Charlie's Diary

I find this interesting because, to me, this feels like the second time the natural price floor has been broken. In my (fallible) memory, until about 5-8 years ago the computer I wanted for home use, one or two steps down from the high end, was always out of my league at around $5000. I'm typing this on my (expensive, niche) Mac and it's nowhere near that. I think it was around 5 years ago that non-toy PCs really dropped under $1000, later under $500, and that people weren't afraid to buy them for serious use.

I'll be delighted if you're right and we're really in for another price drop of a factor 5-10. Think of all the fun things we can do if programmable general-purpose computers, running a commodity hardware platform, drop below $100. (Yes, I know embedded systems are below that today. They often cannot be re-programmed or run a non-standard CPU, which means I cannot just slap standard software on them, especially codecs.)

Seriously cheap PCs will also mean the end of the shrink-wrapped software industry - who will pay for an extra $600 (Windows OS + Office) if the whole computer is $100 and comes with Linux/xxxBSD and OpenOffice/KOffice?

Posted by: Hildo | November 25, 2007 2:16 PM

Hildo: Seriously cheap PCs will also mean the end of the shrink-wrapped software industry - who will pay for an extra $600 (Windows OS + Office) if the whole computer is $100 and comes with Linux/xxxBSD and OpenOffice/KOffice?

Now consider the combined market cap of the shrinkwrap software industry and the PC business, and contemplate what deflating it by, oh, 80-90% will do to the western economies.

Posted by: Charlie Stross | November 25, 2007 2:39 PM

I have to admit, I tend to find myself wondering how much the licence for the software is costing me when I buy a new PC these days. There was a brief, hopeful period a few years ago, back when the MS anti-competition business was going on, where I saw the possibility of a brave new future with the chance of being able to purchase a PC without necessarily having to accept whichever MS OS was being preloaded onto it as the OEM standard today. Alas, that fitful vision flickered and wavered and disappeared, which means my current situation is that I'm not going to be even thinking seriously about purchasing a new PC (despite the fact that my current one is reaching the end of its usable life - no working USB ports, floppy drive not connected, no DVD drive etc) until Windows Vista reaches at least service pack 2. As far as I'm concerned, the notion of paying Microsoft an additional $700 (AU) for the joys of beta-testing and being nagged to upgrade their software is just ridiculous.

Here in Australia, PCs are still largely in the range of $1000 to $5000 - for $1000 you're generally getting a "bitsa" box, put together by chip wranglers somewhere in the local area and with no manufacturer name on the box (so, something much like all of my previous boxen); for $5000 you're getting something with a "guaranteed" brand name (like Dell) or you're getting the latest and greatest in gaming machines with all the bells, whistles, chrome, and go-faster-stripes on the modem.

I already use Open Office for a word-processing package (have done so since I first ran into it as Star Office 5.0, from a CD on the front cover of a PC magazine) and wouldn't switch if I was paid to. (My greatest rave about OO is the size of their proprietary file formats - when a file of 50K words is 663KB in .doc format, but only 161KB as a .sxw file, it's impressive to someone who carried most of their files around on a floppy disk for years. In these days of multi-gigabyte personal storage which is carryable on a lanyard, it's not much of a selling point, but I still find myself impressed by it). I've tried my hand with Linux a couple of times, but not in the last 4 - 5 years. I've no doubt things have improved greatly in terms of user-friendliness for the less Unix-knowledgeable of us out there, which makes me more inclined to use it, rather than less.

Posted by: Meg Thornton | November 25, 2007 2:59 PM

In Intel's case, the word "recession" doesn't even start to describe what would happen. For the last 30 years, Intel has had a business model based on Moore's Law, and the inevitable outcome is the the corollary that every 18 months the cost of running the assembly line for a PC component doubles, as does the cost of the engineering required for a new component. So Intel (and other IC companies, to a much lesser extent) has been in a Red Queen's Race with itself, constantly having to pay off the investment in the previous generation by adding huge fixed costs to the production cost of all their components, while simultaneously doubling that investment for the next generation, which is the only tool they will have to prevent the commoditization that would destroy the advantage they have over their competitors. This cycle meant that Intel, in a very real financial sense, has financed the development of all the hyper-integrated technology that makes modern computing possible. Take away the market acceptance of that constant drive for higher capability, and Intel doesn't have a workable business model anymore. Being a large corporation, they won't notice that right away, and will probably do some extreme damage to themselves while they try to turn the battleship around.

The greatest effect of the silicon revolution has always been based on putting as much as possible of the functionality on as few as possible components, and making as many of them silicon chips as possible. The cheapest and most functional solutions are ones that can be built on a single chip, with as few external components as possible. We can do that right now with a commodity PC (although nobody has, as yet; I predict someone is going to make a bazillion dollars by doing it sometime soon).

The only real survival strategy I can think of for companies like Intel that depend on a competitive edge based on high R&D investment is to start diversifying into other application areas and other system architectures than PCs. There are other markets the bleeding edge junkies can go after and the long-term consequences of them should make any SF writer or reader salivate. The server market is one that the big silicon houses haven't really gone after, which is why the most cost-effective server solutions right now are based on cheap consumer components in huge fault-tolerant clusters. What happens when someone puts a server, cpu, ram, and flash storage*, all on one chip and sells it for USD 25 in lots of 10,000? Or when a complete internet node with webcam, mike, speaker, wireless connection, and a multi-farad supercapacitor and solar cell, capable of powering the chip indefinitely in low duty-cycle mode can be manufactured for USD 5?**

These sorts of markets haven't been a high priority for the industry up to know because they've been making their nut in the consumer PC market. When the bottom falls out of that, we may see the world change drastically as the industry moves into other areas in a big way to replace the losses.

* and maybe even power-conditioning circuitry, so the power efficiency of large bunches of the things goes toward maximum? And perhaps ventilation as well, in the form of electrostatic air movers?
** This is known in the trade as "smart dust", and Vernor Vinge has had a few things to say about what it means.

Posted by: Bruce Cohen (SpeakerToManagers) | November 25, 2007 3:17 PM

Charlie Stross @2 said:

Consider the combined market cap of the shrinkwrap software industry and the PC business, and contemplate what deflating it by, oh, 80-90% will do to the western economies.

Hmmm. The main thing I can see happening is things like Operating Systems, office suites, graphics manipulation packages, and similar all dropping in price to the point where they're matching prices with things like games. The market for PC games is pretty much "whatever the traffic can bear" already, and this means there's a range of games available at prices from $120 (AU) down to $20 (AU) - the cheaper ones are usually the ones which are being re-issued, which don't require the latest and greatest equipment to run, or which have been selling so slowly the retailers want to get the shelf space back so they can put another copy of "World of Warcraft" on it. The median price is about $80-ish - a bit higher if it's new or popular, a bit lower if it's older or past the first flush of popularity. PC gaming in itself is starting to face serious competition from the console market, from what I've seen locally - things like the PS2, the PS3, the XBox and the Wii are able to do all the things that a high-end gaming PC can do, and they cost less (the most expensive console at the moment is the PS3 - that's about $999, roughly equal with the "bitsa" PC I mention in #3). So do the games, by and large, and the games are what people buy the consoles for.

So, the shrinkwrap software industry as a whole will probably start resembling the game software industry a lot more - in order to be able to charge high prices, you have to provide something new, something interesting and something genuinely different to the majority of the stuff out there. It doesn't matter how many man-hours you put into your product, you're not going to be able to charge people more for it than they'd pay for the PC to run it (see the pricing of games for the hand-held consoles - Gameboy, DS, PSP etc - for further reference). Up at the corporate end of town, we'll probably see a lot of the bigger players starting to take major hits, because this puts the ball back in the customer's court - and many of those customers will be asking for things like better interaction between various tools they're using. As a result of this, I think we might see things like the various Microsoft products becoming much better at playing with others.

