Posts by James 100

Had it in May

I had LASIK on both eyes (by Optical Express, May 3). £2,500.

Apart from allergic reactions to two of the eyedrops and a small corneal injury (sharp stray piece of plastic from the disposable vials), no "complications" as such, just significant under-correction in the right eye.

End result? My vision is *nearly* as good as it was wearing specs - just short of 20/20 in the left eye - but with a LOT of extra 'floaters', and all goes to pot in low light when my pupils dilate beyond the correction zone. Dry eye greatly improved, but still needing drops a couple of times a day.

Improvement? No chance. If I could undo it, I would in a heartbeat. If I could somehow pay another £2,500 to have it all undone, I'd jump at the chance. Technically, though, this qualifies as a "success".

It searches for a pair of file/directory names which hashes to a particular value, then uses a salted hash of that name as the key. Because the added salt changes the hash completely, there's no realistic way to get from the (known) unsalted hash value to the unknown hash which is used as a key besides finding the filename it's looking for.

One interesting aspect is that the path entries themselves seem to be being used as a sort of salt, or one of them at least - further obfuscating the file it's actually looking for in %PROGRAMFILES%.

Parts of it are eerily familiar to me - I implemented something vaguely similar a few years ago as a simple licensing hack (the published code checks for a filename with a particular SHA1 hash). Of course, it's not the first time this has been used, malware was doing this (with CRC checksums of filenames) back in DOS days...

Bigger problems

Personally, I quit Carbonite for Crashplan a while ago - not because of the performance, more because when I moved offices and computers, I thought 'ah, I'll just install Carbonite here and restore my backup to the new machine' ... a few days later, it had given up on a bunch of files. It didn't actually say WHICH files, just that "some" files had failed to restore properly. Once I worked out which were missing, I tried restoring just those files ... long wait, same error. Customer support: "hm, looks like we can't get those files back, would you like a cookie instead?" (Actually, a free service extension. As if I'm going to keep using a backup service which has already given up on protecting my data!)

A backup service that can't restore the data again? Even free it's overpriced.

I had other backups of course, so nothing was actually lost, except confidence in Carbonite. Since that, I've backed up and restored plenty with Crashplan, without the slightest hitch - and the local backup is a big help (I back all my machines up over the LAN to a big external drive, as well as their servers, so I get nice fast restore times there.)

A 100kbit/sec throttle really is taking the mickey. I was about to compare it to my local Indian buffet having a policy of "after the 1st plate, you have to eat through a straw", then thought that bit could be misinterpreted - but when they are capping you to literally dialup speeds - slower than bonded ISDN, slower than my very first half-megabit cable modem from over a decade ago - it really is that extreme.

Dilemma

If India really did have such a key, you'd expect them to keep quiet about it so people would continue using it in a vain attempt to protect secrets from them. On the other hand, if RIM *did* have a master key or some way of generating one, we'd expect them to deny its existence for similar reasons...

With the tight network integration, even BES still involves RIM-controlled servers in the traffic. Now, that might be a purely opaque encrypted tunnel, with the RIM kit only knowing which BES server a handset is contacting, and vice versa, but since it's a proprietary protocol, who knows? With a tame CA, you can do a man-in-the-middle 99% of people won't be able to detect: how do we know there isn't some equivalent in the BES/BIS protocol, allowing the Indian government to have the equivalent of a wildcard SSL certificate?

Ultimately, either the Indian government is lying and doesn't really have the key, RIM were lying to them and provided a duff key, or they really do have a genuine working key. The middle option should become obvious as soon as they try using the duff key and crack down on the handsets and RIM themselves...

Supporting infrastructure?

HP are having to pay Intel to keep the chips coming in the short term, they've had to sue Oracle to keep the database engine coming - what does Oracle need to do that? Will their C compilers, test/dev hardware etc still be available? Will they get operating system source access, to ensure parity with their Linux/Solaris/Windows offerings?

