AMD Doesn't Make $100 for Every Console Sold

[I]f these talks were to result in Microsoft owning AMD, it would save the firm a lot of money. Microsoft, as well as Sony, pay AMD in the range of $100 on each console produced. On top of the savings Microsoft would be making on each console, sales of PlayStation 4 would also directly benefit Microsoft. [nonsense]

Today we are going to unpack what's on display at VG247. This is why you need to pay more than minimum wage for writers to fill your news feeds if you want to get something that provides positive value for readers rather than misleading them with speculation that doesn't stand up to any scrutiny.

These numbers, which we'll go on to show are created by people misinterpreting what is just speculation, smell completely wrong. Any editor who has an understanding of chip sales should have flagged this. $100 is not the important number to look at. Sony has not added ($100 * 22.3m) $2.23bn in profit to AMD's financials in the last 18 months due to per-console fees. Note that AMD's market cap is currently $1.82bn.

ARM, who publicise their offers for IP licensing so you can get a chip design and ask anyone to make it for you (or customise it), charge in the region of 1-2% per wafer. After Microsoft got burned by nVidia on the original xbox, they're highly incentivised to pay a bit up front for R&D and minimise their licensing fees. They want to own the semi-custom design they paid to get developed. Sony have in-house expertise and are unlikely to be taken for a ride by AMD when they've done so many deals with people who generally only make this 1-2% profit on supplying the IP/chip designs.

So where did the number come from?

AMD makes more profit than Sony on the PS4 [...] This sees AMD making quite a healthy profit per console. [nope]

The Inquirer failed to understand a report from the IHS Teardown Analysis service. In this report, IHS estimated that a large (~350mm²) chip like this is likely to cost Sony or MS about $100. This is assuming a good deal from TSMC or GlobalFoundries and very little profit for anyone. So a cost of $100 for a part gets written down as a profit of $100 on that part. This is very stupid but also not the end of the journey. Note that I didn't say AMD was making the SoC.

AMD used to be able to make chips but they spun off that semiconductor fabrication plant operation, following the foundry model. They have completely divested themselves of GlobalFoundries so it doesn't matter who makes the chips, AMD doesn't profit from the actual production being profitable. I'm not sure if some of this analysis is done by people who think that GF is still part of AMD and something MS would be buying.

When MS or Sony pays AMD $100 for a SoC for their console (assuming they can't buy directly from a foundry - something ARM contracts often promote as the normal way of doing business) then AMD need to get a foundry to make the chip and provide it to the console manufacturer. This is not a highly profitable venture as that is not a small chip to make. The IHS estimate of $100 doesn't seem completely off but is an estimate that assumes AMD make virtually no profit, the actual number (or how close this guess is) isn't actually all that important. That 1-2% per-wafer fee sounds not unreasonable. That's cents to a couple of dollars. Not $100.

This would be the profits that MS could save on production of the XBOne and could extract from Sony for production of the PS4. And this is assuming Sony don't have the rights to produce the PS4 SoC paid for in full via their R&D budget and so have to buy SoCs via AMD rather than buying them directly. It's pennies, not almost a third of the current price of the console.

But this isn't to say Microsoft are definitely not interested in AMD. The failure to crush nVidia with their new generation of GPUs (despite having a 6-9 month head-start on nVidia with stacked DRAM) and pinning their hopes at clawing back a competitive x86 part that uses more than 3 Watts being entirely reliant on a 40% IPC increase from Zen means AMD are not in a good space. They're ripe for acquisition. MS have lots of money, especially as AMD are rather small compared to people with similar IP portfolios like nVidia (who Intel is still paying to not develop x86/claim they could if they wanted to - something AMD do have). MS also have that hardware division that does Surface tablets and the xbox consoles. Although they use the best Intel has to offer for an x86 tablet, they could probably move to AMD without ruining the devices. There is some synergy there. It's just not a $100 per console obvious move.

Legalised theft

I've used the phrase "legalised theft" a few times talking about Microsoft's policy of currency conversion that moved their digital stores from MSP (Microsoft Points) to local currency balances/credit.