I also suspect that there'll be a greater appreciation for the work done by tech support staff. I foresee the realisation on the part of some of the larger firms that the support desk is where they're going to be making the money now. So, it'll probably be paid support, in that you'll have to pay for support calls, but in return, you'll get good support. Buying a warranty will actually mean something (rather than the present situation, where it just says you have money to spend and a willingness to be ignored/put on hold for long periods of time).

Posted by: Meg Thornton | November 25, 2007 3:24 PM

The Eee isn't quite the disposable computing resource I've been wanting — they'll have to shave a zero off the price tag for that...

The obvious question - how long ago was it that the average consumer mobile phone sold for £220? Because we've certainly had most of that order-of-magnitude drop.

Posted by: Andrew Gray | November 25, 2007 3:29 PM

Huge problem with the pointer, Charlie. Indium and other rare metals used in computers. The price of Indium current stands at over ten times what it did a mere five years ago, and this has badly affected the price of LCD screens. This is going to get worse. Much worse.

Also, as an interesting point, the GPU's companies have stopped doing something. They've stopped pricing graphics cards as $=£. The European cards (when you look at them minus VAT) have less than a 10% markup.

Anyway, £220 is far too much for something with the limits of the Eee - the battery life is simply far too low and the only way to get a spare battery is to buy another Eee.

I also have to disagree that offloading the work to the server makes sense in the home. In business, perhaps, but the nature of home connections (UK actual broadband connection speeds and reliability have fallen in the last two years, and will continue to fall - especially given the massive costs of the 21CN network to ISP's) means it's not even remotely viable outside the LAN.

Meg? Er... the PS2, Xbox 1 and Wii don't have as much computing power as PC's of some years ago, and the top end PC's now allready outstrip the gaming performance of the PS3 and 360.

Posted by: Andrew Crystall | November 25, 2007 3:38 PM

Not all components of a PC obey Moore's law. The case and keyboard are already at commodity levels. The LCD screens follow a much slower price/performance reduction trajectory - although OLEDs may force a big drop in the next few years.

The biggest drive to commoditization of prices is that the CPU has topped out in clock speed. Moore's law is being used to add more cores per CPU chip, but applications cannot yet make this extra power work for them - so even CPU hogs can only use a fraction of the available power at the moment. As a result, there is little reason to buy a premium machine as the performance gains are not perceivable.

I think Intel understood the issues quite clearly over a decade ago. They encouraged and invested in applications that were CPU intensive and Microsft obliged by making Windows ever more bloated. [As an aside, I once installed Windows 95 on an ancient HP 486, stripped out every piece of crap I could to slim down the OS and ended up with machine that booted up in a few seconds - yet still ran basic applications normally].

Note that Microsoft is offering a Chinese version of Windows at $3 a copy, to counter the Linux and pirated Windows OSs there.

So all the signs are there that basic PCs should become like simple calculators - you give them away in cereal boxes.

BTW - I really like the extremely low energy consumption of the OLPC - that makes it quite feasible to run off a small solar panel. Removing the weighty batteries from a PC would be the largest contribution to weight reduction and form factor changes.

Posted by: Alex Tolley | November 25, 2007 4:08 PM

Andrew: indium, LCDs, yes. But there are OLED and IMOD and epaper waiting in the wings, all with different input materials and different performance parameters. Worst case, if indium turns into a huge bottleneck for LCDs, it'll make it cost-effective to spend the development bucks to bring OLED up to speed as a replacement technology. And so on.

Now, batteries are indeed a bottleneck technology -- they don't follow Moore's Law at all. But again, look at epaper. Today it suffers from rubbish latency and a slight lack of colour (in any escaped-from-the-lab form, anyway), but if they can speed it up by an order of magnitude and get colour working, then we'll have something that will simultaneously make the LCDs obsolete and reduce the power drain of one of the two most juice-hungry components of a PC by more than an order of magnitude.

Broadband connections ... yes: but I'm not convinced that's a technological bottleneck rather than a f*cked-up-industry bottleneck. I suspect it's also hugely aggravated by the fact that we haven't gotten the idiot film and music cartels to switch from a 19th century copyright model onto a compulsory licensing scheme. Once compulsory licensing is in place (with a blanket exemption for downloading content, because users are paying for a license as part of their ISP fees) we can begin to clear up a lot of the p2p traffic that's clogging our networks and replace it with a sensible content-on-demand system with local caching. (I hope.)

The one constant I can predict and be certain of is that the future will not resemble the past. (There may be bits of the past embedded in it, but the future will invariably be more complicated ...)

Posted by: Charlie Stross | November 25, 2007 4:22 PM

Speaking parenthetically, there are three things that could improve the Eee experience hugely for me (apart from dropping Xandros and switching to Ubuntu):

1. More FLASH memory. 4Gb is marginal. (But they're promising that for later in 2008, anyway.)

2. Built-in 3G broadband. (Again: promised for 2008.)

3. A non-backlit purely reflective display using something like epaper or IMOD. Doesn't have to be fast enough to play games or video -- just for email/word processing/web -- but if you can cut the power drain from that display to zip, then you can go from a 3.5 hour battery life to about a 6-7 hour life using the same cells. Going to 1024x600 resolution would be great, too (and cheap-ish 7" LCD display units in that resolution are available -- for example, Kohjinsha and some of the UMPC vendors use them).

Posted by: Charlie Stross | November 25, 2007 4:27 PM

You think the Asus Eee is cheap and small? Check out these mini pc's, one of them under 100 u.s.--

http://www.extremetech.com/article2/0,1697,2194862,00.asp

http://www.linuxdevices.com/news/NS6828123924.html

Posted by: coltrane | November 25, 2007 4:27 PM

Charlie - I think there's actually an OLED product finally due out in Japan soon. Sure, there are other techs but in the next two to three years it's going to push the price of the mini-screens up nastily.

As for broadband, yes. It is an industry problem, but one which isn't going to get sorted in the UK in the next few decades unless something radical happens because of BT and an utterly, utterly incompetent and impotent industry regulator.

"sensible content-on-demand system with local caching"

So something like...oh...NNTP? (I hate NNTP specifically, but something done even slightly differently..)

Posted by: Andrew Crystall | November 25, 2007 4:28 PM

Charlie @2: Now consider the combined market cap of the shrinkwrap software industry and the PC business, and contemplate what deflating it by, oh, 80-90% will do to the western economies.

I've seen various estimates that 60-70% of all software is custom-made, not shrink-wrap. In my workplace (an investment bank) our IT costs are over $3B, and a small fraction of that is spent on shrink-wrapped or ready-made "enterprise" software. We have over 1500 software developers in NY alone, writing code that only our firm runs.

So I don't think the deflation of software houses will be a big deal. The importance of in-house software, of customization, and actual customer interaction, will increase. And that's long overdue.

Posted by: Hildo | November 25, 2007 4:31 PM

Another SF ref for ultracheap computing would be Bruce Sterling's "spun straw" stuff in "Distraction" -- fueled in part by a Chinese release of virtually all intellectual property they can get a hold of to the public.

But the computing universe has gone through interesting cycles of centralized-decentralized (google for "Dogbert's First Rule of Consulting"): The web almost brought back the mainframe, and it looked like we might just carry ID keyfobs to public thin terminals to do most of our play^h^h^h^hwork. Malware has made that a little less attractive, but it's still an option.

We're still a long way from goggles-and-rings for user interface, that stuff is still pricey. But phone "slates" like Crackberry and rumored Nintendophones bring this convergence a little closer.

Posted by: joelfinkle | November 25, 2007 4:32 PM

Hildo @13: dead right. But the shrinkwrap market is the chunk of the software iceberg that's visible above the industry waterline.

Posted by: Charlie Stross | November 25, 2007 4:43 PM

I view the ultimate problem at this stage as being power — electrical power — and not component costs. The "killer micronotebook" must be able to handle a twelve-hour day without getting near a power outlet: Cab to airport to commuter-type aircraft to airport to cab to meeting to cab to hotel. (Or, for the fen among us, a full day in the dealer's room, art show, four or five panels, a smoke-filled bar with one's editor/agent/favorite author/favorite Klingon, and a couple of hours in the con suite before getting back to one's room.)

