Civ VII Post-mortem: Crafting a redemption arc

[GeneralZift]

YouTubers are damned when they say something positive and damned when they say something negative, it's rather funny and reminds me of what game journalists go through.

Funnier is that both of these groups have this stage where they're revered and assumed to be correct. Then a backlash phase. And then upcoming, a phase where they are ignored entirely.

You have to take things at face value. YouTubers and journalists get flown out, they get early access, they have a certain level of inherent bias towards the thing that they're covering.

They're also pressured to come up with opinions quite quickly, it's not like they have proper time to take it in. We have to add the new-game bias too - where the first dozen or so hours of any game is usually pretty good, and the real weakness comes in the next couple dozen (from repetitive gameplay or gameplay that gets boring over time)

Most people also come with some prejudice based on what they like, what they think the game will be like, and popular opinion. Most games from Major franchises are rated 8,9,10 before anyone even sinks their teeth. How do you know the game is perfect within 2 days of release?

So long story short, I just don't trust anyone's word upfront. The sum of many opinions is a better way to gauge the truth about a subject. But also keep in mind that even the sum of opinions is dependant on the source (is it the target playerbase or not?)

 

[Kenshiro70]

Updated, with a section I really didn't want to have to write: "Were the layoffs avoidable?"

Last week has been devastating for a lot of people, and they deserve our sympathy and empathy.

But it's important to understand how organizations make these decisions. That understanding is key to ensuring it doesn't happen again.

 

[Siddorm]

At the end of 2024, the back of the envelope cost of Civ VII is $162m, all inclusive.

Wow. This shows how out of touch I am. I would have imagined that that would have been the cost through the end of year 2030.

 

[Kenshiro70]

Wow. This shows how out of touch I am. I would have imagined that that would have been the cost through the end of year 2030.

It's ballooned in the past few years. It's crazy, but $500m in revenue is no longer necessarily a home run.


(Source: https://www.brsoftech.com/blog/aaa-game-development-cost/)

 

[Kenshiro70]

I think I finally found my white whale. Civ VII had too many chefs in the kitchen - based on the game credits (link), while the number of Developers at Firaxis doubled from Civ VI and VII, the number of Designers tripled.

Note: this was after de-duplicating and removing people who were listed as Designers in the game credits but had other primary roles. Originally the credits listed 34 people as Designers.

If everyone in the room gets to have one of their ideas implemented, that dilutes the focus needed to get the features to a release-ready state.

Fewer developers per designer also means less resources to build each of those out, burning the candle at both ends.

If you’re looking for the smoking gun on the Rule of Thirds violation, this is probably it.

 

[itix]

I think I finally found my white whale. Civ VII had too many chefs in the kitchen - based on the game credits (link), while the number of Developers at Firaxis doubled from Civ VI and VII, the number of Designers tripled.

View attachment 742053​

Note: this was after de-duplicating and removing people who were listed as Designers in the game credits but had other primary roles. Originally the credits listed 34 people as Designers.

If everyone in the room gets to have one of their ideas implemented, that dilutes the focus needed to get the features to a release-ready state.

Fewer developers per designer also means less resources to build each of those out, burning the candle at both ends.

If you’re looking for the smoking gun on the Rule of Thirds violation, this is probably it.

This is interesting but please note that at least Anton Strenger and David McDonough left Firaxis.

Edit: You may also find developers who left Firaxis, changing the ratio again...

 

[Kenshiro70]

This is interesting but please note that at least Anton Strenger and David McDonough left Firaxis.

Edit: You may also find developers who left Firaxis, changing the ratio again...

Great minds think alike!

I'd bemoaned the exact problem that you're raising (credits tend to overstate the numbers)in the Team Sizes section (link).

Then, when I was looking for evidence of the rumored UI layoff last year, I managed to figure out something about the credits. In comparing the credits to LinkedIn profiles, I noticed that there's a euphemism in the credits for people who have departed. Anyone who left before the release was credited with the prefix “Additional” (e.g., “Additional Interface engineering”) and that you can use to filter them out.

Since the people hired to replace those people were also credited, I removed the leavers from the cost estimates, and as a result the estimates should be much closer to the actual numbers. For example, originally I had Firaxis at 279, but it dropped to 236 once I pulled out the people tagged with "Additional".

I'll never be as accurate as the internal numbers of course, but now it's enough within striking range that I feel confident enough to make basic statements about the data. For example, if there were 286 QA engineers in total (including 2K QA personnel), why is the quality in such shambles?

