A Cloud Appears Over Sega of America's Rapid Progress (new article translation)

[Gryson]

This article is from March 1995 and looks at Sega of America through the eyes of Nakayama and Japan.

A Cloud Appears Over Sega of America's Rapid Progress

Cliffs:

• The 16-bit market decline was in full effect by March 1995.
• Nakayama was considering giving SOA "head office" status. As I understand it, this meant SOA would have been more in charge of its own market, with less going through Japan.
• Nakayama apparently placed great trust in Joe Miller (hello 32X).
• There is a claim that Sega only sold 2/3 of its sales target for its best-selling game at the end of 1994 (presumably Sonic & Knuckles).

It's an interesting article that puts into context a lot of the decisions being made in 1995.

[fr0zenbuttox]

nice work gryson, is it possible you will find the confirmation that sonic xtreme was infact planned to be a 32x game anytime soon?

[Team Andromeda]

nice work gryson, is it possible you will find the confirmation that sonic xtreme was infact planned to be a 32x game anytime soon?

It was planned to be on almost every SEGA system at one stage, that was part of its trouble and aren't there video's around of the 32X version

[Gryson]

nice work gryson, is it possible you will find the confirmation that sonic xtreme was infact planned to be a 32x game anytime soon?

Isn't this already well documented?

Sonic X-Treme at one point was called Sonic Mars for the 32X. Chris Senn has talked about it a lot.

https://web.archive.org/web/20080422...treme/FAQ.html

http://info.sonicretro.org/Sonic_Mars

[fr0zenbuttox]

Isn't this already well documented?

Sonic X-Treme at one point was called Sonic Mars for the 32X. Chris Senn has talked about it a lot.

https://web.archive.org/web/20080422...treme/FAQ.html

http://info.sonicretro.org/Sonic_Mars

i never thought there was a official confirmation til now, i wasn't 100% convinced the rumor was squashed on the majority internet consenses, i was hoping a more recent source confirmed my interest that the game was planned for 32x, unfortunately these types of sites get lost to time, only for unfortunate consequences to continue festering to this day

edit: rip team andromeda

[Zeus]

Link to the original Japanese article?

[Gryson]

No link - it's only available in Nikkei's online database via library.

[Folco]

The most stunning nugget of information from the article, if true, is that Sega of America had around 600 employees dedicated to game development which is quite appalling knowing the the actual output produced.

[Gryson]

The most stunning nugget of information from the article, if true, is that Sega of America had around 600 employees dedicated to game development which is quite appalling knowing the the actual output produced.

Yeah. I'm curious what the actual breakdown was. I know SOA employed many play testers (like, dozens and dozens).

I wonder if the 600 included all of the development divisions: STI, Sega Interactive, Sega Multimedia Studio, Sega Midwest. Am I missing any?

I assume hardware development was also included in those 600, so throw in anybody working on things like Sega VR (which SOA spent $10 million developing).

There were presumably a ton of people working in support roles for SOA's second party developers. Producers, technical support, package/manual designers, etc. And people working to localize Japanese games.

But overall, the output was... not that impressive. Especially when you focus on just the output of the four development studios.

[zyrobs]

It all depends on how you count that. If you include any outside studios whose games they published (for ex. Novotrade who made Ecco the Dolphin, Blue Sky Software who did Vectorman), then the number can shoot up fast. The devil is in the details when you use statistics like that.

[Gryson]

I don't think they would include outside developers - those were independent companies, definitely not on payroll. It's not clear if the studios they acquired were included, like Sega Interactive, but probably yes.

[Barone]

Yeah. I'm curious what the actual breakdown was. I know SOA employed many play testers (like, dozens and dozens).

I wonder if the 600 included all of the development divisions: STI, Sega Interactive, Sega Multimedia Studio, Sega Midwest. Am I missing any?

I assume hardware development was also included in those 600, so throw in anybody working on things like Sega VR (which SOA spent $10 million developing).

There were presumably a ton of people working in support roles for SOA's second-party developers. Producers, technical support, package/manual designers, etc. And people working to localize Japanese games.

But overall, the output was... not that impressive. Especially when you focus on just the output of the four development studios.

SOA had a 24/7 QA department at the time, so that thing, alone, may have had 100-200 people allocated.
Their QA structure seemed highly inefficient. You'd get there any time, and a random game and the related testing task would be assigned to you. Anyone who has professional experience in QA will tell you that such a thing doesn't really work well.
Also, per interviews of former SOA testers, a lot of those people were hired fresh from high school without any background in the gaming industry and quality assurance, so you're talking about really unproven expertise in the field.