Any Microsoft Points that you had remaining in your Microsoft account have been retired, and we’ve added to your Microsoft account an amount in your local currency equal to or greater than the Xbox stores’ value of your Microsoft Points. This value we added is promotional and will remain in your Microsoft account until 1 June 2015. However, the currency you purchase and add to your account will not expire.

You cannot, for a single moment, believe that last line. Adding credit to your current MS account will provide absolutely no guarantee that the currency you purchase and add to your account will not expire. We know this because the entire transition to local currency described before this explains how they have stripped the non-expiring nature of the MSPs you purchased before. This is theft, but done in a way as to be technically legal.

You should never buy credit for an MS store (say, in a sale or other offer) that you do not intend to immediately use because their statement on credit expiry is known false. You may use this example of dishonesty as a reason to blacklist MS and totally avoid their digital stores, I know I will be less likely to spend money in their stores after this behaviour.

But some people seem to think I'm grossly exaggerating or lying when I make the claim this is theft, even if qualified to be a spin on theft that has been crafted to be technically legal. Let's look at exactly what happened and could happen again to see if we think this looks like theft.

1. Someone offers for sale currency for their store (advertised as not expiring, can be used forever) and a supply of goods to be purchased using this store credit. Items have a real value of the real money required to be converted into this store credit to make that purchase.
2. People use this store and buy credits in advance of needing them (the store actually makes it impossibly to by exactly the credit you need, you have to buy it in blocks), converting their real money into the equivalent value of store credit.
3. The store owner, after accumulating significant real world money that is not spent but sitting as store credit that cannot be converted back to real money (thus avoiding the legal requirements applied to actual banks who provide an internal balance for people paying in real money), decides to end this situation.
4. The store owner removes all items from sale using the store currency, making the value (in purchasable goods) of outstanding credit to be zero. But the credit has been guaranteed to not expire so cannot be removed. It is simply made worthless by removal of places to spend it. This is the opening for theft.
5. A new store is created by the original store owner that sells the same goods and for the same real world prices, only there is a new store currency. This is where the people realise they've been stolen from.
6. The store owner offers to convert the old, worthless currency to the new currency for use in the new store but the conversion process will create new currency credit that does expire. This is the scam that softens the theft, allowing users to sign over the non-expiring nature of their store credit for access to the credit in the new store (credit they previously thought they could spend on goods in the store forever).
7. The store owner assures everyone that the new store currency you can buy will result in credit that never expires. Somehow the store owner doesn't think the people are on to their scam.

Commercial mods

Wouldn't it be great if there was some way for mod teams who make expansive mods to get paid for their significant work, maybe on Steam. Shame they can't and PC gaming is dead according to the backlash to the backlash about Skyrim mods.

The actual implementation chosen for the paid mod marketplace may have been one of those exceptional times when every group loses. It's not easy to build systems that fail all.

Game players now have a search issue that has financial costs rather than just being a "grab and see" process. You now have to possibly buy into some mods to get the game you want; this is a massive barrier that it's easy to understate. Especially due to "value" (games are (for the content) cheap, DLC is generally far less cheap, mods appear to be coming in at DLC or worse value levels) concerns.

Valve, who have spent forever locking down their system to avoid having too much DMCA takedown and other copyright infringement work, are now getting financial gain from an unmoderated store. Their earlier UGC stuff was all carefully crafted to avoid this issue (at the cost of making each piece expensive to put up) and a small curated store meant every item could be popular.

The publisher now has competition for the commercial DLC they put out in the form of commercial mods. This was always what "no you can't sell it" stipulations were about - it means you could sell your expansion pack. Yes, their cut doesn't have the costs associate with Valve's work (although they probably need those lawyers to be primed for the copyright infringement claims they'll be co-defendant on if Valve don't take down content fast enough and hand over any money). Their game can also lose value by perception (game players now see the PC release as worth less due to increased "cost" to find the mods they want to play with).

