Mixed Signals
Any sufficiently advanced information is indistinguishable from noise
Wednesday 11 June 2008, 10:07 PM
Apple vs Microsoft - the chips are down
Should you buy your chips ready-made, or cook them yourself? Depends whether you're Apple or Microsoft.
Let's take Microsoft first. It has its own Xbox Semiconductor Design Group, which among other things designed the graphics processor for the Xbox 360. Unfortunately for the group and Microsoft this is the component that's being fingered as the culprit behind the Red Ring Of Death. The chip dissipated too much power, ran far too hot and frequently expired. According to EE Times (apologies for the long quote):
"The Xbox 360 recall a year ago happened because "Microsoft wanted to avoid an ASIC vendor," said Lewis [Bryan Lewis, chief analyst at Gartner]. Microsoft designed the graphic chip on its own, cut a traditional ASIC vendor out of the process and went straight to Taiwan Semiconductor Manufacturing Co. Ltd., he explained. But in the end, by going cheap--hoping to save tens of millions of dollars in ASIC design costs, Microsoft ended up paying more than $1 billion for its Xbox 360 recall. To fix the problem, Microsoft went back to an unnamed ASIC vendor based in the United States and redesigned the chip, Lewis added. ([...] most likely the former ATI Technologies [...]) Asked the moral of the story, Lewis said: "Had Microsoft left the graphics processor design to an ASIC vendor in the first place, would they have been able to avoid this problem? Probably. The ASIC vendor could have been able to design a graphics processor that dissipates much less power."
Lewis goes on to conclude that systems houses have no business designing their own chips, and should leave it to external companies that know what they're doing and have plenty of capacity at the moment.
Logical enough. But Apple's now giving out a bit more information about its purchase of PA Semi, the processor design company it bought recently amid speculation that it's looking to cut Intel out of some of the action. At the time, I thought it might be for some custom high-performance circuitry aimed at new appliances, but since then there have been rumours that PA Semi was looking for a white knight, was friendly with Apple anyway and may even have had projects on the go for Cupertino. In which case, Apple had stuff to rescue and saw an offer too good to refuse.
Apple is indeed going to design its own chips, but Intel shouldn't be too worried: they're going to be for places Intel doesn't yet go. PA Semi is going to do system-on-chips for iPhones and iPods,, Steve Jobs told the New York Times. Given that one of the iPhone's big problems is its cost estimates range between $400 and $460 wholesale keeping Apple reliant on the handset discount model and the power in the hands of the networks, then this looks plausible. You get the best cost reduction by making chips that are precisely tuned to the needs of your particular hardware. If they work.
So, who's right? The two cases aren't directly comparable. PA Semi has exactly the sort of real-world experience bought in a commercial environment that Microsoft's design group most keenly missed. But that ends now: one of the big problems of having an in-house group is that it doesn't get exposed to lots of different demands by lots of different companies, And other parts of the company feel compelled to use that group instead of getting a more appropriate deal outside: valuable cues are missed, politics blossoms and competitive advantage is lost. There are places and times when having your own chip-designing expertise is a good thing Sinclair Research's use of Ferranti's ground-breaking ULA in the ZX81 was a good example -- but I'm not sure either are common at the moment.
I'm with Gartner on this one. The Red Ring Of Death should hang above every technology company CEO's desk a baleful anti-halo, warning to stick to the knitting.
Wednesday 11 June 2008, 1:19 PM
UK P2P user? Hope you like US prison food
This is one of the most frightening things I've learned in a long time. Over in the US, a bill has passed the House of Representatives and is heading to Congress with a huge amount of support.
The PRO-IP bill, H.R.4279, significantly increases the state's power to detect and prosecute IP infringement, carrying with it a whole host of new law enforcement positions and capabilities. It establishes an IP Czar, someone with the job of overseeing zealous action on behalf of copyright and trademark owners, and includes such powers as the ability to seize equipment if it contains just one file thought to infringe.