You can check that some SOA-tested games have 20-30+ testers credited, which is kind of absurd when compared to what some other companies were doing at the time.
And SOA-tested games have a history of game-breaking bugs, tech issues (slowdown, poor camera, poor music/sfx, unbalanced volume, etc.), blind hits, and nonsense difficulty spikes; stuff that smaller but more dedicated teams would usually be able to prevent (yeah, I know about the whole rental-related difficulty spikes thing, but even so, Virgin and other 3rd parties overall made a far better job).
Some of these same testers would then be able to ascend to producers, mostly thanks to Latham.

This whole thing never really made sense to me.
Let's check a few examples:

Tempo had no less than 38 testers. Thirty fucking eight testers for a game that has horrible camera issues from the start, terrible boss battles, etc.

https://www.mobygames.com/game/sega-32x/tempo/credits
Yet, I'm sure, the blame for such issues will be put on the Japanese developers and SOJ.


Chaotix, a game known for having game-freezing bugs, had 26 testers:

https://www.mobygames.com/game/sega-...haotix/credits


Star Wars Arcade is another one that is mind-boggling:

https://www.mobygames.com/game/sega-...arcade/credits

Lead Testers, Assistant Lead Testers, and Testers; 33 people in total for a game that has bugs that allow you to skip an entire stage.

The proportion between developers and testers also doesn't add up. 4 programmers vs 33 testers in the case of Star Wars Arcade.
As I said, to me it all sounds highly inefficient.


For comparison, Earthworm Jim 2:

https://www.mobygames.com/game/genes...-jim-2/credits
6 testers/QA people.


Global Gladiators:

https://www.mobygames.com/game/genes...iators/credits
8 testers.

[Gryson]

Yeah, that is insane. I can only imagine they were some of the first to go in the 1995 restructuring. I wonder if the huge numbers were just to make up for the fact that a lot of them didn't have any experience - brute-forcing the Q/A process to hopefully catch everything. Seems inefficient.

SOA also had a problem retaining staff. A lot of people left after just a few years, particularly in management. That kind of turnover no doubt contributed to problems related to completing games.

To grow a development company to that size that quickly must be hard to pull off.

[axel]

Yeah, that is insane. I can only imagine they were some of the first to go in the 1995 restructuring. I wonder if the huge numbers were just to make up for the fact that a lot of them didn't have any experience - brute-forcing the Q/A process to hopefully catch everything. Seems inefficient.

SOA also had a problem retaining staff. A lot of people left after just a few years, particularly in management. That kind of turnover no doubt contributed to problems related to completing games.

To grow a development company to that size that quickly must be hard to pull off.

With that many names, I picture most of them being interns who were thrilled simply to have a job where they got to play videogames.

[Barone]

Yeah, that is insane. I can only imagine they were some of the first to go in the 1995 restructuring. I wonder if the huge numbers were just to make up for the fact that a lot of them didn't have any experience - brute-forcing the Q/A process to hopefully catch everything. Seems inefficient.

Another questionable thing is what happened to games such as Dynamite Headdy.
Both Gunstar Heroes and McDonald's Treasure Land Adventure didn't have that many changes across regions. But for Dynamite Headdy...

All of a sudden you begin to replace bosses with the excuse that "oh, in America people won't like that" and change the difficulty of the game.
All of a sudden you need to retest a lot of stuff that was already done and ready for release. Which, "coincidentally", results in you now having an SOA producer and like 30 of their testers assigned and credited in your game.
...
A few months go by, it's the end of the year and you need to report to SOJ the cost of your QA team. At that time, it comes in handy that, all of sudden, you had one more game tested by you to split costs, result in a lower per game cost average and help justify that giant team you have.

A few things I'd also like to highlight:
- Let's assume you have a really offensive-to-the-American-public-looking boss. You then request the art to be replaced. If the new art has the exact same measurements as the original one, do you really need to retest that boss fight? No, you don't.
- Now how about if I request the same boss to be redesigned, with different dimensions and/or AI patterns? Now you need to retest the whole boss fight; oh, well, but don't worry, we have a 24/7 QA team, "the best in the world", standing by.

- By the same token, if I request a new difficulty setting to be added to the game and/or request the default setting to be the hardest one already available, do I need to retest the whole game? No, not all.
- Now, if I tell you that your game is unbalanced/too easy to survive a weekend rental and needs to be recalibrated following the instructions of our American market-specialist producer, do you need to retest the whole game multiple times? Yes, you do. But, hey, don't worry, we have a 24/7 QA team, "the best in the world", standing by.

SOA also had a problem retaining staff. A lot of people left after just a few years, particularly in management. That kind of turnover no doubt contributed to problems related to completing games.

To grow a development company to that size that quickly must be hard to pull off.

Yeah, companies such Atari and Crystal Dynamics were always approaching hiring people from SOA.