but stock options too. There's just no comparison what Microsoft, et al can throw to vacuum up talent if they want to and that's a better resume building opportunity, again depending what your career goals are. I won't chide people who go to greener pastures but it's a heck of a big ask to throw many tens of thousands of dollars away annually unless you want to be in gaming given the costs of everything nowadays if you want to settle in the PNW.

[ShadowZone]

9 hours ago, TLTay said:

Did anyone spill the beans on the other project the majority of the team was working on since EA release?

Less than 10 people were working on the second game according to my information.

[ShadowZone]

8 hours ago, J.Random said:

There is something that I don't actually understand from this video: salaries. It seems like @ShadowZone (or whoever he was speaking to) tries to argue - at the same time - that KSP2 developers were all juniors with no experience AND at the same time that they should have been paid above the market. idk, maybe Seattle is some sort of CA corner in WA, but 150k/yr sounds like A LOT for a junior coder, especially considering that it looks like it's after the tax (later on, there are estimations of 200k/employee).

Ah, maybe I didn't explain this correctly in the video:

The "only hired junior people" part was during the Star Theory days when they still tried to do it cheap and quick (10 million, two years). I don't know what the compensation cap was at that time.

The 150k was the maximum compensation for engineers at IG. They had trouble getting senior and lead people for that amount and more trouble keeping them. There was internal pushback against this cap, but corporate finance refused to budge. Again a decision that was made outside of the team directly and adversely affecting it.

The 200k/employee is "fully loaded" costs (I hope I'm using the correct term, we call it something else outside the English speaking world), not just salary. So per employee you also need equipment (PC hardware, office chairs, desks etc), rent (office space) and other overhead (HR department, legal etc). While some of the latter was offloaded to PD, there's still a more costs that pour into employing somebody. For instance in my country, an employee that gets 50k a year into their bank account (what actually lands in their pocket) overall costs a company double that amount.

The 200k/employee average was used as a rule of thumb number, but it was given to me by LGG as well as my sources independently from each other.

[WatchClarkBand]

5 minutes ago, ShadowZone said:

The 200k/employee is "fully loaded" costs (I hope I'm using the correct term, we call it something else outside the English speaking world), not just salary. So per employee you also need equipment (PC hardware, office chairs, desks etc), rent (office space) and other overhead (HR department, legal etc). While some of the latter was offloaded to PD, there's still a more costs that pour into employing somebody. For instance in my country, an employee that gets 50k a year into their bank account (what actually lands in their pocket) overall costs a company double that amount.

Yes, in the States, we managers call it the "fully burdened cost", and it tends to run from 33% to 50% over whatever an employees base salary is. Health insurance, unemployment insurance (paid to the state to cover support payment should the employee be terminated), taxes, the overhead of hardware (as you called out), all factor into this. At Amazon, I believe we used an internal number of $300k/year average for the FBC of any "professional" (non Fulfillment Center) employee, as the median total compensation (salary plus bonuses plus stock) for all employees below VP level was probably $225k/year (I could give breakdowns by role/experience level, but it's easy to Google or check on Glassdoor). Game engineers (and PMs, and UX folks) tend to be paid about 60-70% of the market rate for the same role in a non-entertainment company, based on my 25 years in software development in and out of games (that is to say, if I were to survey all the game studios at which I'd worked, people made, on average 1/3rd less than they could have made at a non-game company in the same role).  In case someone wants to nitpick, at the time Amazon also had a salary cap of $160k/year, but salary was a small percentage of total compensation for anyone approaching that cap, as stocks and bonuses could equal multiples of base salary at the Senior or Principal Engineer or Engineering Manager level. So it's definitely not an apples to apples comparison.

For obvious reasons, I will only speak about compensation at other companies based on my experience working for them.

[dave1904]

13 hours ago, K^2 said:

Well, I don't know, maybe when you see someone with experience making a thing defend someone else's attempt at making a thing, instead of being "sick" of it, try to understand why?

Maybe, just maybe, a person who knows how games are made has a better idea of whose fault it is that something didn't get made right? Just a possibility, maybe?

I know why they failed. Incompetence. It's pretty simple. Both T2 and IG or whatever they called themselves in the past made mistakes that are unacceptable. It doesn't matter what industry you work in. You start a project with a goal and you work towards that. If you own a bakery and your goal is to make a pie. Only an idiot decides to change the goal to a cake halfway through. Same goes to everything without exception. Just like the current goalless moon missions. They have no idea what the plan is. 