And modders now have the same commercial "app store" myth to push them to try and make money when 99% of people just make something no one will play. The few success stories push everyone to try and charge and find out that no one is buying their small mod (that can't compete with the value of Skyrim being $5 for the base game, which took 100+ people several years to make - your solo mod would take you literally hundreds of years to get the same work done so you're always going to be offering something like 1% of the "value" of the $5 game - so a fair price is 5 cents, or nothing when rounding). A few individual modders will get to do it for a living but as a group, modders now have a more volatile group of consumers who are expected to provide more search labour to avoid being scammed. Those who will make money this way, if they did work that wasn't incremental, could be making those mods into commercial indie releases. Those who do incremental work may well be better served working as a contractor or inside the studio system making those assets for the next game. The ideas of open collaboration and remix culture are destroyed by commercial concerns while the extrinsic motivators suppress the intrinsic ones.

We Don't Have the Money, But We've Got a Plan

The last month has involved quite a lot of getting lost in some deep PC games. While the PS4 facilitated my gaming highlights last year, the games I've been dropping a lot of time into recently have all been on the PC.

An interesting thing about Pillars of Eternity, which is looking like one of the best CRPGs ever to be released: it was built on a budget that didn't facilitate anything cinematic. Obsidian didn't have the funds to make this a big 3D world, to provide glorious cut-scenes that compete with Blizzard, or to even provide a fully-voiced script. And those limitations have only made this game better. The game was built to hark back to the original Infinity Engine CRPGs which many of the developers had previously shepherded.

We now have the tools to create those worlds and script it together without spending as big as the industry can spend. A lot of that is the AAA tier of games has moved up in price, with the development costs alone going beyond $100 million. While Baldur's Gate and its ilk were almost certainly cheaper than $10m each, wages are much much higher compared to the end of the '90s so even a few million wouldn't bring that sort of team back and Obsidian didn't have $10m. There is a lot more rapid prototyping and efficient development possible with a modern tool-chain and knowing what you're going to build before you build it. Obsidian could map out areas using a fast 3D package, render those to both 3D data for real-time effects & occlusion for vision cones and 2D maps, and let the artists draw over the 2D maps to add in the detail without building it all in the 3D package. The game looks great, showing off the same painted backdrop love that originally wowed us playing those games in the '90s.

The script, because it didn't need to be 100% nailed down months before release so it could be given to the voice talent and because it didn't need to be paid for in VO costs, could reflect what made those old games great. This is an RPG that, if it is the child of any other medium, was descended from books. It is something you read, painting pictures with words that bring the rendered scenery to life and allow the story to dive into the areas too expensive to spend a few person-years of work rendering in a cinematic style for a short aside. The lack of any cinematic intent doesn't just provide an experience that reminds us of what was lost when CRPGs moved to making the dialogue into movies, it also allows for more options. We've already mentioned dialogue but this also extends to character skill checks. As it's nothing more than some text and a painted illustration of the challenge, the game can provide multiple paths to conquer an obstacle with real consequences depending on the skills of your party and what you attempt from the options provided. A cinematic RPG could do this, but not without heavily investing in building custom animations or pre-rendered videos for each option.

Pillars of Eternity is as good as it is because they had a plan, knew how to get to where they wanted to go, and knew that their aims fitted the constrained budget of crowd-funding. Without having to invent the typewriter or create a new style of prose, they had more time to write the book.

Another game released this month on PC is the game that EA failed to make when they mistakenly injected always-on DRM into SimCity in 2013. A small team who had previously tested their abilities building the Cities in Motion series of traffic building city-management games decided that they could take their existing knowledge and build the game that the large team at Maxis couldn't. Colossal Order is 13 people. They built on Unity (as did Obsidian) to give them a basic rendering and rapid development platform which can easily support several platforms and did their best to build the city simulator that was obvious from the pre-release enthusiasm for SimCity 2013.