A quick search on hp.com for Itanium products suggests perhaps I wanted 'titanium' instead - whoops! If I wanted to develop Itanium software, in the past I could have used Dell or HP workstations - oops, those have been discontinued already. Virtual machine? I can fire up Linux, *BSD, Windows, Solaris, even OS/2 (eComStation) virtual machines for development purposes - anything on Itanium? Never mind that Windows and the Linux distros have both dropped it already (does Oracle have to keep supporting the database on those?)...

It's an alarming demand: support a dying fringe platform in perpetuity?! Oracle don't have to keep supporting Windows, or z/OS, or Solaris if there's no market for it - but suddenly, supporting HP-UX (or NonStop? OpenVMS?) is mandatory whatever happens - until HP finally drops the whole architecture completely.

If I were Oracle, I might be tempted to reply to the judge "OK, but HP will have to provide X Itanium workstations running each version of the OS they want supported, with supported standard-compliant C compilers and libraries to match", where X is at least the number of people needed to maintain and support the products. How can HP drop half the product range itself, then demand other people keep supporting it regardless?!

General principles

As a general principle, I would like to see what the media persists in calling "hacking" resulting in custodial sentences, and trial in the country where the target computer was located. If someone were to shoot me from across the border* or mail me a parcel bomb from another country, I would want them extradited here to stand trial for it.

I abhor the notion that breaking into and corrupting computers is somehow not a "real" crime, too, whether it was easy or not. Why should compromising a PC by guessing the password be regarded more lightly than breaking into someone's office by wiggling a credit card in the door?

As for this particular case, having dragged it out for over a decade seems bonkers. To have gone on a tour of almost every court in the land, without even reaching the actual trial, just arguing over whether or not the trial should actually happen ... What on earth takes so long?! OK, the courts are busy places, but still, ten years?

* This is actually one of the example cases on the CPS's website somewhere: supposing someone in Wales shoots somebody standing across the border in England, where does the trial take place? Oddly, the answer is not clear-cut, it's supposed to be determined by factors like the location of witnesses/evidence.

Conflicted

It's already been reported elsewhere that Yeo has a rather big conflict of interests here, being employed by several of the "green" companies he's advocating funding so generously!

Of course he'd love his businesses to get massive guaranteed subsidies - who wouldn't? The question that should be considered is whether we're happy to pay for them - which I for one am NOT. If his companies can't deliver the product for a viable price, why should we pay to make up for their failings?

Like the handover note said in 2010, we're out of money. In a sane world, that would mean we stop giving it to failed businesses for products we don't want that they can't sell. Sadly, that's not the world politicians live in!

Efficiency

In terms of shuffling the paperwork, this sounds like a big improvement (and this is only partly influenced by having spent a few days earlier this year preparing c 10,000 pages of evidence - for a case which was ultimately settled anyway, but because of the time involved, all the paper evidence had to be prepared before that happened). The time and money involved in all the photocopying, filing and transportation is staggering - and completely pointless, when it would have been *better* exchanged electronically in the first place! (No search facility on a massive paper dump, for one thing.)

I do understand concerns about swifter justice risking a loss of accuracy: in some parts of the process that's certainly the case, but simply exchanging documents electronically rather than on dead tree, by email rather than LP or DX? Where's the harm in that? Not to mention access; easy for a lawyer to turn up without some document, a secure comms link would let them retrieve the document in seconds instead of having to delay the court while it's physically retrieved from another building.

Virtual machines?

Amazon stores the EC2 virtual machine volume snapshots (but not the volumes themselves, EBS is separate) in S3 behind the scenes, but each is a single large file. If the virtual machines stored either the actual files, or used smaller chunks, that could easily balloon the count enormously. (100k virtual machines, 10 snapshots of each, 1m files - bang, that's a quarter of the 4 trillion straight away!)

Or of course if they're using it to hold the mail and other data for Office 365/Live@Edu, that would account for a huge chunk. Tens of millions of students/staff each with tens of thousands of messages in their mailstore - easy to hit another trillion that way.

Nice to see a bit of competition for Amazon, anyway: I like - and use - S3 myself, but good to know they're not the only big player in the market.