You're expirence doesn't make your argument any stronger because so far you've been clearly wrong over the years. You make up your own version without any evidence to support your claims. I have evidence. The game itself. After 7 years thats all we have? Sorry but chimps could do it better. We knew in 2020 that it was looking bad. So did you. You can make excuses for them then and try to understand why. The awnser is simple. T2 hired a studio that had no idea what they were doing. Plain and simple. AsI said. I can forgive that. It's human nature. I cannot forgive lies.He said they were playing KSP2 multilayer in 2020 and having great fun. The fact is they were playing KSP1 with mods. You can say that t2 forced him to say it. We know it's nonsense. Even if they did he didn't have the integrity to just keep silent. 

So forgive me for being the bad guy here but I'm not going to feel sorry for someone that gets what was coming for him. 

[Bej Kerman]

1 hour ago, dave1904 said:

I know why they failed. Incompetence. It's pretty simple. Both T2 and IG or whatever they called themselves in the past made mistakes that are unacceptable. It doesn't matter what industry you work in. You start a project with a goal and you work towards that. If you own a bakery and your goal is to make a pie. Only an idiot decides to change the goal to a cake halfway through. Same goes to everything without exception. Just like the current goalless moon missions. They have no idea what the plan is. 

You're expirence doesn't make your argument any stronger because so far you've been clearly wrong over the years. You make up your own version without any evidence to support your claims. I have evidence. The game itself. After 7 years thats all we have? Sorry but chimps could do it better. We knew in 2020 that it was looking bad. So did you. You can make excuses for them then and try to understand why. The awnser is simple. T2 hired a studio that had no idea what they were doing. Plain and simple. AsI said. I can forgive that. It's human nature. I cannot forgive lies.He said they were playing KSP2 multilayer in 2020 and having great fun. The fact is they were playing KSP1 with mods. You can say that t2 forced him to say it. We know it's nonsense. Even if they did he didn't have the integrity to just keep silent.

Funny how your "full picture" is to blame Star Theory for what T2 was doing and to deflect answers coming from someone who has actual experience in the industry. I would, at this point, admit you're just looking for someone to throw knives at an image of, and that you're being overly dramatic over a game.

1 hour ago, dave1904 said:

So forgive me for being the bad guy here but I'm not going to feel sorry for someone that gets what was coming for him. 

Everyone's the hero of their own story. Lol. 

[KincaidFrankMF]

One other thing that occurs to me. @ShadowZone so you're saying KSP2 was designed from the outset to be an on-the-cheap reskin of an indie game? But IIRC (anyone?) from the first announcements it was sold to us as a full price AAA game.

So Take Two decided to take us for every penny they could get, then penny-pinched the project to the point where it couldn't possibly succeed, got stuck in the sunk costs fallacy, and eventually lost millions on the project. I'm just... speechless.

[cocoscacao]

@ShadowZone I has a question. You mentioned colonies where supposed to be in EA launch, but the developer working on them was shuffled to different duties. This somehow implies they're nearly completed, yet, months have passed in developing them. Have I misunderstood something?

Great work by the way. Really appreciated. 

[ShadowZone]

51 minutes ago, KincaidFrankMF said:

One other thing that occurs to me. @ShadowZone so you're saying KSP2 was designed from the outset to be an on-the-cheap reskin of an indie game? But IIRC (anyone?) from the first announcements it was sold to us as a full price AAA game.

Yeah, the idea was to just take KSP1, freshen it up, stabilize the code base (no idea how they should have done that if nobody teaches you how it works) and add a bit more content. Think FIFA or Assassin's Creed franchise. The core mechanics remain the same, but some fluff around it changes.

Then they got enamored with the "interstellar, colonies and multiplayer" vision but somewhere along the line neither timeline nor budget were increased to accomodate such a re-imagining.

24 minutes ago, cocoscacao said:

@ShadowZone I has a question. You mentioned colonies where supposed to be in EA launch, but the developer working on them was shuffled to different duties. This somehow implies they're nearly completed, yet, months have passed in developing them. Have I misunderstood something?

Great work by the way. Really appreciated. 

So, a bit more detail maybe:
When T2 gave them the hard deadline of February 2023, everything was still in the game in some form (colonies, interstellar, multiplayer) to varying degrees of completion. Around May '22 there was the notion "it won't be pretty, but we might be able to ship this". The decision to go EA and rip stuff out again was made around September. They did try to get at least colonies in there, but somebody pulled the plug because the deadline was already close and they wanted some other thing in there (don't know which one).