And they did it. Thirteen people and a plan, with some modest funding from their publisher Paradox (who also did the publishing work for Obsidian, although not the funding of the project - these games are twins in many ways; not least that they, via backer pricing or cheap offers, both cost only £13 to get them in the first week of release), managed to build the game with the depth missing from the Maxis offering. No draconian DRM that mandates permanent server connectivity, a more deeply realised simulation of the virtual people and vehicles, and the trust that the community would enjoy finding the edges. This game has thousands of mods and props and tweaks in the official workshop, which is highlighted on the main menu of the game as a great place to check out. They already managed to scale the simulation up to vastly larger spaces than Maxis did, but mods arrived within hours that remove the limits they has put in, if your computer is up to running the simulation or you'd rather play at 10fps and build big.

It's a triumph and a massive success for Paradox and Colossal Order. Because small teams don't need to sell 10 million to break even. Reaching half a million sales so quickly triggered celebrations; quarter of a million in the first day was a record for the publisher. And we all get a game that is launched at a budget price to prevent it being a play-though for those rich enough to throw $60 at something they're not certain about while the rest of us wait for it to get somewhat closer to the price of other entertainment media.

Also Available on Tumblr

A full blog post is still coming this month (we're back to that schedule after missing three months in the latter part of last year); this is just a quick administrative post.

Due to the wishy washy way in which Google have talked about their interest in retaining Blogger as an open platform for personal expression, I've decided to start crossposting content to Tumblr. If you use that platform and would rather follow this blog there, you can do so here. On a related note, I've been crossposting the articles related to computer and video games to GiantBomb for a while, which has elicited comments both great and hostile.

Edit [July 2015]: Due to a database error on GB, the archive of blog posts and their attached comments now seem to be missing.

The "Death of Classic Reviews"

Man, the "death" of classic reviews. We previously had most launch-window reviews of (PC) games as software reviews because most games had glaring UX issues, problems in their mechanics, or simply a lot of bugs that led to weirdness. It was important to talk about how the game controlled, how it provided feedback, how it developed the new ideas and taught you how to play the game, and, only then, say if you enjoyed it. That's where previews (where I really enjoyed writing around the turn of the millennium era when I was doing that stuff seriously) could be really interesting - a lot of that stuff was exposed in a preview build so you could lay it out. This was the era of Dave Perry explaining game design as building the best thing you could around one totally new idea, the hook, that would sell your game.

There was no single, established way to do everything. FPS controls were all slightly different and how you interacted was being developed. I remember the amazing point in our LAN games of Quake where we all migrated to the mouse, because we'd played Doom and Duke 3D with the keyboard. It wasn't obvious that you had to use the mouse, we still didn't know how sensitivity and so on should be set to best optimise our frictionless input onto our avatars. What systems were involved and how you interacted was often bespoke. It's not an action adventure title with this sort of crafting and thumbstick reaction curves lifted from a previous game (pick if you want full-lock to continue to accelerate or be flat, set 80% lock to be flat, do or don't smooth) that you already know intimately. A game was something that you wanted to know about before you tried it because it could be anything, often objectively worse that other potential choices. Reviewers could tell you about the sharp edges and defects that might make it worth sitting this game out.

But now most games are polished in a way that means they are functional. They may not have been refined iteratively in the developer sense of "polished" but the importing of best practices/lifting from genre-defining titles means they don't actually have to. X came from Y, where they got it right and everyone played Y or a derivative so understands the systems involved and the interactions. Everyone is trained in "how things work", you don't need to teach the user how to shoot. We are now trapped in this place of many local maxima. So what's the point of a review when it's not a 50/50 guess if the basic functionality is even there? When 90%+ of games that get a PR push will also be mechanically solid (and people seriously suggest those that aren't shouldn't even be purchasable) and not crashy (crashes being the kiss of death you'll hear about as soon as a game is released) then the only thing left is if the reviewer liked the game.