The Designer case had its own twist - it seems as though Firaxis handed out Designer credits like candy to people only tangentially involved. That actually required me to go through and find everyone with multiple credits and vet each person individually. You can probably tell that I'm a stickler for DQ.

I documented the assumptions here (link), but I'm planing to make my data set available. I'm in the middle of cleaning up the spreadsheet, annotating it with the assumptions and de-duplication logic, and parameterizing it so others can generate their own results. I'll let you know when it's ready for prime time.

 

[Gorbles]

Note: this was after de-duplicating and removing people who were listed as Designers in the game credits but had other primary roles. Originally the credits listed 34 people as Designers.

If everyone in the room gets to have one of their ideas implemented, that dilutes the focus needed to get the features to a release-ready state.

Fewer developers per designer also means less resources to build each of those out, burning the candle at both ends.

If you’re looking for the smoking gun on the Rule of Thirds violation, this is probably it.

The thing is, you can't diagnose on role alone. In games development, people often wear multiple "hats", or a broad "hat" that encompasses multiple responsibilities.

Take a non-games example: "senior software engineer". A common title, but it can mean completely different things to different companies. To different BUs within the same company, even (less likely, but I've seen standardised company title progression skewed due to complications from the size of the BU or product in question).

A lot of designers in games development often do bits in other areas. Some can code, some can't. It's very malleable.

And on top of this, you have a balance of mechanics-to-designer to account for. If a designer has 1 - 2 programmers whereas before they that 3 - 4 (subtracting art resource from the equation as that's normally a pool that gets allocated upfront based on the theorised cost for any given feature) . . . is that necessarily a negative? If the designer is responsible for less systems than before?

 

[Kenshiro70]

You're absolutely right; there's no black and white and people do wear multiple hats. But if I didn't de-duplicate, the number jumped to 34. Because it was such an important item, I scrutinized each of those 34 people individually when I winnowed them down to 12.

Also, I view Designers as the game industry equivalent of Product Managers on the enterprise side. At the end of the day, the requirements a PM documents and the gameplay mechanics a Designer creates become the authoritative statement on how the game plays. Both the PM and Designers are directing the engineers on the results they need to program to create. In effect, they are are "spending" a developer's time (=£££) when they do this. For that reason, there's only a limited number of people with that spend authorization.

As to the issue you raised around the ratio, fewer developers means they have less time to consider and deal with the exception cases. They spend most of their time handling the "golden path" case (i.e., when everything goes right), but when there are interactions with other systems, error conditions, and additional permutations (e.g., new Leaders and Civs) they don't have the time to anticipate those situations and create a robust solution in advance. Add in a brutal schedule, and it's very likely the code will need more upkeep later, which over time builds up a level of tech debt that slows down new development.

Sorry if I'm babbling on a bit - it's not every day that someone shows enough curiosity about these things that they'd be interested in discussing it. You made my day.

 

[Gorbles]

It's something I'm having to consider more and more at work. It really depends on the resource allocated to projects, the complications that come up, and how much domain experience the developer has. It's not strictly-speaking a competency thing. A senior developer that's come in fresh to an established stack (which games projects often are, and given Firaxis has a decades-old in-house engine, this matches up perfectly) is going to struggle (but still catch up faster with the proper onboarding) compared to a mid-level that's been there a few years.

I work with a guy whose title doesn't distinguish him for most of the other developers I work with, and yet he's the go-to guy for anything in a particular project (or language, even). That's domain expertise (plus some heavy backend specialisation ). He can cover the golden path and robust handling of edge cases faster than I can, in that project (and by title, I'm one up from him, and have committed plenty of improvements and additions to said project over the years).

I find the stats you've provided interesting, don't get me wrong. I just worry they'll be taken as a strong correlation to the end result when we don't know a lot of the internals. For example, you could have a fantastic QA department, and all bugs could get shunted off to an endless backlog. Then it wouldn't matter how many designers, programmers (or QA testers) there are.

I'm rambling a bit, mainly a you to me to you kind of thing, I guess it's not that interesting to the rest of the thread / will just be taken as posting in defense of Firaxis. It could be too many chefs, for sure, but it could be a pre-existing resource problem (given programmer salaries) that requires designers to double-dip more and more, or it could be something else.

 

[UWHabs]

I'd never put too much faith in job titles. What one company calls a designer might be very different from another. I mean, nothing says someone can't "design" a feature, and also spend 80% of their time programming it as well. Or they might have been a programmer for the last 5 years, but they got promoted to a designer role in the lead up to the launch, and didn't actually design anything that made it to the launch version.