That's a problem of chemistry and basic physics, not of engineering. Almost all of the advances we've seen in the last 25 years have resulted from engineering solutions to reasonably well-understood chemistry/physics/materials problems (at the basic science level, anyway). Unfortunately, what we don't understand at the basic science level about energy storage/production is running right at the edge of our current engineering prowess for energy storage/production. That's one of the reasons that the Eee is so power-handicapped: Its batteries can't get better and lighter with current knowledge... just bigger, and being drawn down less.

The "stuff it full of AA/AAA batteries and let people use rechargeables if they want" solution isn't any better, as the power loss in the connections wastes almost all of the purported extended life! And the less said about a "multi-farad supercapacitor" the better... because that's bounded by some pretty basic physics.

So, for at least the foreseeable future, I don't see things getting really much better than the Eee, and it isn't quite good enough for me to consider switching. Of course, all it will take is a fundamental breakthrough of some kind to change "foreseeable" from "five to ten years" into "three months." The problem with fundamental breakthroughs is that they're quite unpredictable (one of my major annoyances with Civilization-type tech trees in games).

Posted by: CEP | November 25, 2007 4:45 PM

CEP: We've got an energy bottleneck, true. But look at it from the other end: we have a bunch of components with various power drains and if we optimize the hell out of them we can draw less power. The core components of a current laptop are: fixed storage (hard disk or FLASH disk), CPU, and display. The fixed storage we can work on by switching to FLASH or using physically smaller rotating media. The CPU ... do we really need much more than a 624MHz ARM9 core for toting around in an ultramobile notebook? Those things will run off milliwatts. Even an Intel core-architecture CPU doesn't have to be a stupendous power hog if you underclock it and optimize for power efficiency.

The problem we're left with is the backlit screen. Staring into a light bulb is something we've come to associate with laptops, but it ain't necessary; back in the days of the early 1990s when we still used B&W screens, I had an HP Omnibook 300. It had a non-backlit black and white LCD -- and a battery life of up to 10 hours. Those backlights suck juice, but they were really only required because colour TFT screens are lousy reflectors. Switch to a reflective rather than a transmissive display technology, and I'd expect the power consumption of the display unit to drop like a stone. Switch to one that only draws power on refresh (like epaper) rather than thirty or sixty times a second, and you'll save even more.

Posted by: Charlie Stross | November 25, 2007 5:17 PM

Charlie,

There's an additional power savings with using a slow core: you get to use slow DRAM, which uses considerably less power. If we could get flash ram in random access arrays that had reasonably fast access times at reasonable prices, the power drain would go way down.

Hildo,

The shrinkwrap industry isn't composed of the same companies as the enterprise industry, and it's a much larger market cap, just because there's no inhouse development backing it up. If Microsoft alone were actually to have to do a major restructuring and lost most of its value, it would do a lot of damage to the stock market, as well as put a lot of people out of work, and take a lot of aftermarket and service companies with it.

CEP,

The best route to an order of magnitude increase in power storage I know of is nanostructured battery elements. If that works well, it has the potential for a Moore's Law-style development curve that could easily last a couple of decades before structure size hits the monatomic wall.

That the PC is the mainstay product of the computer business is somewhat of an accident of history; it happened to be the market that, in the technical and financial environment of the 1980s provided the least resistance to the imposition of a 2 to 3 year product life, thus allowing easy recouping of the massive R&D investments required to keep Moore's Law trundling along. And we're talking here about a specific market segment: ultralight notebooks. The technical constraints on desktops are not the same, nor are the marketing considerations. The same kind of deflation will eventually occur in desktops too, but that market, and the TV/Entertainment device market could serve as a cushion to give the industry some time to find a new business model that's more profitable than commodity but more resistant to technical development. I predict they won't do that, but will instead thrash around trying to keep the ancien regime going.

Posted by: Bruce Cohen (SpeakerToManagers) | November 25, 2007 6:06 PM

For those here who know a little basic economic jargon, it's worth remembering that hardware and software are complementary goods, not substitutes. Yes, people sometimes balk at paying more for their software than for their hardware, but when the software is something they feel willing to pay for, the less they're spending on hardware the more they can ante up for the software.

The real story with the OS pricing and Linux ending up on more and more inexpensive computers - Dell is starting to gradually push it more - is that the basic functions of operating systems have become a commodity, after a couple generations of computer science development. Only Windows' monopoly dominance plus a hell of a lot of marketing has enabled them to maintain the illusion that an operating system is worth a lot of money. Ditto for many standard applications like word processing, browsers, basic databases, software frameworks, etc. (Programming environments and languages hit this point a while ago; the GCC toolchain, Perl, Python, Ruby, Eclipse, and the end effect is that MS gives away a 6-month free version of Visual Studio to keep developers interested, and has cut the purchase price of basic VS versions to well under $100 IIRC.)

But the same complementary good relationship is true between operating systems and applications components, and the applications themselves. If your OS is free or nearly so, your WP is free or nearly so, your DB server is free or nearly so, then if there's some highly specialized or personalized software you really want or need, you can now afford it. These are the software development areas I'd expect to boom. Unfortunately I have no idea what they are.

Incidentally, I would expect these factors to mean that Apple suddenly takes a big profitability hit a few years down the road; they're finally perfectly positioned against the last software and hardware market, so they should get a couple more years of great profit and market share gains, and then I'd expect it all goes blooey.

Posted by: Clifton Royston | November 25, 2007 7:17 PM

Bruce: I'd argue that the market for low-end desktops has already undergone most of that deflation. Just a few years back, the minimum price for a desktop was over $1000, with the manufacturers making every effort to push you up to $2000 via add-on options. Now Dell's pushing desktop systems with a starting price of $349 (without monitor.) That's roughly a 3-fold price drop, and of course you get more power in that desktop than you did in most desktops of a few years ago. (The next step up, $499 with a monitor, is dual-core, so probably more equal to the typical servers of 4 years ago.)

Posted by: Clifton Royston | November 25, 2007 7:28 PM

CEP@17: They can start by selling me (not 50% unit price, either!) additional batteries. Cheaply. Oh, and a charger unit. And a connector which I can plug into the unit on one end and the battery on the other so I can keep it running on battery power when having to shut down for a minute isn't really viable.

Posted by: Andrew Crystall | November 25, 2007 8:07 PM

"There is no intrinsic reason why a kilogram of plastic and metal with a couple of silicon chips in it should sell for more than its weight in silver." I'm sure that argument could be applied to the constituents of the human body to terrible effect...

There are still problems all over - Symbian 60 seems to need more RAM to operate effectively on my Nokia phone, than Windows XP does on a PC, and takes longer to boot. I also wonder how effectively Open Office et al will continue to develop without Oracle, Google and Sun pouring megabucks into them simply to kill the Microsoft cash cow. The commoditization may in the end be paid for by advertising space on every spare pixel.

I haven't looked it into it lately, but has anything been done to break through the problem of write endurance limits on Flash memory. If you have a database application writing record-level updates, then you can wear out your storage medium extremely quickly.

Posted by: Mike | November 25, 2007 8:37 PM

Hmmm... having a thread like this on LJ at the moment and I find myself more or less in the opposite camp. I think the phone business is an interesting place to look at. For a long time (well, a couple of years around 2002) many people were demanding that the phone business should grow up and become commoditised like the PC business - of course, the problem was, it already had.

The software costs on typical phone platforms are a pretty small fraction of the BOM costs, even something frighteningly expensive like WinMo or Series60 is a small part of the over all costs. The real costs for software rack up in adapting the essentially low price stuff you license and putting it together in a way you like. Depending on your starting point you're looking at somewhere between 20K and 100K man hours, depending on the feature set. Assuming around $75/hr costs for Western engineers and $20(ish) at Eastern, that's still a lot of money that needs to be recovered in unit sales. (And that will go for Android too, if half of what I'm hearing about it is true...)