48V and more PoE would be nice. I wonder which direction monitors will support this in - or will they support both? Having my laptop charge down the monitor cable would be nice (more or less what Apple do with the Thunderbolt displays: displayport+magsafe), but so would having the monitor powered from the desktop rather than needing its own power cord (like Apple used to with some of the G5 towers I think?).

5A at 20V should be enough for a lot of things - laptop charging, some printers, scanners, any kind of external storage, decent USB hubs without needing a power brick or being limited to self-powered devices only. The combined laptop-charger+USB-hub route sounds good to me: I could have one at home, one at work, connect the power supply in to the network, printer etc and just have a single cable to handle.

I await with interest his proposals for bringing all the EU's own websites into full compliance with Iranian and North Korean law, then. Better delete any references to Israel's existence, then, to comply with the laws of various Arab countries - and Kashmir will be tricky, needing to comply with both Pakistan and India's rules on that subject.

He did publish his ruling simultaneously in English and Canadian French, as required by the laws of Quebec, I hope, and make sure it's not available during any public holidays anywhere in the world in case of violating national or local holiday ordnances?

I'm in the UK, my personal server's in the US. I have no problem with complying with those laws, they are actually relevant - but the idea I now need to comply with laws of countries I have never visited, just because someone in that country looks at my website? Utterly crazy.

23m customers - until this happened

"According to their corporate information RBS Group has 26m customers in the UK."

That was LAST week... expect smaller numbers now.

"The fiver was moved between accounts, using faster payments,"

FPS is separate from BACS - something wrong with the RBS BACS connection?

Quite why the whole RBS-Natwest-UlsterBank-everythingelse behemoth was allowed to grow to that size in the first place, I don't know. (Well, I have some suspicions...)

Audio only?!

I didn't bother starting it - why no transcript? This is meant to be a website, not a radio station!

As it should be

'I worked for a large company in the early 90s and they always got what they wanted. If it wasn't on the approved list (and not able to be had via the purchasing dept.) they'd put it on their corporate credit card and then have IT install and support it.'

That sounds about right: within reason, the function of the IT department should be to support what the clients need, not to dictate what the clients get. If the head of one of our law schools wants a slimline ultrabook and an engineering prof wants a multi-Xeon monster to do his CAD on, so be it: IMO, forcing them to have something "standard" to make our own jobs easier at their expense would be a dereliction of duty and detrimental to the organisation as a whole.

OK, Sony Vaios need to be purged from the face of the planet, and I'm not exactly suggesting deploying Angry Birds through central provisioning servers, but I've seen far too much dictatorial "standard" imposition lately, often for little more than the sake of a power-trip by those imposing them. I'm reminded sometimes of tales of American Home Owners Associations which dictate the colour of your front door to suit their whims...

I was surprised by the focus of this too: I've always used nVidia rather than ATI in my machines because of their *better* driver support for Linux - and as others have pointed out, ATI manage to suck for drivers even on Windows. Laptop with an ATI chipset? No drivers for you! Talk to the manufacturer ... yes, they went bust months ago and never bothered providing support or drivers anyway, so you're stuffed. Thanks for buying ATI - won't be making that mistake again, will you?

The Optimus thing is interesting: it seems nVidia needed access to the DMA-BUF buffering stuff, but some developers were objecting because their driver is partly closed-source.

I can understand the desire for GPLed drivers from nVidia, but wearing an end-user hat this does look very much like "users screwed by ideology". ATI don't seem any better in this respect - at least, their Linux drivers are also proprietary and will hit exactly this issue too - it's just nVidia were actively involved in Linux kernel development and seem to have hit it first. Maybe if Intel manage to produce competitive graphics cards with open-source drivers, it'll force nVidia and ATI to re-think - but for now, taking users hostage and obstructing nVidia supporting them better is hardly helpful.

I can see quite a lot of use for this in various scenarios. The obvious one is migrating to another cloud, private or otherwise, but there are others. Testing and development, for example: pull the image of your virtual server down to local hardware, run tests - some of which might be expensive or impossible 'in the cloud', like trying attacks that would upset your host if done on their systems, simulating packet loss, component failures...

