You shouldn’t make changes for the sake of changes, while you may not like the UI that Steam uses, the changes you propose would decrease usability of the client for me.

The ‘boilerplate’ is far more useful to me than the details you want first, and in general, the stability of the Steam UI makes it far more usable than rapidly-changing UIs like Youtube’s, where you need to relearn features every other month… just when you’d gotten used to the changes.

jonathan /b/low

Pretty much all of the system services are implemented in as naive a way as possible, with the idea that they can always provide a better way later – which they never do. Often these naive approaches also preclude the developer from doing anything useful as a workaround, too; for example, the file system is designed such that any seek operation has to rewind to the beginning of the file and then read through it while discarding all the data you don’t want. Basically, anything I/O that doesn’t involve linearly reading through a file from beginning to end becomess an O(n^2) operation.

The virtual memory system is built on top of that, so it becomes so poorly-performing as to be useless. The virtual memory system also does not allow you to override the page fault handler, so you can’t use it to do anything USEFUL like using it to remap physical memory access (which is very handy if your memory space gets fragmented).

The C++ toolchain only begrudgingly supports exceptions, and buggily so, because “nobody uses exceptions” (which is only the case because the toolchain support is so poor). Often you have to rewrite entire codebases when porting to PS3 to make it not use exceptions, despite exception handling having been a part of the language since the 80s and generally being significantly faster than checked error returns (when used properly).

Once upon a time, the Sony tools folks were developing PSGL, basically a version of OpenGL ES which ran on top of the native low-level graphics library (gcm). The intent was to make it easier to program graphics while also improving performance in a central optimization point (because they had skilled people who could write SPU code for doing matrix math and culling and so on). For a while they were even saying, quite explicitly, to use PSGL for all new graphics code, because it was the future for the PS3. A couple years later they quietly dropped it and any support requests for it were responded with “PSGL is deprecated, just use gcm.”

There is a robust save API, but it’s built on top of a house of cards (so it still doesn’t do anything to prevent the hard disk corruption issue) and uses a ridiculously restrictive “dinosaur model” (as Jonathan puts it) which doesn’t allow for automatic background saves, so very few games use it (and the ones which do always generate groans of “Oh no, not THIS”).

The title-safe area stuff is further compounded by REALLY terrible support for resolution selection, and inconsistencies about legacy things like 4:3 displays and 480i vs 480p. But developers have to support it, even though the APIs for supporting it are TERRIBLE. It doesn’t help that the PS3 still ships with a composite cable and nothing else; so much for it being built for HD and nothing else, right? I wonder how many PS3s out there are hooked up to an HDTV via composite cable.

Pretty much all of the PS3 problems boil down to bad cases of not-invented-here syndrome, where the Sony engineers decide not to see how other people did it anyway because it’s just important to get something out the door (regardless of how cumbersome and clunky it is, and regardless of how bad the design is from a high level).

You must be also very influential to MS since I worked in localization for several major games (mass effect, forza, etc…) and we had a hard time to get waived some TRC violations even being first party dev. sometimes…
I couldn’t agree more than the 3 big ones are resting in their laurels when it comes to digital entertainment…

I have a question: Jon, have you ever considered doing a lecture/assembly about the value of more interesting, thought-provoking games for an audience of children at all? Seeing as they are essentially the ‘next generation’ of gamers, it would be very interesting to see exactly what is on their minds amidst their disillusionment that Call of Duty and FIFA are the epitome of video games, and the fact that you’re quite possibly the only person that would ever give them the opportunity to show them what is outside the box (because, let’s face it, the aforementioned titles are never going to do that for them) would be really valuable to them, I reckon.

Is this something that you have ever considered?

Thanks!

Gary

xx

Jonathon Blow using internet memes… well now I just don’t know WHAT to think!

Building screen-edge/corner relative UI (and/or UI scaling) is something we almost all do anyway when display resolution is a variable we need to support. Getting the actual display region, offsetting from there, and not punishing people with decent TVs is a win for everybody.

Compliance with ratings: because consumers want to be informed about the age/cultural appropriateness of the game before they buy.
(Steam is the only platform not to have a rating system in place)

UI consistency: because players expect to be able to pause the game with the START button, cancel with the B button and that the Memory Unit is consistently called “Memory Unit” and not “SD Card.” Being user friendly shouldn’t stifle innovation.

Legal checks: they do exist, at least in the form of warnings in cert reports.

Game stability: hire a new QA team. A crash or progress stopping bug is almost a guarantee for a certification fail.