There will be some things that people will be happy to do for free, but there's a lot of dull, crap, hard stuff, especially in the phone business, which won't get done, or will get done so badly it won't be good enough to ship in a mass market market device. So while software prices are going to drop, there will become a point where the needs of the mass market and the costs of actual mass market quality software development intersect.

I have no idea where that price point is, it'll be lower than MS currently charge, but probably more than they charge for WinCE professional ($16).

It could mean the end of Intel's drive for ever faster multi-core processors, but there's still a lot of development work needed in low power, higher speed mobile and embedded processors where there's a lot more competition; ST, Freescale, Qualcomm, Marvel, TI, Infineon, Philips, Toshiba, Samsung etc...

Of course, lots of silicon means lots of NRE porting stuff to it that isn't free either, which in turn...

Posted by: Dave O'Neill | November 25, 2007 8:48 PM

Symbian 60 seems to need more RAM to operate effectively on my Nokia phone, than Windows XP does on a PC

I'm assuming that you're talking Series 60 running Symbian? I'm pretty sure that that's a 64MB OS, probably running 64RAM/64ROM (128MB total) on a sub-400Mhz processor with an ARM 9 core. The processor is probably the real problem here. One of the reasons that the iPhone looks better and runs better than other phones is it's processor runs roughly twice as fast and it's got a lot more memory.

The real issue here is that Apple, by dint of who they are and what they sell, don't have the price pressures that people do on a phone and have been able to throw away the BOM rule book.

You can build a GSM phone for sub-$15 (probably sub-$10) but it won't do much. The component costs alone for something that will be interesting, look good and can run a Smart OS and Platform are still up over $150 - which is still a good $50 more than the global carriers want it to be (they have a target of sub-$100 and then they can start weaning the users off subsidised handsets).

The killer is how you amortise your NRE related to getting the product developed, tested and certified and into the shops and why the economics of the phone business are so bad. For every RAZR or T610 you'll have box loads of duds, all of which will have basically cost you the same fixed cost in software development. Making the basic core software free doesn't solve that problem if you still have to pay people to do the dev work, even if that dev work is done by a thousand dirt cheap software engineers in Nihzny Novgorod.

Posted by: Dave O'Neill | November 25, 2007 9:06 PM

CEP@17, the OLPC uses crank power. Would you find that acceptable for 24-hour power?

Charlie, the 10-year-old Mitsubishi Amity subnotebook (http://www.businessweek.com/1997/46/b3553067.htm) I have was $500 with all the extras I could add. It has a 7.5" color LCD screen, the display is 640x480, and it weighs 2.4 pounds. There are two problems: 1) Mitsubishi stopped making subnotebooks, so the upgrades I expected didn't happen, and b) the peripheral market changed to USB, so there aren't other companies providing add-ons.

So the big difference between what I have and the ASUS Eee is really better upgrades and peripheral use. It'll probably come out more expensive than the Amity, too. And I definitely think I'll wait for the next edition since it will take me a month or so in the new year to recover from Medicare Plan D from this year.

Posted by: Marilee J. Layman | November 25, 2007 9:23 PM

My Z88 ran 40 hours between battery changes, has a full sized keyboard, sufficient memory for what I did, and fitted into the top of my bag for visits.

The EeeeeeeeePC does indeed look like the current tool, but ...

Posted by: Adrian Midgley | November 25, 2007 9:37 PM

Adrian: I've still got a Z88.

You omit the fact that it ran for 40 hours on a four-pack of alkaline AA cells, that if it ran down you had to change cells PDQ before the backup capacitor discharged and it lost its memory, and that the only non-volatile backup memory cartridges for it were EEPROMs that took half an hour under an ultraviolet lamp to reset once you'd filled them.

I'll grant you the 40 hour battery life, but I'd like to point out that the Eee's power supply is a whole lot smaller than a Z88's brick. Indeed, it's about the size of a last-generation mobile phone charger. And how often do you have to work for four hours without any access to a mains socket (or enough privacy to change a battery)?

Posted by: Charlie Stross | November 25, 2007 9:45 PM

Hah, scooped you!

Actually the OLPC crank has been replaced by a pull-string, but a small solar panel also works.

Posted by: Randolph Fritz | November 25, 2007 9:52 PM

Dave @23: You raise a good point about licensing issues. There's a lot of IP to pay for - MP3 and other codecs (not all of course), DVD, USB blah blah blah. Don't forget language issues - side from UI translation, developing tools for language parsing, speech reco, handwriting recognition generally involve licensing lists, texts and samples from third parties. Building your own isn't necessarily cheaper by the time you've assembled enough computational linguists to do the work. Want to provide the tools for one of the many many language with less than 10 million speakers? - the rates go up due to lack of competition.

Want it all really really thoroughly tested before it goes out prematurely, whether it be from MS, Apple, Nokia or Mozilla? Then you're going to have to pay pay pay.

Posted by: Mike | November 25, 2007 10:03 PM

Mike: Where the PC industry does have it easier than the phone business is that there's no significant 3rd party certification requirements on the final device. Launching a device for the global phone community will require 3rd party GCF testing by an acredited testing agency and, typically, approval from the Operators on which you plan to deploy a device.

It's not all window dressing either. A few thousand phones with poor radio management can seriously affect the base station controllers in a network. There are things you can do with 3G QOS (Quality of Service) demands which are plain scary when you start to think of an era of more open radio and software defined radio.

Posted by: Dave O'Neill | November 25, 2007 10:14 PM

Dave @30: I can't speak to the hardware, but PC software will usually have to be certified it complies with various standards for accessibility (US government), privacy and encryption(various EU), ... in order to get sales to any government agencies, including schools.

I've been out of the game for a few years, but every year there were more and more of these to work through.

Posted by: Mike | November 25, 2007 10:28 PM

No, the crank isn't going to cut it. Just try deploying it on an airplane that's been sitting on the tarmac for 45 minutes, pushed back from the gate due to a weather hold at the destination. The same goes for pulling a string (which I'd expect to result in some stupid doll-like phrases anyway).

My point was that battery life is going to be a limfac (limiting factor) for the foreseeable future; no matter how much more efficient we make screens, processors, etc., we're still going to be dealing with the same power budget, in a way that is different from the progression in processor power, memory/drive density, LDC size, etc. over the last couple of decades.

Posted by: CEP | November 25, 2007 10:30 PM

7W is a fairly small solar panel, I think. I realise that would mean going outside...

On a related note, has anyone ever made a laptop with a translucent cover, so ambient light would pass through the LCD, and you could turn off the backlight? I'm tempted to try it with my elderly Thinkpad.

Posted by: Chris | November 25, 2007 11:07 PM

Power: I know it costs considerably more than £300, but I can't complain about my Sony Vaio TX1. It's put up with almost 2 years of my business travel, a half dozen major trade shows and upteen drops and dings and still gives me a solid 6-7 hours out of it's battery, can play 2 DVDs on a plane in DVD only mode and weighs in at under 2kgs. With my Kensington 70W power everything I carry pack it's an amazing piece of kit.

I can do London-Seattle with doing some work and watching some TV and still not have to plug in to the seat.

Posted by: Dave O'Neill | November 25, 2007 11:09 PM

The low end notebook market in the US is really dropping. While they aren't as light weight as the Eee PC or other subnotebooks, you do get a larger screen and more storage for around the same price or less. I just picked up an Acer Aspire for $350, which is more powerful than a desktop computer I was using 5 years ago.

In the near future I wouldn't be surprised if low end notebooks replace low end desktops as the starter computer of choice for light users. The costs aren't much different, and neither are the capabilities. You get a bit more in a desktop, and they'll always be more expandable, but most light users aren't going to max out their machine's capabilities or go poking around inside the case.

And notebooks offer the advantage of being at least semi-portable and having a smaller footprint around the home. It seems like every college student who isn't a PC gamer has a notebook now, and I find it likely they'll stick with them after graduation and once they move into the professional world.