Importing and exporting infringing material will attract harsh penalties, and there's a $30,000 per-track fine on music (so that's half a million dollars for an album), The list goes on, and I thoroughly recommend you go out and Google to educate yourself on the many quite overwhelming powers the US government wants to give itself in its apparent determination to put file sharing on a par with drug dealing, gangsterism and other great crimes against society.
Thank goodness we're not in America? That hardly helps. Among the many provisions is the establishment of "five additional Intellectual Property Law Enforcement Coordinators in foreign countries to protect the intellectual property rights of U.S. citizens [...] increase DOJ training and assistance to foreign governments to combat counterfeiting and piracy of intellectual property." -- and if you think their job is just to lead the rest of the world in the way of American righteousness, think again.
Transferring a file between the US and the UK, however you do it, will most certainly count as importing or exporting: that will include BitTorrent, web hosting, FTP, Usenet and all those popular ways of moving stuff around the Internet without worrying about who's doing it or where they are. Even if you're scrupulous in avoiding the obvious areas of naughtiness, I doubt very much you know how much of the stuff you've browsed today was hosted in the US, let alone how legal or illegal it strictly was. And if you're a torrent fan, well, good luck to you.
As a UK citizen, you no longer have any effective defence against a US demand for deportation. Under the Extradition Act 2003 the US can apply for a UK citizen to be extradited without having to present any evidence to face charges of a crime committed in the US for which the UK citizen need not have been actually present.
For an example of how this works, take a look at the case of Brian Howes and Kerry Ann Shanks, who have had seven months detention in the UK and are now facing deportation to Arizona and potential 100 year sentences. They ran an Internet chemical supplies company selling, among other things, iodine and red phosphorus which is perfectly legal in the UK but not in the US, where they're controlled as precursors to methamphetamine production. Quite a lot of these chemicals found their way into quite a lot of meth labs in the US meaning, say the Americans, that Howes and Shanks knowingly supplied the substances.
They haven't been charged with anything in the UK, but the only defence they have against the extradition is to show that it didn't follow the right procedures. There's no test of evidence, and should they be deported they'll have no access to legal aid.
It can certainly be argued that it's wrong to supply drug factories with chemicals on an industrial scale. In that case, the normal rule of law should apply: make it illegal in this country and test the cases in UK courts.
It's far harder to argue that IP infringement still basically a civil matter, frequently of debatable harm in most cases should put UK citizens in peril of a foreign judiciary with no effective safeguards whatsoever.
But this could happen. It could happen to you or me are you in total compliance with all your copyright material? I know I'm not: I actually read EULAs.
See you in the state pen.
Wednesday 11 June 2008, 10:59 AM
Bugs haunt iPhone SDK, Jobs comes out in sympathy
As Steve Jobs spoke, Apple pushed out a new version of the iPhone SDK to the 4000 developers previously deemed worthy to join the programme. As I type, many of them still haven't got it working properly.
It seems from various concerned threads on the Apple developer forum, problems have been introduced in this version that prevent applications being moved onto a physical iPhone (the simulator within the SDK works fine). As is usual in such matters, some people claim to have fixed the problem and posted fixes while others say that these fixes don't work for them.
"It's almost understandable", one rather disgruntled developer told me, "but definitely a touch of the rushed release syndrome. Releasing a buggered SDK in the middle of your big conference isn't the best thing you can do for your developers."
And this wasn't the only bug-laden system shown off to the world during the keynote. Lots of people noticed that Steve Jobs was absent for much of the time - his presence replaced by an endless stream of application show-and-tells - and when he was on stage, he looked emaciated.
As he was previously very ill with pancreatic cancer, an illness that Apple kept secret until his recovery, the rumours started to fly almost as quickly as the iPhone 3G's spec sheet. Apple's PR felt it wise to issue a statement. According to the Wall Street Journal's Biztech blog:
"In response to a question about his health Tuesday, an Apple spokeswoman said Jobs was hit with a common bug in recent weeks but he still felt it was important to participate in the Apple conference. The spokeswoman said hes now on the mend with the aid of antibiotics."
