To HTML5, or not to HTML5: is that the question?

Please note: There was an epic comment thread discussing WebGL implementation of 
Quake 2 and outlining a lot of performance, compatibility and segregation issues 
HTML5 currently has. The thread was pulled out a day later along with my comments 
focused on importance of killer titles that every technology must have to "show the 
way" for both developers and amaze end-users. Luckily, I save my comments, so I've reposted them below.

A note to Gamasutra moderators: deleting MY comments in MY blog post and 
derailing further conversation WITHOUT NOTIFICATION AND CONSENT is WRONG. It was the only time when I made my blog post on Gamasutra EXCLUSIVE and now I regret that. Please comment WHY an important thread was deleted. 

I've been quite busy with traveling in last few weeks: from Moscow to New York launching new games here and there, thus the lack of updates.

Yesterday could've been a big day for web gaming. Google has started to push HTML5 content through it's new Chrome Web Store. As you might remember, Google's first large-scale HTML5 experiment was to embed an epic Pac-Man game into their logo, but wasn't really a "showcase" product since it did not expose the power of HTML5 - small canvas, no performance demands or memory hogging. (Tsshh, they even used Flash sound in their demonstration.)

At the same time, EA has issued an HTML5-focused press-release:  "Poppit! has never run faster and never looked better". This time EA's Pogo showcases a new HTML5 version of Poppit! called Poppit HD.  

I should say that I have some prejudice against Poppit. Actually FOR it, not AGAINST. I love Poppit. I played this game a lot, both original, online and Show versions. We love the genre: SameGame. Team members have experience from GameHouse' Super Collapse! 2 for Xbox Live Arcade and even did a 3D "same game" title called Archipelago.

The downloadable casual title Poppit! Show was one of the casual masterpieces both scale- and tech- wise. It used both PopCap Framework and Flash animations for main character. The game communicates with Pogo servers and used their Pogo account system. It's just like any modern online social puzzle game or Nexon's online casual titles.

It is plain awesome by online gameplay standards, allowing players to chill out in public chat/game rooms and actually exchange power-ups with other players while playing "survival" sessions. Show! was so cool that it inspired me to create GamingOS.com, a mix of Xbox LIVE and Game Trust online platform, now used by a local social network and online casinos.

So what's up with HTML5 Poppit! HD? Even though there are plenty of AS3 games working fine at 800x600 and even above, Pogo decided to port a single level of the game to HTML5. (Google has hinted that you need to use HTML5 to get featured in a web store).

Judging from EA's fancy press release, it looks like they have a special arrangement to bundle a trial version of a puzzle game along with a browser. Kind of a... KILLER TITLE?! Think of X360+Hexic or GameBoy+Tetris but just with 1 level (and not even a survival or blitz mode). 

Technically speaking, Poppit! HD is a basic JS game, not a HTML5 masterpiece. It uses canvas (and Google's own excanvas library), sound libraries and webm which could be easily compared to BitmapData/Sound/Video methods of Flash API. The game doesn't have any fancy resource packing.

It just loads dozens of unoptimized transparent PNG images and is an intense traffic hog, yet launches quick and seamlessly, thanks to native browser image decompression. Good parts: good performance for a single screen JS game without scrolling, but after just a single round my notebook started to overheat, which hasn't happened to Flash for quite a long time.

This exposes the underlying problem: JavaScript. Doesn't matter if it's the last iteration of V8 or a newly announced blazingsquirrelquirrelhedgehog CrankShaft engine that will work even faster, JS will be a major resource and memory hog. A quick session puzzle game is one thing (and yet it overheats!), a long-session MMO AJAX game that eats gigs of memory and has 50x the code to process is another story. 

Just as Microsoft doesn't get Silverlight, Google doesn't get HTML5. Unless you will showcase a real-world example delivered to millions of players, the tech will remain irrelevant. A real world example is either MMO or a social game.

Zynga's Dextrose acquisition is also as weird, as acquisition of GarageGames/InstantAction by IAC (and we know how that plugin-based story has ended). HTML5 has a poor penetration, cross-platform issues, and cross-browser performance discrepancies. iOS and Android devices are just TOO SLOW - Poppit does not work on iOS and Entanglement takes forever to pre-load. Will IE9 and Firefox 4 increase penetration of HTML5? Yes, they will, but it might still be irrelevant due to compatibility and performance issues.

So what did Google and EA actually prove today? The obvious part - you can produce basic JavaScript games, just as you were able to produce them years ago. Flash is still superior. Perhaps they've missed their own EA 2D Flash MMO content on Kongregate? :)

I'd say that reality is far more harsh. With "superior" HTML5 version of Poppit! HD Google sounds like "we want to bundle a mediocre JavaScript game with our browser". It's not Peggle or Bejeweled and looking at PopCap's minimalist PvZ flash demo for Chrome Web Store, it appears PopCap didn't spend their precious resources producing HTML5 version for a reason; the tech is too rough for the current rate of development and the traffic that could be acquired from Google Chrome Web Store is not worth the hassle.

By now Poppit was installed by whopping 8000 players at Chrome Web Store. Meanwhile in last 24 hours innoWate had 80k mini-game installs in just a single local social network for AquaDreams. Ironically, Bejeweled 3 was launched on the same day and was probably snagged a mil time, overshadowing both Chrome OS, Chrome Web Store and Poppit. And perhaps was far more playable than Cataclysm that day.

Summary:

HTML5 without an actual working dev tool chain, first-party classy authoring tools and improved JS engine performance across platforms will not be the best choice for at least another year.

Right now real-world penetration is as big as Unity or Silverlight due to quirks and segregation, meaning it "won't work everywhere" when you want to utilize all those APIs at once. You can install a missing Unity/Silverlight plugin for your favorite browser, but not so for in incompatible HTML5 implementation.  You'll have to switch your browser due to quirks and compatibility issues. 

Does anyone remember SVG buzz? Adobe was battling Macromedia with SVG+JavaScript until they figured out it just won't work and acquired Macromedia. Will it repeat once again? Hmm...

Everyone is pushing soooo hard not to repeat that. Let's wait and see, but EA didn't revolutionize browser gaming with this latest release.