OK, you could go and build a "matching" system ... Server 2008 ... current patches - region settings? File system parameters and cluster size? I've had a third party web application fail when deployed to EC2 - one dependency on the locale had crept in, so on the (UK) test server it worked fine but on the (US-built, though running in Dublin), one operation failed. Having a test server which was genuinely a clone of the target would be quicker and easier than hand-building a duplicate, as well as replicating it more accurately.

Within minutes of having to handle a BB for work, I wanted the company to die a painful death. The pointy little keys digging into my fingers, the clunky menus obviously scrounged from the "MS DOS Shell" (back before Microsoft were sure these new-fangled GUI things would catch on) ... that turkey just wasn't designed to survive in the 21st century.

As for security ... my mail all comes over SSL straight from my servers, and has had very very few remote exploits of any kind ... would diverting it through an extra potential point of failure really help at all? If and when RIMM folds and the servers go quiet, the handsets are little more than paperweights - any standards-based handset will keep handling IMAP/SMTP mail quite happily regardless.

I was with Nildram, and loved them .. then they became Pipex, and the slide started. Fortunately, I bailed before they reached rock bottom in TalkTalk, the packet graveyard; a happy two years with Entanet, then over to a C&W/ex-Bulldog reseller when Entanet fell off BT's "21cn/WBC" cliff.

Now I'm with Virgin. Nice fast downloads, most of the time, decent upload, with staff who can't even spell IP and a few irritating limitations. All my outbound mail goes over port 465 to another country, though, so it doesn't matter what they screw up SMTP-wise!

Tethering

Personally, I jailbreak so I can use tethering - being on Giffgaff with a 'Gigabag', some bizarre three-way blamestorm between Apple, O2 and Giffgaff means the official tethering route doesn't work. (iOS can't tell the difference between Giffgaff and O2's own PAYG service, which is set not to allow tethering.) I'll probably switch networks in another 2-3 months once my current credit's used up; Orange should only be about £1-2/month more expensive for me, and a bit more flexible. (Can't combine goodybags and gigabags on Giffgaff, for some stupid reason.)

I just wish the ASA would shoot down the "fibre" lie that Virgin's copper coax cables are somehow "fibre" because there's fibre as far as the copper conversion point, unlike ADSL which only has fibre as far as the copper conversion point...

The traffic shaping also makes a big difference - though at least the traffic limits seem high enough to be acceptable, and only kneecap you for the rest of the day instead of costing you a fortune or wrecking the rest of the month.

I remember work rolling out a new management system (an optional part of our existing mail server suite) for non-BlackBerry handsets a while ago. The upshot was that everything could connect to either BES or the non-BB equivalent - the big difference being that the BES cost money over and above the mail server, making BB handsets more expensive for the same job.

Now mail is outsourced, non-BB handsets can all connect natively, BB kit still needs its dedicated babysitter in the middle at extra cost. It's a big enough, diverse enough workforce there are people around who know each major platform well anyway, so BYOD shouldn't entail any extra costs anyway.

Besides, most non-enterprisey kit is designed and supported by the manufacturer/vendor for self-sufficient use. For email, feed in your IMAP/SMTP/ActiveSync credentials and away you go. Virtually all our BB support issues amounted to the BES having fouled the nest anyway...

Mainframe/PC deja vu

It's deja vu all over again - I can see exactly the same arguments a few decades back, that nobody would advocate replacing the million dollar three-phase monster with the six figure IBM support contract with mere 'personal computers', it would be career suicide ... meanwhile, the bean-counters were discovering they could get the answers they needed faster on a PC in 123 than from the mainframe. Suddenly, somebody notices the mainframe's not actually being used any more: the PCs have taken over.

At work, we had a six figure SAN with a team of acolytes tending to its every whim, sacrificing goats as ordered in an attempt to keep the clunky 80s email system grunting along. Chucked now, even Microsoft can do a far better job of email than that fiasco, for a tiny fraction of the cost.