So in the coming years the place where we'll likely see the most competition in the market is between low end notebooks and subnotebooks. Desktops will be mainly something reserved for the office, professionals who need a lot of power, or gamers.

Posted by: Andrew G. | November 26, 2007 12:03 AM

Thanks a bunch for your very helpful rundown of the Eee's pros and cons -- I think it's going to be my next computer after my 12" iBook gives up the ghost, probably in another couple of years. Like CEP said, the short battery life's a killer for me; hopefully by then they'll have figured something out.

Now...is 12 Gb of storage enough for a primary computer? I suppose if one stored music, etc., on external hard drives...

Posted by: Tlonista | November 26, 2007 12:04 AM

Remove the screen, sound hardware and input devices-- and add some storage (and ideally upgrade to gigabit ethernet) -- and you have the basis for a nice, quiet, storage / backup appliance.

Also of interest in the storage world: BitMicro are to start shipping (in volume in Q2) 3.5in flash HDDs with up to 1.6TB capacity each. No prices as yet, and I'm still waiting to hear how many re-writes you get per sector - and more importantly, how cool they run. But an indication as to the direction things might go..

Posted by: David | November 26, 2007 12:28 AM

Clifton @ 20

I don't think desktops have bottomed out yet. Exclusive of the display, a low-end PC isn't much more complicated than a commodity DVD player or a Tivo (which, really, is a low-end PC), and they retail for 80-100 USD. That looks like a bottom price-point for consumer electronics to me; down where the components are dirt cheap, and most of the cost is assembly and supply-chain. So with the display, commodity PCs should sell for around $150-180. Once we have a mass production on a display that doesn't use scarce materials or require expensive assembly techniques, I think the price will go down to $100. I give that four years max.

Posted by: Bruce Cohen (SpeakerToManagers) | November 26, 2007 7:28 AM

Interesting discussion, but premature, I think. We're still well within the feature-driven phase of the computing revolution (vs. cost-driven). If people were ready to start pushing the market based on cost, Linux would be on millions more desktops, no one would make PC games any more, and items like the iPhone wouldn't sell like crazy. No, people are still choosing their electronics on a feature (and "cool factor") basis.

The only way we'll really see a move to commodity PCs in the near future is if the technology advances to the point where someone can produce the same features at commodity prices. Not "almost as good" features, like many of the iPod clones; not "just as good but different" like Linux vs. Windows. Until then, this stuff is going to remain a purely niche market because your average consumer is willing to pay for those extra features on the higher-cost PC (even if he never uses them).

Posted by: Skip Franklin | November 26, 2007 10:17 AM

In the Royal Society archive is a video of the 2006 Clifford Paterson Lecture given by Professor Richard Friend discussing plastic/polymer semiconductors, which can be used for displays and laid out using ink jet printers. Well -- at least in the lab. It does seem an interesting direction and should help the costs to come down. It could open up a whole new approach to chip fabrication once you've got an inkjet that can print semiconductors.
reprap for chips, anyone?

I can't link directly to the video but here's the link to the main page http://www.royalsoc.ac.uk

Posted by: Richard Kennedy | November 26, 2007 12:48 PM

Andrew@7: the eeePC has been out for less than a month. There will be third-party batteries available by the time the original one won't hold a charge. I'll bow to wiser heads than mine and agree they probably won't be much better, although my old 3G iPod runs much better on its 3rd party replacement battery than the one it came with.

Charlie: you mentioned you upgraded the RAM. I understand this voids the warranty - can you confirm?

Posted by: Harry Payne | November 26, 2007 1:22 PM

There are some questionable assumptions being made here.

The first is that increased processor power does nothing to improve our lives, it just allows hardware makers to fleece us for more cash, whilst lazy software developers write sloppy bloatware to suck up the extra processor cycles without any functional benefit.

The second assumption is that higher bandwidth, ubiquetous internet access naturally implies that we will return to a thin-client, thick server model of computing.

I think both assumptions are false, or at least require independent justification rather than being held as self-evident.

Faster computers have several benefits. Firstly the sales of expensive, fast computers subsidise the cost of cheap, slower hardware such as the iPod Touch I'm writing this on - a machine comparable in power and price to the laptop you describe. It has always been the case that fast expensive chips have subsidized the slower, cheaper version's. If we stopped pushing forward the state of the art, then the new generation would not exist to become the old generation which ends up in budget hardware.

The concept that we could focus on reducing cost INSTEAD of increasing power is a misnomer. It is only by increasing power that we can pay for the reduction in cost of the older technology. Unless you are proposing that the electronics industry simply stops making faster machines, and lowers its profit margins year on year to keep the cost of hardware falling at the same rate.

As for the idea that servers will become more powerful, this also makes no economic sense. A server is a computer bound by the same rules as any other. If a thousand users of a web site all buy a computer half as powerful and push all the processing to the server, the server must now take on a thousand timesthe workload that each individual client has offloaded. The cost of that hardware upgrade will then be pushed back to the consumers, and on top of that they will also have to pay for the extra bandwidth required to shuffle data to and from the remote server that could have stayed on the client. Then there's all the overhead of wrapping that data on layers of encryption to stop it being snooped or tampered with in transit.

Posted by: Nick | November 26, 2007 1:43 PM

David@37: You do realise that a lot of routers can have USB hard disks plugged in right? Heck, one of the ASUS routers has a built-in hard disk. No need for another box :)

Harry@41: Except I'd want 2 batteries straight off - I do travel for longer than 3 hours on a fairly frequent basis. (Well, it shouldn't be longer than 3 hours, but given the English transport system...)

Incidentally, looks like ASUS has managed *again* to violate the GPL on the Eee release. I'm sure it'll get sorted (again), but you'd think they'd not need prodding to do it one day...

Posted by: Andrew Crystall | November 26, 2007 1:57 PM

Harry: there's a sticker across the memory hatch that says "warranty void if removed or broken". I would be interested to see whether that holds up in court in the UK, as I suspect it violates the Sale of Goods Act. On the other hand, the Eee is basically solid state, with no moving parts, so after a basic burn-in period it's hard to see what might go wrong with it within a one year warranty period.

Posted by: Charlie Stross | November 26, 2007 3:20 PM

Talking of computers ...

Someone has won the Wolfram Turing prize - see here:
http://news.bbc.co.uk/1/hi/programmes/more_or_less/7112637.stm

Posted by: G. Tingey | November 26, 2007 5:46 PM

An interesting detail re: EEE PC & RAM comes from ThoughtFix:

"It [the EEE PC] takes both 1GB and 2GB SO-DIMMs, but only 1GB is recognized in Xandros. XP [which runs on EEE PCs, too] recognizes 2GB. It's just a kernel option on the Xandros side."

Posted by: FrF | November 26, 2007 7:03 PM

Xandros is, IMO, a rather crappy distro; it's a Debian fork (so installing Debian or Ubuntu packages can break it badly), it's an out of date Debian fork at that (hence things like OpenOffice being stuck at 2.0 on the Eee, rather than the current version 2.3), and there's lots of stupid nonsense stuck to it. For example, Xandros supply a frankly inferior file manager for KDE that isn't as functional or elegant as Konqueror (which is still there if you know what to look for); they've got that silly maximum memory limit compiled into their kernel; they've left out HFS and HFS+ filesystem support (hello? Ever heard of Macintosh users? Or Mac-formatted iPods?); the version of Python is out of date; and so on.

Once someone digs up a workable driver for the wifi hardware (so I don't have to dick around with ndiswrapper) I will probably blow Xandros off my Eee and stick Kubuntu on it, clawing back about a gigabyte of disk space and adding functionality while bringing the apps up to date.

But you may take this to be a complaint by a Linux gearhead. As a second machine for someone who is not a Linux gearhead, it's fine.

Posted by: Charlie Stross | November 26, 2007 8:13 PM

Asus have apparently released the source code required under the GPL: see

http://www.engadget.com/2007/11/26/asus-posts-more-eee-source-code-to-comply-with-the-gpl/'

Sounds like an oversight, rather than deliberate breaking of the GPL.