I loved that as well! I was expecting a begging animation and a menu but I just got droped into the iconic bridge in front of the sunrise/burning city and I was this strange character in the dark! I didn’t even wait like 10 seconds and I was playing already. Without a warning I was playing, it was the game, no menu or animation or cutscene!

In that very moment at the begging, just seconds after leaving the console hub I was Tim and was exiting the bridge and going into the house and I said “This game is fucking serious!” Best game ever, really.

I hope The Witness has the most Epic opening ever, A dark hallway? That could be potentially cool, just depends on what’s behind the door at the end of the hallway! I hope The Witness is as fast or faster than Braid to load and begging but being 3D it will be difficult! I wish the whole team good luck! I hope you can get away with alot more that you couldn’t with Braid!

Kudos on getting those requirements waived / adjusted – it definitely had a “magic moment” effect on me as a gamer.

I spent about a year in the Xbox Platform team and at times found myself in the set of people making the call on developer requests to waive cert requirements. The discussion was always primarily about user experience robustness. In general, product / UX quality was valued much more at Xbox than at any other group I’d seen at MS.

The console industry’s DNA was forged in the memory that Nintendo’s invention of the certification process had saved it from collapse in the early 80s. Agreed that heavy robustness certification is irrelevant now thanks to ubiquitous connectivity (thus easy patching), crowd-voting / social media, and a self-publishing model that increases content supply by orders of magnitude, thus lowering prices and expectations. I agree it’s likely that next-gen consoles will undercorrect for this shift, but both MS and Nintendo have both demonstrated the ability to make bold (at the time) decisions and radical shifts from one console generation to the next. Of the three console makers, I’d bet that MS will go the furthest, given its experience with the more open model on Windows Phone (and Windows), its platform-driven mentality as a company and its relative youth in the console industry. Sony / Nintendo’s best bet would be to just adopt Steam itself as their primary content system and ride the wave more easily… but something tells me they won’t do that.

Don’t… Don’t do that please… You are so much better than that…
Yes, the guy is an imbecile but please don’t use internet jokes…
…Please no meme talk.

Much better is to tell him to go fuck himself or that white on black is the best combination of them all or whatever.

Why is ESRB ratings compliance a necessity? Steam doesn’t require it. Apple doesn’t require it.

Why is UI consistency a necessity, when the person who is making the UI rules is a far worse game designer than a lot of the people trying to make games for the platform? Doesn’t this prevent improvements and stifle innovation?

Certification processes don’t include any legal checks; how could they? There is no way to certify that. That is something that goes into a contract and is standard pretty much everywhere, so I don’t see it entering into this issue.

I agree that a game should not crash, but in my experience with testing on these platforms, the testers are not spending most of their time trying to make the game crash; they are spending most of their time going through a bunch of mechanical steps to test a bunch of standard TCR things, most of which are unnecessary.

Cool story, bro.

Yes, I wanted to get rid of a lot of the ugliness and bureaucracy that happens when so many games start up.

Note also that the game did not have a screen where it tells you to press A to start, if you only have one controller active (I got away with this because GTA4 had already shipped doing the same thing) and that it doesn’t ask you where you want to save if you only have one save device. (Though maybe it always asks now, since there are cloud saves and thus almost always two save devices).

These are all technically violations of the requirements that I was fortunate enough to get waived because I did the programming work to get around them.

But it would be better for the platform holders to do this stuff once than for everyone to have to do it all the time. (Well, they wouldn’t be doing exactly these things, but rather, related things to make the system more streamlined.)

Glad I read till all the way down here! This comment was the most informative thing in an already highly-informative article. It took me back to the first time I played Braid on my Xbox and how the game “just started”… earlier than I expected. The opening experience was a lot more surreal and impactful because of the suddenness of its onset. Of course a lot of that had to do with the way the opening screen presents itself (with Tim in the dark), but I imagine the lack of the save warning helped a lot too.

Now that I have this context I’m curious to know which was the effect and which the cause. I.e. did you put in the extra effort to skip MS’s save screen requirement because you wanted that suddenness in Braid’s opening? Going even further, did you implement your more robust save system in order to skip MS’s save screen requirement, in order to achieve the sudden opening?

Ironically, the presence of the save screen in every other game made Braid opening even more sudden than it would have been if MS had no save screen requirement to begin with.

Sometimes having rules is important though, it creates a language and unity for the users, but it would be nice if Microsoft/Sony/Nintendo at least once in a while worked together to unify these rules. For example, why the hell is save zone on PS3 and 360 different for 720p?!

Great letter, thanks for bringing this issue to the public. The certification process used by all three console manufacturers is a bureaucratic nightmare that makes the Soviet Union seem like a well-oiled machinery in comparison.