Some companies you need to be on the job for like 10+ years to be called a Senior Software Developer. Other companies probably have someone one year out of university as a Senior Developer.

Now, that said, there's definitely some sort of change between 6 and 7. But we really don't have any idea what it all means, if it's at all related to any of the issues or feel of the game, or what it means.

 

[j51]

I finally got around to reading the OP and have to say that it and the current conversation in this thread is really illuminating. I’m learning a lot about managing software development. Thanks!

 

[stealth_nsk]

I finally read the significant portion of OP too and it was quite painful to read, honestly. I respect the amount of work put there, but there's a reason why essays usually require multiple people to make. When you build logical conclusions in your head alone, they could look fine, but you don't notice how questionable the things are and when you build something on top of this shaky foundation, it usually goes wrong.

I started taking notes, but stopped after first two subsections, because that would lead to the reading of equal size. But here are those two just to show where it goes:

I disagree badly on the section "Civilization VII’s war on Agency". Each of it's points is untrue:

• The game doesn't take medium and long term decisions out of player's hands. Things which affected those decisions, like city placement, specialization or wonder building, still work.
• Arbitrary strategies and milestones for win conditions are part of each civ game. Civ7 is actually less affected than, say, Civ6, because you don't have to choose your victory until the last age.
• There are no arbitrary caps on settlement limit or trade range. Settlement limit is a soft cap and there are ways to increase trade range. Both areas are for strategic decisions.
• I only partially agree that age reset destroys your momentum. Yes, it's kind of abrupt and some yield numbers temporary fall, but all the things you've done continue working.

Legacy paths are connected to one more turn killing, but in a different way. Victory conditions were always disconnected from significant part of the game in Civ and one more turn can't be defined by victories or legacy paths because it by definition happens after the victory. The problem for Civ7 here is really far from core - it calculates game score based on legacy paths, instead of taking into account all player achievements. Yes, the fix to one more turn is to implement game score similar to previous game, leaderboard and end game graph.

 

[AD1730]

The Designer case had its own twist - it seems as though Firaxis handed out Designer credits like candy to people only tangentially involved. That actually required me to go through and find everyone with multiple credits and vet each person individually. You can probably tell that I'm a stickler for DQ.

In film production, there are legal requirements to receive certain types of credits (these have been negotiated by the relevant unions, i.e. producers, actors, crew, cinematographers, etc.). Seems like it would be good to have a similar thing in the game production space.

 

[SupremacyKing2]

Not sure if this has been mentioned before but another issue I would add to the civ7 post-mortem is not to overuse % bonuses. There are videos out there where players take advantage of stacking % bonuses to get insane yields that totally break the game. So I feel like % bonuses should be used sparingly and carefully so not as not to create these issues of OP yields that ruin the game.

 

[Krikkit1]

Not sure if this has been mentioned before but another issue I would add to the civ7 post-mortem is not to overuse % bonuses. There are videos out there where players take advantage of stacking % bonuses to get insane yields that totally break the game. So I feel like % bonuses should be used sparingly and carefully so not as not to create these issues of OP yields that ruin the game.

I think most of those were fixed once they used the correct math (+% Growth=/= -% Growth time)...+% is fine

 

[Kenshiro70]

Ok, I finally was able to get the credits data cleaned up so that you all could take a look and play with it to your heart's content (link). You can go to the File menu and "Make a copy" and have your own to play with. I did a lot of work marking Product babies, Emotional Support Animals, etc. so that they could be filtered out.

As others have already mentioned, be careful of reading too much into the data. I did a lot of DQ work, but it's still looking from the outside in -- assume it's 70-80% accurate at best. Some caveats:

• It's a snapshot from February 2025, and doesn't reflect anything after.
• The credits for Firaxis list everyone who was involved over the development period. I'm guessing that started in 2020 or 2021 base on a a LinkedIn profile of a contractor mentioned testing out a multi-platform approach for a an "unannounced AAA game" in 2021, which I'm guessing was some of the early single-stack architecture pilots.
• As the credit reflect everyone involved, they would overstate the numbers since they would include people who left. However, I noticed (and confirmed with spot-checks on LinkedIn profiles) that Firaxis labeled people who left as "Additional" (e.g., "Additional Writers") so I marked all of those so they'd be filtered out. Still nowhere near spot on, but at least in the same state as opposed to a different country.
• The Firaxis and Outsourcing credits reflect people who really worked on the Civ VI and VII, but the 2K ones are much less dependable. Those are more indicative of cost-accounting weirdness (think Hollywood film accounting where hit movies don't make a profit). I didn't put too much faith in those or use them heavily.