Yes, a giant SaaS provider needs to address ... actually, exactly the same problems any other very large outfit would do, which makes it ironic they are capturing the smaller users first.

To a small business, it's often a choice between an SaaS provider's clustered, replicated servers or an elderly Dell out of warranty. Of course the SaaS provider can be much, much more reliable! In a big company, they probably run to multiple servers, some load balancing or failover, decent backups - but 24 hour staffing? We don't, with nine figure turnover and an IT dept headcount well in double figures: it would cost too much. End result is that almost any commercial service is likely to deliver lower downtime, just because problems have to wait until 9-5 M-F to be fixed.

In related news, Sony have given Betamax an exciting new facelift! New shiny buttons to press!

They really do have the same problem Betamax did against VHS, but with the handicap of being late to the party. DVDs had an uphill struggle to unseat VHS, for exactly that reason: why switch, when VHS is so entrenched? In the DVD's case, of course, there were a string of real advantages: better picture quality, no rewinding, more reliable media, more physically compact - so it actually worked. To unseat Facebook, G+ will have to become not just as good, not just better, but so much better than Facebook that people actually switch - and that's exactly what Buzz tried and failed to do with Twitter.

Do the Wonkas really believe they've bet their ad company's future on out-Facebooking Facebook, or will G+ go the way of Buzz, Wave and the others in a year or so?

(Of course, Facebook could park their tanks on Google's lawn, starting to do serious well-targeted ads on third-party sites like AdWords - probably Google's big long term fear, but do Facebook want to go that route?)

The current asynchronous system, of employers paying and deducting semi-random amounts each month, occasionally being told by HMRC to deduct different random amounts from that point on, then trying to balance it all up afterwards for each tax year, is stupid. "Tax codes"? WTF does HMRC have to invent new gibberish like "D0", "BR", "L8500"? The new system sounds like a huge improvement: they say "we are paying Fred (NI#AB12345C) £1500 this month", HMRC say "OK, pay us £700, pay him the other £1000" (because they like to pretend part of the tax comes from the employer), that should be the end of it.

No doubt they'll screw the pooch horribly, blowing billions on a system to get the answers wrong then billions more to fix it, but it's a nice idea.

Trademark should never have been granted for that

If I start selling pre-arranged rockeries for gardens under the name Stonehenge, do I get to sue Google any time someone else calls their rockery Stonehenge? That's nuts. They aren't even the first people to use that name *for that class of product*, let alone first to use it at all!

Weak argument

I don't like Sun/Oracle's "field of use" restrictions - surely, if I want to play Pong on a Cray or shoehorn Ubuntu onto an old 386, that's my problem, whether it works or not?

Having said that, "it's freely available so you can't enforce copyright" is a really, really dumb argument. Will they be redistributing modified copies of the Windows 8 preview release? That's free, too, but I suspect Microsoft might just get all lawyered up about that idea! Even open-source/free-software doesn't mean you can do what you like regardless of licence - the FSF of all people should be on board with that distinction.

I hate the idea of software patents and of trying to restrict use of a protocol or interface (look at the Linux kernel fiasco with BitKeeper?) - but it does look as if (a) the room wasn't quite 'clean' after all (Sun stuff slipping in) and (b) Sun/Oracle have patents which cover things Android does, at which point no clone, clean-room or not, is safe.

With hindsight, and given their love of Java, maybe Google should have bought Sun themselves...

Vested interests feeling "threatened"

No surprise there - remember where the term "Luddite" actually comes from? In related news, Tube "drivers" are also strongly opposed to being replaced by machines...

If they really don't work reliably, and can't be made to do so, that's different - as it stands, we have a story about turkeys not liking the smell of stuffing or the wrapping of presents.

The "same playing field as the rest of us" from a rival and Microsoft's "like thousands of other partners and utilizing our standing pricing and licensing terms" seem to indicate the real issue: OnLive is NOT just another startup doing VDI, it's one with close ties to MS.