Posted by: Spike | November 26, 2007 8:13 PM

Spike, I'm pretty sure that's the ACPI driver, not the wifi chipset. Still handy, though.

Posted by: Charlie Stross | November 26, 2007 8:22 PM

Skip @39.

re Asus being a niche market - I have current visibility of lots of orders (note: I don't want to get fired for giving out numbers..but lets say 'many' thousand); that's in the first month of actual release and with people willing to wait post xmas for delivery.

They are in such demand that I know certain folk are making befs at taking orders for 40,000 units before 08.

And that's from selling to a niche market.

You forget that the only clowns that want to pay £3k for a laptop are (generally) the ones who will have no idea what to use it for.

The Eee however, appeals to exactly the right people. Our children, who by and large will be/are far more tech savvy than we'll ever be.

Posted by: serraphin | November 26, 2007 8:29 PM

Incidentally, I would expect these factors to mean that Apple suddenly takes a big profitability hit a few years down the road; they're finally perfectly positioned against the last software and hardware market, so they should get a couple more years of great profit and market share gains, and then I'd expect it all goes blooey.

Apple is rapidly transitioning from being a PC manufacturer to being the highest-margin consumer electronics manufacturer. I'd say in a few years most of their revenue will come from the iPhone and similar gadgets. They're Bose on steroids, only instead of being based around loudspeakers, they're based around anything that needs a decent user interface, good industrial design and the ability to run software applications. They're already years ahead of Dell, HP and other PC-industry rivals in moving into the general consumer electronics space - those firms will probably never catch up (look for HP to sell off their PC division to the Koreans or Chinese in the next 5 years).

Microsoft is also screwed. Nintendo is eating them alive in the games market, their mobile platform is under siege from above (iPhone) and below (everybody else), desktop computers are rapidly becoming commodities, and they still aren't a dominant player in the server space even with all of their monopoly money. When that dries up, they dry up.

Posted by: Dan Flanery | November 26, 2007 9:04 PM

Wow, what a fascinating discussion.

True, low end laptop have gotten quite cheap. However, as Charlie mentioned, what may be game changing about the Asus Eee is its weight. All other two pound laptops are significantly more expensive. Most UMPCs weigh around two pounds, but they're also more expensive (and don't offer real keyboards). The laptops which cost close to the Eee's cost probably weigh around 5lbs. The Eee may spur more interest, and development, of really lightweight laptops.

As for computational power, for what an ultralightweight low-cost computing device is expected to do, a 624MHz ARM9 undoubtedly serves fine. However, I think there's a difference between what people need and what people want. The affordable people that would suffice, and the product that people are willing to buy may not be the same product. (OTOH, drop a zero from the end of the price, and no one may care. "The cheaper it is, the better the perceived performance" may be an axiom. Obviously, this fails at the limit.)

As for the "lower price" vs. "increase power" argument, commodization is clearly happening. IIRC, I read about Walmart selling a $200 Linux box. It's not hard to find a sub-$1000 laptop. The interesting question is what happens to the computing industry in the face of this. Not every computing device will be a commodity device, at least in the near future. (We won't get there until we have more computing power, and storage than anyone knows what to do with. For some people, we're already there. But there are still lots of people who, whether out of need or want, are willing to pay extra for more computational power.) It'll be interesting to see how the market deals with this.

Posted by: John Chu | November 26, 2007 11:04 PM

Microsoft is also screwed. Nintendo is eating them alive in the games market, their mobile platform is under siege from above (iPhone) and below (everybody else),

Hmmm.... it makes nice rhetoric but it's not quite the case.

Nintendo is eating everybody in the games sector, but the XBox is doing pretty solid business with the Xbox.

In the phone sector, WinMo isn't, by any real definition under seige from iPhone either. They shipped about 8 million WinMo devices in 2006, Symbian shipped something like 52 million devices. That's separate to the WinCE licenses they shipped which I'm actually gathering some numbers on.

iPhone had a great launch, but even at their target numbers they still are hitting a small part of what is an emormous market.

As I said before, the NRE of building phones cripples your cost basis unless you are going to be shipping multiple millions.

In the server sector, I know that their London (Financial group) blew away their targets and took in nearly $1BN in deals last fiscal and this year looks even bigger.

Posted by: Dave O'Neill | November 26, 2007 11:24 PM

But there are still lots of people who, whether out of need or want, are willing to pay extra for more computational power.

I wouldn't be so quick to say that. I work in a science department (biology), and the number of people here who would actually notice the difference with a faster machine is pretty small. And for those people, modern "faster" machines aren't actually any faster, because as someone else pointed out, stuff like GIS and statistical software doesn't really know what to do with multiple cores, etc. I know for a fact that R, which is probably the most up-to-date statistical software, can't thread hardly at all.

That kind of work benefits from more memory and better bandwidth into the processor, rather than sheer grunt.

Nick @42: slower chips have traditionally been underclocked versions of faster cheaps. They're the ones that don't pass the full-speed test at the end of the production line, so they get their multipliers locked at a lower speed and off they go to be sold at a discount. I don't see that model being a whole lot of use when the push is increasingly towards low-power rather than high-speed. Developing any new product costs a fortune and manufacturers defray the expense in whatever way they can. There are plenty of fields out there where the new product doesn't really do anything really new, but it still had to be developed, and production lines paid for. Most of the things that Charlie is calling a commodity fit that bill: new speakers, for instance, don't really do anything that old speakers didn't do. Same with food processors, and cars, and... you get the point, I think.

Posted by: Chris | November 27, 2007 12:14 AM

Hey Chris @54:

I do bioinformatics at UC Berkeley. I'll take ALL your spare cycles, plus all of your RAM.

My colleague in the CS dept is doing dynamic programming with FPGAs, mapping Solexa data onto the genome, and other colleagues have been talking about playstation clusters for years.

Then there's BOINC (ex Seti@home). Compbio can ALWAYS use more cycles.

Posted by: Iron Thighs | November 27, 2007 7:34 AM

Nintendo is eating everybody in the games sector, but the XBox is doing pretty solid business with the Xbox.

The Wii has already buried the Xbox 360, and it's only been on the market, what, a third the time? See http://www.vgchartz.com/?p=60 Worse, Microsoft invested a reported $21 billion (!!!!) in the development of their videogame platforms and their launch according to Forbes - heck, they supposedly sank a billion into just extending the 360's warranty when so many of them started to go tits up with overheating issues. They'll never break even on that investment, let alone turn the kind of profit it would take to justify an investment that massive.

Whereas the folks at Nintendo could already roll naked in the millions in profit the Wii has generated less than a year after launch, as the Wii is widely believed to have cost Nintendo a mere fraction of what Microsoft spent to develop its next generation console.

In the phone sector, WinMo isn't, by any real definition under seige from iPhone either.

Ha! Apple, selling just a single device, has already moved more than a million units in the US alone in under 6 months. Their plan is to sell 10 million by the end of 2008, which will place them within spitting distance of WinMo sales, and with a vastly higher profit margin. Not only that, but Apple's getting a cut of the monthly phone bill, a deal Microsoft certainly can't command. That may ultimately prove more profitable than selling the phones.

It's hard to see how Microsoft's going to compete with that, especially as hardware prices continue to decline and platforms capable of running the mobile version of OS X grow cheaper and more capable. As with the iPod, it's only a matter of time before Apple dominates the only profitable portion of the market - the high end. Let Microsoft have the junk. I'm sure the phone companies will be delighted to pay Microsoft a buck a phone to use their crappy OS, assuming they don't switch to Linux and get the same functionality for free.

Whoops.

Some software and hardware companies are going to make the transition to the post-PC world, and some aren't going to make it. It's becoming pretty clear Microsoft's not going to make it - certainly not in the consumer space. If they can't make it on the server side, the only real question now is, how long before they implode?

Posted by: Dan Flanery | November 27, 2007 9:30 AM

Here's something that might be a pointer to the software future.