I personally work with localization, and the translation requirements are just as insane as some of the technical ones. For instance, how do you operate a touch screen? By *touching* it, right? No, not if you’re Swedish and you play on the PS Vita. Then you *caress* it. If you’re Danish and want to play a Quick Game (this is an MS term) on Xbox LIVE, you have to look for the equivalent of “Quick Profile”. It makes no sense, but call it something else and MS will fail the game. Or how about this? Sony and MS couldn’t agree on how to translate the word “button” into Dutch. There’s at least one dev studio out there right now (that I know of) that is busy getting rid of every instance of that word to avoid having to create separate strings for the two different platforms. They can’t just leave it out of the Dutch translations, because that would mean that Dutch players lose “vital” information – the better option is, obviously, that EVERYONE loses out on this information.

Absolute madness, I tell you.

I don’t know why I exactly feel compelled to talk about this, but I believe it was a couple of years ago the entire steam platform went through a relatively large refresh and added the community section and the notion of groups. As I recall it did change the UI.

And since then they have added cloud-saves, the steam workshop, and screenshot/video upload features. Also there’s the promising sounding Greenlight feature coming up soon as well. Again I think Valve is fairly aggressive at adding features and updating steam.

While the cert process could certainly be a lot more streamlined, getting rid of it entirely seems like a very bad idea.

I was able to convince them to skip that requirement because I implemented a robust save of my own.

*Community page

The Steam client is updated regularly, but I have never seen a gui change in any of the sections I use and no features I have ever used have been added.

The detail view is horrible. Half the detail section is filled the the same default “game Info” , which is just in fast a paragraph of boiler plate and a link to their coming page.

And then you have the news, that mentioned how there was a sale 5 years ago, and before that an update.

I’ve met Nintendo in person, and he is a strange fellow indeed. Nintendo doesn’t know what XBOX Live is, or what is cross-game chat. Nintendo is fiercely protective of their platform’s image, to the extent that they shun microtransaction-based games because they fear developers might trick users somehow into spending real-world currencies. Nintendo is also paranoid in the extreme with regards to what third-party titles do to their console, to the extent that they can’t conceive of an online game where each patch does not go through an extended lot-check process. Yup – lot-check – Nintendo still uses a manufacturing metaphor when thinking about software! To me, this explains a great deal.

Anyways – I’ve only played the PC version of Braid, but I’m pretty sure the XBLA version doesn’t have an animated icon saving screen. Were you able to get Braid certified without passing all of the reqs?

This is true in theory but I don’t think it’s the reason Sony or Microsoft do this. Maybe Nintendo, though — they seem to have weird mental problems about independent developers.

I think it depends on what the specific complaints are. In general, certification does not cover quality of controls, so it’s hard to see how certification would fix that kind of issue.

But my impression about iOS controls is that they are generally fine if the game is designed to be a touchscreen game, and they generally are terrible if it’s a joypad game ported to a touchscreen (2D platformers, etc). But that isn’t really a cert kind of issue, it is more just a game design kind of issue.

I’m sorry, but all your attempted points are wrong!

To the first point, nobody is going to complain, because we are just talking about a robust write operation. It is the same as what people would currently be doing, except it is better in every way. Nobody is going to complain. If someone wants to do something different, they just can. You aren’t taking the rest of the API away from them.

About screen saftey, it is the same issue of a “lowest common denominator” either way. The question is just how it’s implemented. In one case, every developer has to do all the work separately; in the other case, the platform holder does something once. People aren’t going to return consoles for title-safe region problems, especially if you run the screen calibration at startup the first time so that you know it’s right!

This point (“… unless you forced all titles to render at the same resolutions”) is just nonsense. I almost don’t know what to say. It is straightforward to scale an arbitrary resolution down by an arbitrary amount. I don’t know why you think this is an issue, but I surmise you are not a programmer or technical person.

To the last point, this letter is not about XBLA. It is about all certification processes for all sizes of games across all the major consoles. And no, XBLIG was not designed around me or anyone else who designs reasonably ambitious games written in C++ and who wants to sell them to a reasonably-sized audience.

In this case I am talking largely about autosaves, in which case you don’t know exactly when it’s saving anyway. What should happen for all games is that we establish baseline behavior where people don’t worry about whether their progress is saved, and just don’t care, because they trust that the game has done it.

If you actually go to a menu and pick a manual save, I have no problem with putting up a “Saving…” message until you know saving is complete. That is kind of a dinosaur way to do things, though, except in special cases.

He is talking about the certification process for games being rolled out on each system (Steam/iOS/xBox/PS3/Wii)… Apparently you missed the whole point of the article/letter.