Microsoft's dilemma is that if they admit to giving OnLive a nice special licence exception for VDI, all their other customers/partners will either demand the same, or cry foul - and I'm sure the last time that happened is still etched in minds throughout Redmond as something to be avoided like the plague. We'll probably see them buying OnLive and rebranding the service as their own - but as a new service like this, an agile "startup" environment probably works better than being one small part of a giant corporation. The thousand-strong army of lawyers to check whether the shape of the letter T on the boot logo complies with Pakistan's calligraphy regulations is fine for billion-dollar product launches like Windows 7 - not so good for delivering a new web service...

They could and should introduce a bar on 0800 calls being any more expensive than any geographic call - i.e. your unlimited or 250 landline minutes must include 0800 calls too. There's no excuse for any telco charging more for an 0800 call than a landline one - particularly since the recipient pays them for it too.

I'd like to see revenue sharing banned outright except on 09 numbers; 0871 and 0844 were mistakes all round. Yes, I know some people have set up nice little earners screwing people through the phone bill - that's no excuse for letting them continue. The AC with 275 different 0845 numbers is obviously doing something very weird, but there's no reason it couldn't be done on 03xx numbers instead.

On premium rate numbers, Ofcom have actually proposed something very sensible (and long overdue): that calls be priced in terms of the surcharge on top of your regular call price. Instead of the current "calls to this number cost 95p/min from BT lines, but random multiples of transcendental numbers on other operators", just "this call is 90p/min plus your normal call charge". Nice and simple, and of course already in place for SMS.

The AC has the right idea - and the CDMA systems Verizon and Sprint use do that already. Make the identify a public key certificate - maybe injected over USB/Bluetooth. Order a new handset from O2, they ship it with your certificate preloaded; buy an unlocked phone and sign up for a "non-SIM only" plan, they email you a file to copy over USB/Bluetooth. Particularly with micro-USB being the new standard for charging, all it needs is to have the data pins connected to something with a kilobyte or so of non-volatile storage.

Apple already proposed almost exactly that, but certain companies replied "waaaah, we demand SIMs". Maybe something to do with having patents which get them royalties from anyone making or using SIMs?

Overvaluing football

"nobody can compete with the kind of prices Sky will pay for that "best content"."

Surely the whole problem there was that OnDigital could and did outbid Sky - by paying more than the football was actually worth? Ultimately, Sky doesn't pay for the football, Sky Sports customers do, via Sky. As long as I don't have to pay for it, I don't care: Sports viewers can pay whatever the price may be for sports, or not. Too expensive, nobody will pay, so no income; too cheap, not enough income, so the sports clubs have to cut spending or change something else. That's business, between them and their customers.

Ads don't bother me - paying through the nose for a channel or two does. My subscription to Virgin gets me a dozen channels that actually have stuff I watch - and yes, the ads make it a lot cheaper than it would be otherwise; the BBC, on the other hand, charge me far, far more per channel and produce less content that I actually watch. Delete them from my channel lineup, would I even notice?

I felt for a while that it could either be cut down to a simple enough service to be profitable from ads alone, building volume that way - the hosting requirements, done properly, should be trivial - or ... bolt on massive amounts of shiny tat in the hope eventually someone will buy it for more than you paid? It worked once, with ITV, but I can't see it happening again.

Having said that, DCT were only interested in the Genes Reunited bit anyway - the rest wasn't much more than baggage along for the ride. A shame they didn't try spinning it back off as an independent little startup to cut overheads and experiment some more; this approach looks rather like a Frankenstein job, trying to bolt the leftovers they have lying around from other commercial work to make FR bigger and shinier - the last thing it needs.

Having the BB10 handset looming will probably slash BB7 handset sales for the next six months, and I can't see Playbook sales exploding any time soon either.

I've always been impressed by the customer loyalty they've enjoyed - even after the massive outages - but the last surveys I've seen indicated huge numbers of current customers planning to switch brands next time they upgrade; at work, even when the choice was between a work-paid BB or paying for your own iPhone, a lot went for the latter, and work funds both now anyway.