Poser is a program at the bottom end of the CGI market. It started out as a digital equivalent of an artist's lay figure--one of those jointed wooden dolls--and became a fairly decent posing and rendering system. The render engine is comparable to professional software in its features, but it's stuck with a user interface with no user-adjustable features. I want to use a slightly larger font for the filenames, and something other than brown on brown for the colour scheme.

It's also stuck in the 32-bit single-processor world, and rendering needs a lot of clock cycles.

There was a promise of an improved version that would make the jump into the future promised by an OS such as Vista (though I gather there's a 64-bit version of XP). and the latest version will support rendering in a seperate process.

And a couple of weeks ago it was sold to a company called Smith Micro.

Looking at the company website, it looks to have a lot to do wuith software for mobile phone apps, and distributes a range of low-end software, such as photo-editing apps which are not Photoshop. Apparently, it was already distributing Poser in the US retail market.

So either Poser is going to be dropped into that low-end niche, and sold more like a computer game (which likely will lead to a flood of Poser-porn), or Smith Micro are trying to claw their way up-market.

But to get a more mass-market approach to work, the Poser manual is going to need a major rewrite. Major sections are written for people who know the jargon. Some of the explanations are as limited as "The Phone node implements a Phong reflection model."

And the 64-bit/multi-core barrier is still there.

But the other side of the Poser business is that a huge number of people sell models designed to work with the program, and part of what Smith Micro has bought is one of the content-retailing sites.

At $10 a time for a set of clothes, which is cheap, you soon can spend more than the software cost. And another pose/render program is already being given away (www.daz3d.com).

I can see the Poser market surviving the changes you're anticipating, because its so different. It's already a market where the program brings in less money than the support. And the support that is being sold is, in a virtual world, more tangible. It's not helpdesk time.

(Sample worksafe poser picture at http://myweb.tiscali.co.uk/zhochaka/Pictures/FurettePirate.jpg )

Posted by: Dave Bell | November 27, 2007 9:31 AM

I'll not argue that the Wii is doing well for Nintendo, but to claim that the Xbox isn't is, well, wrong. Besides, Microsoft don't really have to justify investment, anymore than Google need to. They're still recording record double digit growth in their core client markets (although I have no idea how). There's room for more than one games platform and I'll admit that I'm thinking of getting an XBox in addition to the Wii.

Their plan is to sell 10 million by the end of 2008, which will place them within spitting distance of WinMo sales

Yes, I remember when MS thought they were going to have 100 million devices shipped by 2006 too. I don't see, at this point, where the 10 million sales come from, the US sales seem to have flattened dramatically, and the European sales aren't tracking the US curve (hardly a surprise) - there was some editorial in the press about this yesterday.

Even then, MS shipped nearly 8 million units on their OS in 2006, so we don't know what the official numbers will be for 2007, let alone 8. I've seen some unofficial ones, more than a million Motorola Qs shipped on Verizon alone in 2007, so Apple will still be playing catch up.

The "cut" of the phone bill is going to go away too. I've no idea how they managed to persuade people at AT&T that this was a good idea, but it doesn't really stand up to the GSM phone model, as last week's court action in Germany showed. Apple will end up having to sell unlocked phones to a general market if they want to sustain their sales, and the operators will expect the prices they'll pay for shifting them to come down accordingly, which will squeeze the margins more.

Of course, MS don't really care about the margins on the actual phones because all they ship is the OS and make it as easy as possible for manufacturers to port to their chosen silicon.

Based on the new projects we're seeing from OEMs and ODMs I'd say that there's a lot of mileage in Windows Mobile yet.

Posted by: Dave O'Neill | November 27, 2007 6:43 PM

$21 billion for a gaming platform? Wow. Even with the dollar down the dump, this is much more than the planned cost for the research fusion reactor ITER (that is: construction + running costs for 10 years).

Ok, might be my German genes taking control of me and make me complain ;) but darn it, I'd much rather have ITER running now than the XBox. Why can't any of those firms or one the superrich guys take on these projects? I mean, for once I wouldn't even complain if Bill Gates were to found such a reactor. (And he could!)

Posted by: tp1024 | November 27, 2007 9:41 PM

to claim that the Xbox isn't is, well, wrong

How is it "wrong" to claim that a platform which cost in excess of $20 billion to develop and launch - and hasn't recouped a fraction of that investment - is a failure? If I spent $200,000 building a house, and could only sell it for $50,000, would you consider that a "success"?

XBox is a fiasco. Microsoft outspent Nintendo by at least $15 billion, and is losing in the marketplace. 360 weekly sales are now comparable to that of the (more expensive) PS3, also widely considered to be a flop. Look at this chart if you don't believe me: http://www.vgchartz.com/hwcomps.php?cons1=Wii®1=All&cons2=PS3®2=All&cons3=X360®3=All&start=39047&end=39411 Nothing like spending tens of billions of dollars only to duke it out for 3rd place.

European [iPhone] sales aren't tracking the US curve

I'd be surprised if the current-model iPhone was as successful in Europe as it's been in the US, if only because Europeans already have access to some pretty sweet phones. On the other hand, if you ignore ignorant editorials and go with actual figures reported by the parties involved, it looks as though the iPhone is selling up to - or in excess - of initial expectations. See for example http://money.cnn.com/news/newsfeeds/articles/newstex/AFX-0013-20911760.htm And of course, it's unlikely the current iPhone will last much beyond the New Year before being replaced by a substantially more powerful version. If Apple can address some of the device's more obvious limitations, they stand a good chance of broadening their market substantially in 2008. I'll certainly be looking at a GSM version of the iPhone.

Microsoft has been trying to crack the phone market for, what, the better part of a decade now? They're still in third place, after having spent billions. Likewise, they're battling it out for third place in the videogame market. Their forays into the consumer electronics space have been massive investment sinkholes - they'd have better served their shareholders by investing their money in Apple stock.

When it comes to consumer electronics it's clear Microsoft just doesn't get it, and with smaller, profitable, nimble competitors now dominating or moving to dominate the niches they've attempted to colonize, it's painfully obvious they never will get it. Not when their core business and cash cow - the Windows OS and Office suite - are coming under increasing pressure from ever-cheaper PCs and free (or ultra-low cost) open source alternatives. They simply aren't going to have $10 billion a year leftover to spend trying to give their hardware and software away to disinterested consumer electronics customers.

Posted by: Dan Flanery | November 27, 2007 9:47 PM

Iron Thighs@55: You're in the less than 1% who actually use a computer for computing. And if you aren't using some decent server-grade hardware, your department must be pretty tight. We've got a multi-CPU SGI system in a basement somewhere that the hardcore climate modellers use. Their stuff just doesn't scale down, you can't break it up across ten desktop systems. Which I think is part of Charlie's point; you don't need the supercomputer on your desk, one big server per thousand desktops is a much better way to do the heavy lifting (and ultimately more cost-effective).

I've seen clusters of all descriptions described as what happens when 20 different academics all want control of the pie, rather than letting someone set up what's actually the best system for the job ;)

Posted by: Chris | November 27, 2007 10:10 PM

I've dubbed this "bubblepack" computing since once the price drops a little more it will be selling in laser sealed bubble packs as impulse buys in Target/WalMart/Tesco.

Frankly, I see a huge market occurring, one that will dwarf the losses at microsoft. Frankly, I suspect that Intel will do well w/ all the servers, flash, and other chips in demand.

While they initially run the "same" software as PCs, bublepack (BP?) computers will be more like what microcomputers were to minicomputers. We'll have a lot of PC based servers, both in the googleplex and in each home (to store things w/ questionable IP or moral content). But the BPs will crowd out the PC for a lot of basic functionality just as the PCs did with minis. You will probably use your PC as a NAS, heavy processing (transcoding, rendering, image analysis) and as the gatekeeper for content within your house, but much of your physical interaction will be a combination of BPs and large monitors (HDTVs).