But that’s not to say that we can’t improve on things another way. What about a system that queues the shutdown command, and executes it only after all pending writes have been completed and verified? As long as the isn’t a hardware power button and the only shutdown possible is performed via software, that would work well.

In any case, I think you raise some good points about wasting developer hours by forcing developers to take their focus away from core gameplay implementation and deal with these sorts of details. To the extent that we can extend the framework provided by the vendor to reduce and hopefully eliminate such issues, I say it should be done. Even if Microsoft or Sony don’t make these things a part of their API, as I would hope they would, the developer community could create libraries that extend the base API and accomplish these goals, so that developers don’t have to waste their time re-implementing the same requirements, and can focus on making the game itself.

I suspect a lot of studios probably do have some in house solutions that they at least share internally already, while others may license or sell this work to other developers. But I’d really like to see a community based effort to bring about a Free framework that does this stuff, and makes it available to all developers. Games will only get better as a result.

All this talk about Steam reminded me about Steam Greenlight, which, if all goes well, will lower the barriers to entry even further, as Steam has realized that their certification process, which is far easier than any console certification process, is still fundamentally flawed. Here’s to hoping that consoles change before the companies die out completely, but in the meantime I’ll likely just stick to PC releases for my games.

Being able to set throttles for Steam is a thing I’d like to have too, but to your other points I am not sure what you mean by bad GUI (I’m not a graphics artist or designer), and general oldness. Steam is updated regularly and has had large refreshes in interface, functionality and back-end stuff.

They patch and update their service seemingly aggressively. So, not to be too accusatory, I am unsure what you are talking about?

Self publishing directly to players? Craziness! Steam with its low barrier to entry was disruptive enough – now how will console manufacturers make their millions*

You rebel you.

*I say millions instead of billions or some other “illions” because I assume profit is quite low when the cost of maintaining such a bloated enterprise takes a serious chunk out of any incoming revenue. I’ve done work for Microsoft. They don’t exactly think “lean”.

Because of huge issues like this that Valve could patch in a week if they really cared, I always considered them as people who cared not a whit about user experience.

Anyone know what Mr. Blow is talking about?

Devil’s advocate, because there don’t seem to be any …

Others will claim the ‘Microsoft Save API’ is too restrictive and doesn’t allow them them to implement save data in the style best suited to their game design. This is less creative freedom for developers.

Title screen-safety is a strange thing to complain about. First parties are trying their best to guarantee YOUR game is playable on all of their customers’ televisions. You don’t have to deal with the customer service issues or issuing refunds – they do, and CS is not cheap. Having a ‘lowest common denominator’ makes it fiscally possible to have a modern console ecosystem as it exists today; no one’s gonna buy the shit if the people selling it can’t guarantee it’ll work 100% of the time.

An OS-based screen calibration tool simply wouldn’t work unless you forced all titles to render at the same (or a group of ‘sanctioned’) resolution(s); considering games like Microsoft’s own Halo franchise do some crazy rendering tricks to achieve both excellent looks and high framerate, a one-size-fits-all solution is more creatively restrictive than your suggestion.

As much as they’d like to believe, XBLA is not designed with agile indies in mind, and all the square-peg-in-the-round-hole effort in the world will not change that. If you’re looking for that level of flexibility while still remaining on console, XBLIG was designed around you.

But it’s not like that, because they still already test what happens when you yank a memory card out while saving / loading, or turn off the console! The requirement is that the game does not crash when dealing with the corrupted save, it just discards it and reports an error. On the other hand, if it’s in the API, I claim they don’t need to test it at all once they know it works.

The second case seems to be similar. Instead of them testing if everything looks more or less ok with different calibration configurations, they ask the dev to target the lowest common denominator so Microsoft has to check each screen only once.

As you say, this approach is a dead end.

As always I really appreciate your thoughts Jon, and Braid remains in my top 5 games of all time – although I’m not sure you’re ever going to beat Kingdom Hearts; nostalgia is a powerful thing ;D

These certification rules apply to *all* games, not just independent games. So it has nothing to do with game size or importance to the console.

It’s just that they feel like they can make the developers do this work and so they don’t have to do it themselves. What they don’t realize (or maybe don’t care about) is the immense cost, of time, money and game quality, incurred by making all developers do all this stuff all the time.

To be perfectly honest I have no idea why the system callibration option does not actually exist, if independent developers are able to set aside the time to implement it in their games I don’t see why these large corporations can’t get a couple of people working on it and implement it in an OS patch.

Anyway, thanks again for posting this.

Best Regards,
-RegretZero, Pixxelcraft Website Manager