But reviewers are still operating under the mentality we are taste-makers. Because 15 years ago you could play through a game and explore the "polish" of it and give a good estimation about the objective quality of the software. So you could predict if most people would tolerate a game. But that's not true today, and reviewers don't realise we need to focus our text on "I enjoyed" rather than "this is good/bad". Justifying your enjoyment with objective markers of quality rather than your enjoyment of the text of the game will lead to ruin.

DriveClub gets panned because it doesn't follow the conventional wisdom on controls (it is neither a Forza clone, nor a NfS clone; it's not even a Kart clone) but people who are experienced in the genre have found that it offers its own path that does have significant depth and interesting consequences. Reviewers are painted as lazy when they use shorthands that turn out to be oversimplifications (eg saying AIs stick to a racing line as if on rails). We fail to express our "I don't like this" and go after stuff that is known to be bad design that we may invent to justify our dislike. It's a minefield of ex post facto justification.

Y'know how lots of critics hate so many mainstream films? (I'd say Transformers but those do actually seem like pretty bad movies, even if lots of people enjoy them - but maybe even that helps make my point - classic example might be Pirates of the Caribbean.) Games are there now. Our reviews are filled with a range from hatred to love of games that are technically solid and so entirely reviewed on how the reviewer felt about the story and interactive elements. We're reviewing the game as a piece of media, not a software package (if it becomes a software review the it automatically fails, as it's a crash-prone buggy PoS and none of those need to be given the time of day unless the reviewer really like the game and so ignores those flaws).

Only right now a lot of reviewers are trying to establish our "discerning taste" cred by berating anything not to our palette as "mainstream rubbish, no one should like this, it's just bad". Rather than just saying it wasn't to our tastes. So we get those reviews saying DriveClub doesn't have an engaging driving model or AI, which many fans of driving games (who jumped into that title and waded through the online fiasco) will contest. A game that's compared favourably to Gears and Resi4 is called derivative tosh that's a decade late rather than another solid entry in that genre because reviewers still remember a time when devs were learning how to make games and so sanding away at rough edges.

The Order 1886 is panned for being nothing more than a mix of cinematic and 3rd person shooter in a steampunk world. The high quality of visuals is used as an excuse to pan the title for not being innovative enough. Again, "I don't like this" is warped into "this is objectively bad and justifies my dislike". But people playing in a really nice looking world they've not been to before are transported to a new place where they play a totally solid 3rd person shooter. That's exactly what a lot of people want. And reviewers said it was bad, not just not to the reviewer's tastes, but bad and not worth anyone's time. And bad has always previously meant that the software was either defective as software, often in the UX realm, or the game design failed. But that's not the case.

And now we're here. This is the crossroads we're at. The "death of classic reviews":

"Initially I was caught off guard by the doubt cast by various critics out to smear the game. They ended up doing me a favor, in that I now have a great list of online publications which I know to avoid spending any future time reading." [source]

Flat out sentiment of disbelief between the experienced product and the expectations of how reviews of games software work. "various critics out to smear the game".

Is The Order 1886 any good? No idea, not played it yet and that £45 price tag means there's no chance I'll find out this month. Ask me when it's £20 or less. These early-console-generation prices being jacked up in the UK should end soon and we'll get £30 game releases (if history repeats). Our healthy software market seems to thrive on being cheaper than the $60 US market. But the reviews are 100% useless to me; some friends who buy release games make it sound pretty good. The things it is compared to in reviews, that may not even endorse it, are also a positive sign.

This does seem to mean reviews are basically useless at this point for actually providing consumer advice. And games are too expensive compared to movies to just blindly consume them all (outside of Steam sales - welcome to the pile of shame phenomenon).

It has never been easier to watch someone play a game, from friends or randoms streaming their play to more organised stuff like professional Let's Players. It's never been easier for us to listen to our friends as social media makes everyone a broadcaster of their daily thoughts. It's never been harder for a written review to actually be useful around launch. And attempts to lean into the taste-maker role will only cause this visceral backlash. Maybe it's time to let launch reviews be the domain of YouTube and leave the written work for a month after release when critics can dig deep into the role of deep critical analysis, for the games that generate interesting analysis of their text or mechanics.