If the BPs crowd out the PCs and leak through the digital divide to the point that they are as common as cells, I see a much bigger market for smaller software. I also see metro wifi/wimax turning into a no brainer. Service will probably involve wiping the unit and tossing it into the recondition pile and then handing the customer a new one. Web companies will make out like bandidts, smart content providers will too, once they focus on juggling novelty and availability over IP extremism. I also see a new market "configuration distros", in which the basic BP gets a distro/acessory set which suits the user. A hello kitty distro with a HK skin, panic button and parental controls, a set of student distros tweaked for note taking interfacing with the college network and major specific software. At the cost, getting one for work and one for leisure won't be uncommon. Or the distro may just be in the flash installed allowing for a personality shift as you flip between flash modules.

If Microsoft follows the status quo Redmond will be the next Detroit. But in this specific case, I do belive in the creative destruction that will result.

Posted by: Monopole | November 27, 2007 11:02 PM

Re: XBox - do you have a cite for the $21bn in investment for the xBox handy, the only quote I can find has Robbie Bach saying around "3.8bn", which while a serious chunk of change is a lot less than 21. It'll be interesting to see if the PS3 maintains it's pick up. You've 2 platforms which are almost neck and neck in sales according to the chart you provided, both aimed at different parts of the market. We have a Wii because my wife actually was prepared to have one because it's focused at a different sector of the gaming market. It was a gamble that paid off.

Apple: Microsoft launched Stinger in 2002, so they've been at an actually mobile phone play for 5 years. They're second to Symbian in the Smartphone market and gaining market share. As I said, I base this on the projects we do (I work for a company that puts together the software builds for most of the world's phones) and we're seeing huge amounts of traction in Windows Mobile, especially since they launched 6.0.

I'd not see that quote from Erskine, I had seen the one from the previous day where he said it was probably "tens of thousands". But as you point out the iPhone isn't as big a deal in Europe. BTW - the iPhone is a GSM phone, it's just crippled by being locked to certain networks (apart from in Germany).

Microsoft's phone play is more interesting because they didn't go for the consumer electronics play. They're licensing a fast to integrate and deploy mobile OS which any OEM/ODM can build a phone around. It's proven harder than they expected to get it right, but from what we're seeing of the devices starting to emerge from the leading OEMS (Motorola Q series, Samsung Blackjack, etc...) they're really starting to get some traction, plus while there are limits on what you can do, it's pretty flexible. So you can have something like the 2 versions of the Palm Treo, the Blackjack, the Q plus the touchscreens.

The problem for the phone market is the software, as a proportion of your development costs is a really trivial amount, so it's a better deal to license something at $10(ish) where the core integration work has been done well, than it is to use free software but be looking down the thick end of a 40K hour+ engineering project.

In this regard MS, Symbian (S60 and UIQ) and, probably Android, will end up duking it out for market share with Apple having a nice little niche.

The phone is too much of a commodity good for them to repeat the success of the iPods. Even if they ship the numbers they're aiming for over the next 2 years, they'll still be well short of the MS, Symbian totals and way short of Nokias _monthly_ shipping numbers.

And, finally, while everybody keeps telling me that MS are coming under increasing pressure, they keep delivering double digit sales growth in their core markets. Now that has to stop one of these days...

Posted by: Dave O'Neill | November 28, 2007 12:03 AM

do you have a cite for the $21bn in investment for the xBox handy

http://www.forbes.com/finance/2007/04/18/xbox-microsoft-nintendo-pf-ii-in_re_0418soapbox_inl.html

BTW - the iPhone is a GSM phone

I'm old and getting my computer industry acronyms confused. Meant to say 3G, not GSM.

It's proven harder than they expected to get it right

They've spent a fortune and it's taken them years. Apple launched out of the chute with a better product, which they developed in about a year for a fraction of the cost (hardware AND software). This does not bode well for Microsoft's future in the phone market, let alone the rest of the consumer electronics space.

They're second to Symbian in the Smartphone market and gaining market share.

Not according to Wikipedia they aren't. According to this page they rank a distant third, behind Linux. http://en.wikipedia.org/wiki/Smartphone RIM is threatening to overtake Microsoft in the smartphone market, pushing them to fourth.

Sort of embarrassing for the world's largest software maker. 5 years and they're still a niche player with a commodity product. Apple will probably make more money off of smartphones in 2008 than Microsoft.

Even if they ship the numbers they're aiming for over the next 2 years, they'll still be well short of the MS, Symbian totals and way short of Nokias _monthly_ shipping numbers.

Yes, but Apple will be making money off of every unit sold. The other guys, not so much - they can't command premium pricing. And I expect to see OS X begin to crop up on a slew of other Apple-branded consumer electronics devices over the coming years - we're already seeing it on the iPod Touch.

Posted by: Dan Flanery | November 28, 2007 1:32 AM

I'd note that the iPhone has fairly reliably been scoring really badly on useability tests, and personally after trying one there's simply no way I'd buy one - I prefer tactile feedback, thanks! (this is also why I mourn the demise of the rollerball on notepads - one old laptop I used had it at hinge, to the right, with the buttons on top of the case. This worked really quite well...)

Dave Bell@57: What you mean "lead to" on Poser porn? (Let's just say that being careless with google image search lines can accidentally expand your knowledge in bad, bad ways)

PS, todays great spam subject line: "Beware of beam-controlled government mimes"

Posted by: Andrew Crystall | November 28, 2007 1:53 AM

Hmmm... the Forbes article throws in that number and doesn't back it, plus it's quite out of date hailing from April and not taking into account the huge surge in the XBox sales that your charts you posted have shown. Robbie Bach's quoted number at $3.8bn sounds more credible. I'm slightly aquainted with the operating budgets of that MS unit and $21bn doesn't fit in there.

Phones:
Ugh. That wikipedia article is pretty misleading. I'd like to say more, but take the Linux number with a pinch of salt, in the phone sector not all versions of Linux are actually running what would be known as Smartphones, especially the DoCoMo and chinese market devices, they're really more high end feature phones. By that logic, you could say that all of those are beaten by high end Nucleas or OSE devices running J2ME applications.

Sort of embarrassing for the world's largest software maker. 5 years and they're still a niche player with a commodity product.

No, they're really not. They're supplying a complete phone platform like Series-60. It's being used by 3 out of the 5 Tier 1 OEMs and every single ODM trying to break into the phone market in Europe. There's a bunch of Tier 2s in there too.

I'd be happy to continue this via email, partly because I think it'll get boring for people and also there's lots of stuff I can't talk about publically because, as I said, I work in this stuff.

Not for MS, in case you were wondering. We work on Linux, MS, Symbian, RIM and lots and lots of RTOS based phones.

Posted by: Dave O'Neill | November 28, 2007 4:44 AM

OSX - yeah, one last thing. The Apple kit is currently running with 800MHz processors, and apparently it's about 500MB of code.

As a high end device that will have a loyal "high end" spender appeal that should work. It will cause them problems for the rest of the market. To put it in perspective, the monsterous bloatware that is Windows Mobile 6.x fits into a 64MB ROM (128MB prefered) and will run on a 250MHz processor. It's one of the reasons that the BOM on an iPhone is about $250 compared to most other Smartphones at sub-$150. As long as Apple can stay playing in the high end of the market, the sums will work for them. But it will take a lot of Moore's Law to get the prices for that kind of performance down to meet their competitors, especially some of the Flash based devices that are coming along running on RTOS platforms.

Posted by: Dave O'Neill | November 28, 2007 4:49 AM

re: Poser porn

I'm not sure if they can sell the program in Saudi Arabia.

Posted by: Dave Bell | November 28, 2007 6:15 AM

The 21bn quote isn't for the Xbox. It's for MS's home entertainment division in total:

"Microsoft's Home & Entertainment Division ("H&E"), which includes Xbox 360, Xbox, Xbox Live, Consumer Software and Hardware Products, and IPTV."

You should also keep in mind the box actually makes a loss every time it sells - much like the PS3. The money is made back from the licencing fees on the games, from subscription to the Live service (the thing you need to play online/chat), downloadable content, etc.

It's a subscription model - not a purchase model. I think you'