Tuesday 10 June 2008, 1:03 PM
Security through annoyance
A close personal friend reports a bloody annoyance. The place where he works decided to tighten up its network security, by forcing all users to change their passwords every three months.
As usual, the system checks to make sure that you're not just switching to an old password; the result, of course, is that people will write down their new passwords and stick them on post-it notes, because as soon as they've learned to memorise one it'll be time to change it. Or they (like me) just fiddle with a tiny part of the password, incrementing a digit or similar.
Most security professionals I've discussed this with think that forced password retirement is far more trouble than it's worth, and if anything reduces security while increasing their workload.
That's bad enough. But my close personal friend's place of work is a Microsoft shop, and has doled out Windows Mobile phones for email and other uses. These have to have their ActiveSync passwords synchronised to the new passwords manually - every time you change your password, you need to put it into both the main system and the mobile.
Which itself is more than bad enough - secure passwords should have a mix of letters and digits, and making sure you get those right typing blind into a tiny keyboard where you have to switch modes for the numbers is an absolute usability nightmare.
But more than bad enough isn't as bad as it gets. My close personal friend had to change his password when he was away from the office - and his work phone was at the office, charging, where he'd left it over the weekend. The result was that after the mobile had tried to synch and failed a few times, the network decided that there had been too many failed attempts to log in and locked the account completely.
Lessons? You should never have to enter new passwords into more than one place. You should ensure users have an intrinsically secure password that they can memorise, and let them use it. Increase security through awareness, not annoyance.
Usability is a major part of security: without it, the systems don't work.
Monday 9 June 2008, 1:33 PM
xG's unclear communications
I'm finding it increasingly difficult to know how to handle xG, the Florida based but London listed wireless data company with a breakthrough technology it's kept out of public scrutiny for many years.
To date, none of the company's plans has come to fruition. Its first generation of product has come and gone without deployment or any proof of functionality, the flagship roll-out cancelled amid a cloud of ill-feeling and general confusion. Evaluation deals with various third parties have apparently expired without further interest. The tiniest amount of technical data, uncheckable and largely information-free, has been presented on the xG website. What partners it has are contractors, doing xG's bidding while refusing to say anything more.
What there is, is rumour, and in great profusion. And if the technology remains mysterious and uncheckable, then the rumours - mostly promulgated on a couple of independent bulletin boards - are doubly so. There is - or was, it may have been cancelled but I saw the original filing - a court case against the company that may or may not have been brought by the chief engineer, who may or may not have resigned with a few others over what might or might not be arguments over promised stock options that may or may not have been withheld.
That's just one. No aspect of xG's operation has gone unquestioned, and precious few of those questions have been adequately answered. if you have the slightest interest in what's really going on, good luck to you in finding out.
But if you do have that interest, you'll be practically alone. For one constant I can report from first hand experience: nobody without a financial interest in the company has the slightest belief in it.
Since xG invited me and a handful of other hacks over to Florida in 2005 to see a demonstration, I've talked about the company to engineers, network operators, handset manufacturers and anyone else in the wireless business, from independent analysts to billion-dollar CEOs, who are too polite to get away quickly enough.
The replies have ranged on a spectrum from "Who?" to "Ha!".
Doubtless xG's more enthusiastic supporters will claim this is an industry in denial, running frightened from a seismic event that will overturn empires. But I've found that bona fide industry players take real competition seriously and with respect: you get plenty of spin, sure, but where there are real questions they tend to get real discussion.
That's entirely absent here. On the technical and business front, xG is invisible. It's not behaving like any other public company three years into a strategy. It remains impossible for me to match the company's public statements with anything going on outside its walls, and I'll be extremely surprised if anything resembling a rabbit appears out of this particular hat.
But I'll carry on watching, like those under the command of the apocryphal Army officer whose assessment ran: "His men will follow him anywhere, if only out of curiosity".
Previous
