A few years ago I found myself saying ‘with the iPad everyone flys first class’. My thinking at the time was that Apple had levelled the playing field by creating a device that delivered the same excellent user experience whether you were the CEO of a global company, a regular user, or a kid who’d sold a kidney to buy one.

At the time there was a great deal of attention being paid to iPad’s in the boardroom and C-suite, but the point was that there was bottom up demand for tablets in the workplace, not just top down. This was brought home to me one day whilst I was waiting to meet some vendors – I was checking emails on my iPad 2, and the tea lady was telling me how she’d queued up to get an iPad 2 from the Oxford Street Apple Store on the UK launch day.

In light of my recent posts about abandoning Apple and what people are doing with cheap tablets I thought it would be worth taking a look at the tablet class system…

First Class – iPads

The iPad is still the first class option, and despite my personal reservations regarding iOS 7 it’s still much loved for the user experience it delivers.

With the top end model still below $1000[1], and an entry level of $299 I think it’s still an egalitarian first class – you’d do well to buy first class flights to anywhere for that sort of money.

Business Class – Nexus etc.

I’d consider most of the top brand Android tablets in amongst this group, though there’s not much out there that meaningfully adds to what’s in the Nexus lineup – perhaps the Galaxy Notes. The Kindle Fire range probably belongs here too, though I haven’t had the hands on time to confirm that to myself. One feature that I would call out here is the option to have 3G/LTE connectivity.

When you fly business class you get all the important things of first class – lie flat beds for a good sleep on those night flights, priority boarding etc. What’s missing are the ephemera of luxury, and the price tag that goes along with that.

Tablets like the Nexus get the job done[2], they’re just not priced or marketed as luxury items.

Premium economy

There is an emergent bunch of tablets that are presently around the £119 price point that lack some of the sophistication of the Nexus etc. (e.g. no 3G/LTE option), but that are still pretty decent for regular day to day use. The Tesco Hudl looks like a good example, and I’m sure there’s similar stuff popping off the lines of a multitude of Far East manufacturers.

Economy

This is the land of the £99 (and below) tablet. They can all ostensibly do the same stuff as any other tablet – run apps, browse the web, play music, read ebooks, watch videos. They just do a terrible job of it because the hardware bill of materials has been pared to the minimum – beyond sensible compromises on user experience versus cost. Poor screens and laggy user interfaces are the cramped legroom of the tablet world. The Argos MyTablet provides a perfect example of everything that’s wrong here.

Updates

12 Nov 2013 – Joe asked ‘iPad mini?’ I replied ‘lumped in with the iPad. Consider screen size to be something like flight distance => iPad mini is 1st class short haul’. I probably should have figure out the screen size to flight distance comparison sooner, but it helps refine the model.

Notes

[1] I briefly got my hands on an iPad Air last week at London Cloudcamp, and it’s everything that I expected it to be, which is mostly pretty good. I can’t say I like the new smart cover – the dimensions just feel wrong having moved from a 4 fold design to 3 folds to adapt to the new narrower body.
[2] I’d argue that my Nexus 7 is better for my day to day workflows than my iPad. My HDMI adapter arrived today from China, which means I now have the ability to watch videos on hotel TVs – and it’s even smaller (and thus more convenient for travel) than the old iPad one.


This is Fred Wilson’s question, but I’m repeating it because it’s an interesting one, and because I think I have answers beyond glorified remote controls and recipe books.

Like Fred I was also intrigued by Ben Evan’s post ‘iPads, price and self-selection‘. Also in common with Fred I really like the Nexus 7 – both new and old[1]. I bought a bunch of last year’s model for various family members, and recently got a 2013 LTE one for myself.

Here’s what I see the Nexus 7s that I bought for other people being used for:

  • Kindle
  • Facebook
  • Gmail
  • Some very light browsing
  • Games when the kids get hold of them

Superficially that looks like a similar list to my own use:

  • Feed.ly
  • Twitter
  • Gmail
  • Facebook
  • Casual gaming
  • Browsing
  • Watching videos
  • Kindle

Looking at the lists one might argue that the tablets are used for much the same things, but if you were watching over our shoulders things would be very different.