I mainly used the data for percentage-based comparisons of change between releases -- growth rates in team size from Civ VI to VII, etc.. as that minimizes the impact of data quality issues. For example, 26% of the people who worked ion Civ VI were still there on Civ VII. For a ten-year gap, that's actually pretty in line with standard attrition (about 9% per year).

I realize very few people will be interested in digging into this, but I did feel it was important to "show my work" around statements I made in the post-mortem.

If you think what I did here was horrifying because I made a fair amount of assumptions around an incomplete data set, you'd be correct. But often this is about all you can get and then you have to make hypotheses and test them out.

Most importantly, the Firaxis designers are making similar analyses about our gameplay based on similar data they get from our playthroughs. If anything, their data quality might be lower than mine, as they've been using data badly warped by bugs on base yield calculations and the Modern-Age yield explosion. I at least knew going in that my data would be flawed and took steps to mitigate the impact; I'm extremely worried that the Firaxis crew took that data at face value when making balance tuning and nerf decisions.

 

[stealth_nsk]

Ok, I finally was able to get the credits data cleaned up so that you all could take a look and play with it to your heart's content (link). You can go to the File menu and "Make a copy" and have your own to play with. I did a lot of work marking Product babies, Emotional Support Animals, etc. so that they could be filtered out.

As others have already mentioned, be careful of reading too much into the data. I did a lot of DQ work, but it's still looking from the outside in -- assume it's 70-80% accurate at best. Some caveats:

• It's a snapshot from February 2025, and doesn't reflect anything after.
• The credits for Firaxis list everyone who was involved over the development period. I'm guessing that started in 2020 or 2021 base on a a LinkedIn profile of a contractor mentioned testing out a multi-platform approach for a an "unannounced AAA game" in 2021, which I'm guessing was some of the early single-stack architecture pilots.
• As the credit reflect everyone involved, they would overstate the numbers since they would include people who left. However, I noticed (and confirmed with spot-checks on LinkedIn profiles) that Firaxis labeled people who left as "Additional" (e.g., "Additional Writers") so I marked all of those so they'd be filtered out. Still nowhere near spot on, but at least in the same state as opposed to a different country.
• The Firaxis and Outsourcing credits reflect people who really worked on the Civ VI and VII, but the 2K ones are much less dependable. Those are more indicative of cost-accounting weirdness (think Hollywood film accounting where hit movies don't make a profit). I didn't put too much faith in those or use them heavily.

I mainly used the data for percentage-based comparisons of change between releases -- growth rates in team size from Civ VI to VII, etc.. as that minimizes the impact of data quality issues. For example, 26% of the people who worked ion Civ VI were still there on Civ VII. For a ten-year gap, that's actually pretty in line with standard attrition (about 9% per year).

I realize very few people will be interested in digging into this, but I did feel it was important to "show my work" around statements I made in the post-mortem.

If you think what I did here was horrifying because I made a fair amount of assumptions around an incomplete data set, you'd be correct. But often this is about all you can get and then you have to make hypotheses and test them out.

Most importantly, the Firaxis designers are making similar analyses about our gameplay based on similar data they get from our playthroughs. If anything, their data quality might be lower than mine, as they've been using data badly warped by bugs on base yield calculations and the Modern-Age yield explosion. I at least knew going in that my data would be flawed and took steps to mitigate the impact; I'm extremely worried that the Firaxis crew took that data at face value when making balance tuning and nerf decisions.

Thanks a lot! One minor correction - development started way earlier - Firaxis shown playable prototype to 2K managers in 2019. I'd guess this prototype used Civ6 engine (it's pretty common for early prototypes to be made on previous or common engines) and work on Civ7 engine started later after the general requirements for the engine were fleshed out.

 

[UWHabs]

So if we take that pivot table, basically the overall team grew by about 75% between versions. Of the big teams, the dev team grew by 21%, the art team by 41% (and the audio team by 22%) the 3 teams who were under that mark, and UI, QA, and a lot of the production and management teams grew by more than that amount.

What that means? Who knows.

 

[itix]

For example, 26% of the people who worked ion Civ VI were still there on Civ VII. For a ten-year gap, that's actually pretty in line with standard attrition (about 9% per year).

Thank you for explaining this. I was concerned that so few team members were working on Civ 6, but I had no idea what the standard attrition rate would be.

And of course, some may have been working on Civ 6 DLCs.