I am wondering about the time it takes them to release colonies as well. Apparently "For Science" took already multiple times longer than was expected. One of the issues I was told existed within Intercept Games was a lack of "okay, that's good enough" (and we'll make it prettier later) attitude. If it gets public, it needs to be as shiny and pretty as possible, was the mindset (at least this was how the story was told to me).

In software development you always need to do some kind of tradeoff. Let's take re-entry heating: the mechanics of "part gets hot, part goes boom" with a simple heat gauge or flashing parts or something could have probably been integrated a lot earlier than what we got with "For Science" with all the pretty plasma effects etc. Of course it looks cooler, but with the simple "show me just a gauge of what goes wrong" mechanic you could have tested the math behind it and verify if your assumptions about how which parts will behave under thermal load. The way they did it was that everything had to be "complete" before shipping it. Also kind of misunderstanding what Early Access should be. I don't know who made these calls for re-entry heating. But I believe we could have gotten a stripped down version much sooner, same for science and colonies.

Build the mechanics, test them, test them with the public, make them prettier on the go. That's how I would treat an Early Access development title.

Edited May 25 by ShadowZone

[cocoscacao]

10 minutes ago, ShadowZone said:

everything was still in the game in some form (colonies, interstellar, multiplayer) to varying degrees of completion

Oof. If those features were shoved over KSP 1 reskined version codebase, this thing, even if continued, doesn't have a chance to become a good game... 

[MechBFP]

If the”make them pretty and shiny part” included “functional and reliable” then I can definitely understand that mentality because we all know things already came out rough as it is. I can’t even imagine the state of things without that mindset in play. 
 

If it was just graphics, then that definitely was a mistske, but unfortunately they painted themselves into a corner with their constant overpromising by that point. 

[Vexillar]

53 minutes ago, ShadowZone said:

Build the mechanics, test them, test them with the public, make them prettier on the go. That's how I would treat an Early Access development title.

Absolutely. The fundamental miscalculation seems to have been that the base code is "complete", easily understood by any passing programmer, and easily reworked. The lead work is all about the visuals. Even with all that has happened, IMHO the sound and visual design has been done well - though one of my first impressions was - "where are all the crew interiors?" Perhaps those were another casualty of people being re-assigned before the job was done. Testing seems to have been virtually AWOL until after EA.

[Meecrob]

1 hour ago, ShadowZone said:

Apparently "For Science" took already multiple times longer than was expected. One of the issues I was told existed within Intercept Games was a lack of "okay, that's good enough"

To be 100% honest, and I know I speak for more than just myself here, For Science! simply was not "good enough"

FS! felt like I bought KSP1 off of Wish.com or something.

1 hour ago, ShadowZone said:

If it gets public, it needs to be as shiny and pretty as possible

Lol, yeah, I felt like I had to wear sunglasses playing the game at times.

[dave1904]

2 hours ago, Bej Kerman said:

Funny how your "full picture" is to blame Star Theory for what T2 was doing and to deflect answers coming from someone who has actual experience in the industry. I would, at this point, admit you're just looking for someone to throw knives at an image of, and that you're being overly dramatic over a game.

To be clear. T2 is to blame for this. It's their IP and they had the say about everything. That doesn't mean that the developers are not responsible for their mistakes. Even the guys at the "bottom" are still responsible for working for a company that was run like because they all knew it. That makes them responsible. I have not once personally attacked anyone. I just said they were incompetent and stand by that point. 

Just stop defending the developers. They are not your friends. We payed them for a product and they did not deliver. KSP2 is done mate. I'd let go if I were you. I've accepted that in 2020. Most of us have. 

[m4ti140]

20 hours ago, RayneCloud said:

I respectfully disagree with @linuxgurugamer's statement that KSP2 as it stands at the moment is an entirely unsalvageable "Start over" project. I can't believe people are even entertaining this as an idea. With respect to all of his work over the years, this is just not a statement I agree with at all. As to this video, thanks for covering this @ShadowZone and thanks for bringing on Matt and Scott. Good work.

I have some issues with things said in the video, but I don't see much a point in going over that now. I'm just glad someone is publicly digging like this and calling out T2, but do not expect to ever see anything from them.

I believe he was referring to the code base, and unfortunately no, he's right, it should be nuked and rewritten. Like it should have been done from the start. You can reuse the assets, but the code is a waste of time and resource. It would literally be faster and easier to start over at this point.