June 14, 2018
6 min read

Warsaw video-calling: outsider reflections on Game Localization

Kirill Soloviev
Startup founder and CEO

Last week, I had the pleasure to speak at the inaugural Game Global Forum(organized by LocWorld) in fabulous Warsaw — a city where I’ve attended my first ever localization conference some 12 years ago. My last visit to Warsaw was in 2011 — practically in another life! So what was new, and why should you care?

Amazing new architecture in Warsaw (right) near the stalinist Palac Kultury i Nauki highrise (left)

Game localization is (exciting) business

line underH5 mask

Having spent my entire career almost exclusively in managing localization of enterprise software (initially at an LSP and then at a software company), I admit I was pretty ignorant of the scale — and explosive growth— of the localization solution partners that exclusively focus on adapting games to the needs of the global players!

According to Jaime Gine, COO of Keywords International, the global gaming industry has now grown to approx. $120B and is second only to traditional TV in market size (while dwarfing other media such as film and music). Game localization, estimated by Jaime at approx. $1B globally, is a noticeable part of the larger $5–7B game dev support services market that supports the growth of the gaming industry — and also a highly lucrative segment of the $50B language services market.

This clearly shows that localizing games, while being fun, is also good business. With trends like e-Sports sweeping over the world and China beginning to export more and more digital goods, it’s reasonable to assume that it will stay a good business and quite possibly become even better.

However, game localization seems suffers from the same challenge as the localization market as a whole: extreme supplier fragmentation. Even Keywords, probably the top player in the field, still takes up only ±15% of this market with its €150M annual revenue — despite a very agressive acquisition strategy and a highly diversified service portfolio beyond traditional localization (incl. testing, voiceover, art creation, and multilingual support).

Games are almost software, but not quite

line underH5 mask

Software is eating the world, as Marc Andreessen has famously said in 2011. And all games are software… right?

Well, yeah. Kind of…

Actually, no. At least — not really.

Games are entertainment. Games are fiction. Games are narrative. Games are immersion. Games are multimedia. And only somewhere deep deep down below, where only the most hardcore engineering geeks dwell, games do indeed become software.

From our industry’s perspective, this means several significant differences from the usual software localization routine:

More creative textual content

line underH5 mask

While not all game text is created equal, it’s intuitively understandable that when you want to deliver a compelling experience for your players, you need to approach it a bit differently from writing your marketing copy or your software UI microcopy.

Down the localization supply chain, that often boils down to very different type of LSPs and translators getting to work on games. It also means major hurdles in MT adoption: as of June 2018, even major AAA gaming companies such as EA and Blizzard have still not deployed any MT in a production setting (and I’m not even talking about MT for in-game content, mind you).

Voice is king

line underH5 mask

Both major game localization buyers and large game localization vendors at Game Global Forum were very vocal (pun intended) about localized voiceoverbeing the most important part of the perceived quality for their games.

While neural networks make major leaps in synthesizing realistic human voices, it seems that they are still very far away from the limitless creativity required to deliver an immersive experience for in-game voice acting. It means game localizers need to care about voice quality, a lot, and work with human voice actors all around the globe. And humans make costly mistakes.

Speaking of quality: no attendees I’ve spoken to were aware of any structured framework similar to MQM (which I’ve covered extensively in my talk and workshop during Game Global Forum) that could support systematic evaluation and measurement for voice localization quality.

Perhaps this is something that TAUS could undertake with help of major players like Keywords? This would enable game localizers and producers to make better decisions for their localized games from an entirely holistic view of quality across text, voice, and software aspects.

Testing player experience

line underH5 mask

One thing in common between software localization and game localization is of course the testing component! Both Functional (FQA) and Linguistic (LQA) testing are routinely used as an important part of many game localization projects.

While smaller indie studios might still not be able to afford decent LQA coverage (and this might be a problem that needs solving), it’s still well understood (and accordingly, financed) as a very important driver of global player experience by major gaming companies.

However, testing is both expensive and lengthy even for larger industry players. In order to drive testing costs down without affecting quality risk, game localizers will benefit from learning how to quantify & manage translation quality upstream — and then combine the translation quality perspective with the testing quality perspective.

I believe that MQM-DQF can help provide a unified localization quality perspective between software and language. That’s why it was the subject of my talk “Two sides of one coin” — you can get the slides here.

It’s all about the people — and they ARE different

line underH5 mask

I was able to talk and listen to a lot of different people at Game Global Forum (note: organizers reported around 100 attendees in total). Most of them were localization managers and directors working either for large AAA gaming companies or for highly successful mobile gaming companies. Others were functional or linguistic testing leads for the same. And of course we had our fair share of vendor-side executives, and some technology vendors as well (yours truly included). In other words, your usual localization conference crowd. Or is it?..

Like I’ve said, it was the first time I have spent 2 days locked in a room with gaming (localization and testing) professionals. The more I talked and listened to those people, the more I had a feeling they are subtly but significantly different from our usual software localization kin.

They are slightly more motivated. They are slightly more focused. They are slightly more driven. They are slightly more excited. They are slightly more open. They are slightly more vocal. They are slightly more tenured, majority of them having worked for the same company for 10 to 20 years!

In other words, game localization people are slightly more crazy. And I’m saying this in a very positive way!

Jaime from Keywords mentioned that this industry needs to provide people opportunities to retire while working for the same company. However, the prerequisite to that (beside having sufficiently large corporations) is people being able to thoroughly enjoy what they do for years, and years, and years on end. And from the outside, it looks like employee retention in the game localization industry is very, very strong!

Now, did you also mention AI?

line underH5 mask

Oh, right. I was at LocWorld 37 in Warsaw, too, and we had a bit of a chat on that with some amazing people from Google, Yandex, MemSource, etc.

Good news: Not many of the people working in the localization industry will have to do dumb work in the 21st century. More creative freedom for game localizers, and less room for boredom!

Bad news: When we say “AI”, we still almost universally mean “(Neural) Machine Translation”.

While that’s absolutely reasonable for the non-language professionals (= for 99% of the world), it’s using AI in management automation and decision support that should become an equally hot topic for us industry insiders.

Oh yeah, and there was also talk about blockchain. But more on that later.

Learn More