As the AC says, they've grabbed a surprising market share in the teenage-texter market here, and seem to have a surge in places like South America and Africa. The trouble is, that'll be a lot more fickle - much less lock-in than when your corporation has bought into BIS and mandates BBs company-wide. For the heavy texting aspect, there are other companies doing Android handsets with very similar keyboards - which just leaves BBM, really?

Having seen both sides...

I've seen the good and bad on both sides. I've been in the central bit when a complacent little department discovers that actually, the £10/month we charged to back up a departmental server over the network wasn't such a bad deal after all, as they pay for me salvaging a dead NetWare file system - but I've also been out in the revenue-generating departments, chafing at paying 900% markups on domain registrations (£10 reg fee plus £90 overhead!), choking under 30 Mb quotas because the building full of giant plasma screens apparently can't afford storage any more.

If the central IT provision kept up with commercial levels of service, there would be little or no need for what the article calls "shadow" IT (and I'd prefer a less judgemental term, like 'devolved' or 'distributed' - particularly in an environment where departments generally have their own IT staff).

Give us slow and unreliable 30 megabyte home directories, with extra storage in units of 300 Gb at £5000 each, of course Dropbox gets new users instead. Blow six figures propping up a half-dead dog-slow Groupwise installation with 50 Mb quotas, of course users will end up forwarding to a Gmail account. (To be fair, that particular turkey finally got a late Christmas when we upgraded to Hotmail.) For far too long now, central IT departments have grown accustomed to delivering minimal service at stupid prices. "Shadow IT" is one way of delivering much better results; if "central IT" can't match it, maybe it's time to reconsider their existence?

I wouldn't be surprised if most of the power went to the screen/backlight in the iPad 2 as well. At 43 Whr with a battery life of 10 hours, that's a 4.3 W power drain: 2 W of backlight, 1 W for the comms chipset and another for the processor/RAM/storage, about half on each of those for the previous model, with the lower-res screen and humbler graphics and network connection?

It'll be interesting to see how the WiFi-only model compares, since that'll exclude the LTE chipset/radio but keep everything else the same. If it gets much the same battery life, we know the screen is the culprit; if it's better, with the same battery, we know it's the LTE chipset draining all that power.

Most of the time all the LTE bits should be idle anyway, though, so it shouldn't make a major difference anyway.

If the pattern smears form a loop - for example a rectangle - it could start at any of the four corners, or indeed potentially part-way along an edge, although that might be visible in the smear pattern.

Opening the case and hooking the flash storage up to an external reader should hold up in court - it's essentially what they've been doing for years with regular computers - but the storage may well be encrypted, as it is with the iPhone 3gs and later (though any computer it syncs with will happen to have a copy of the key, which could be handy): I presume Androids will do the same, which is why a little help is needed. (Part of the spec for connecting to MS Exchange using ActiveSync, isn't it?)

"Most of these folk would have been watching catch-up services, such as BBC iPlayer, on their mobile devices. Catch-up viewing accounted for 9.2 per cent of all viewing in 2011, up from 7.1 per cent in 2010."

That must be the bit which scares the ... out of them: they can only collect tax on real-time broadcast viewing, on-demand content is beyond their remit. I could easily swap my current 32" LCD TV (already hooked up to a computer anyway) for a second-hand 30" and stick to downloaded/streamed content: no more TV tax needed, so the savings alone would just about cover the monitor price...

The company running the actual network is MBNL (Mobile Broadband Network Limited) though - a joint venture of Three and EE! I'm not quite sure how that works in terms of the two sets of "separate" frequency licenses, with transparent roaming already in place, but it wouldn't surprise me to see Three grabbing a chunk of this lower spectrum for MBNL to use while EE contribute the huge higher frequency bands.

Personally I'd give the 3 bands to O2, Vodafone and Three respectively - in exchange for coverage commitments. I'd rather have a better network for everyone to use than squeeze more cash into the treasury - but then, I'm not in charge of spending the contents of it...