I use my Nexus 7 online for hours every day – it’s a great device for curation [2] – a device for the 10% [3]. The digital footprints that I’m leaving with it will be at least as large as those I previously left with my iPad. I watch videos when I’m travelling, and I read books on Kindle when I’ve exhausted myself of online content, or just want to disconnect for a while.

My wife uses her Nexus 7 mostly for Kindle. Her digital footprints from it are likely very shallow. The other Nexus 7s in the family get used even less, their digital footprints are virtually non existent.

I think this explains the massive disparity between Android tablet sales, and Android’s share of page impressions across the web. People who’ve been buying Android tablets don’t use them online as much as iPad users, even the relatively high end ones like the Nexus 7.

I think cost comes into play a little here – if you’re using a tool a lot then you’ll buy a good tool. I happen to think that a 2013 Nexus 7 LTE is just as good (perhaps even better) for my own use cases than the equivalent iPad mini, so the price delta (which is almost all margin to Apple) isn’t justified for me. So Ben’s right in part – premium buyers self select in (even if the premium may no longer be deserved or necessary [4]).

There is however something else going on here. If the iPad is a device for the 10%, and cheap Android tablets are a device for the 90% then there’s a very skewed distribution of online engagement. The 90% don’t just consume – they consume a lot less than those who curate, and they in turn consume a lot less than those who create.

So the short answer to ‘What Are People Doing On All Those Cheap Tablets?’ feels to me like ‘very little – certainly a lot less than what you’re doing with your iPad’.

Notes:

[1] I’ve heard from some friends that they tried the original Nexus 7 and didn’t get on with it. My own (somewhat superficial) experiences were all good, but I didn’t have my own and work it hard. I’ve since read that there was a problem with the storage early on (since corrected) that caused major performance issues for heavy users. There were also reports of WiFi issues. To those who were put off I’d say try again (after updating to latest firmware etc.) and certainly give the 2013 model a go.
[2] It was Fred that brought my attention to the 90/9/1 rule ‘Out of 100 people, 1% will create the content, 10% will curate the content, and the other 90% will simply consume it.’
[3] I’m writing this post on my laptop rather than a tablet because even with Swype, speech recognition or a bluetooth keyboard tablets are *much* less effective for creative work.
[4] Though the Nexus 7 is still a premium device. The $349 I paid may be less than $629 for an equivalent iPad Mini, but it’s still a whole lot more than the cheap devices that are now knocking around.


I’ve been a happy iPhone and iPad user for a couple of years, but the time has come to move on. It’s not that I dislike Apple stuff, I just don’t think it’s the best for my needs, and certainly isn’t worth the premium. That said, I’m not very convinced by iOS7, or OS X Mavericks for that matter.

Desktop

The new Mac Pro looks awesome – an object of pure techlust the likes of which I’ve hardly seen since the NeXT cube days. But I last bought a new desktop somewhere in the middle of the last decade.

Laptop

If Apple made an MacBook Air that took 16GB of RAM I’d have bought one already. The 8GB limitation pushed me to getting a Lenovo X230, which I’m very happy with (and was about $500 less than the equivalent MBA – and doesn’t make me carry a little bag of video adaptors).

Phone

I ended up choosing the Samsung Galaxy S4 mini over an iPhone 5 – the dual SIM version was just the icing on the cake. Cost wasn’t a significant factor in the decision, but £335 is a whole lot better than £709 for a 64GB 5s.

Tablet

My iPad hasn’t seen much use since I got a Nexus 7 LTE, and I left it at home when I was on holiday last week, and again for a business trip this week. I think cost was more of a factor here – the Nexus 7 LTE is $349 whilst an equivalent iPad mini is $629.

What I’m missing

I bought the TomTom apps for iOS some time ago, and it’s been great to get free upgrades over the years. I’ve yet to buy a new offline GPS app (like TomTom or maybe Sygic) for Android, so I’m still taking my old iPhone4 with me on trips where I need a GPS.

I had the HDMI adaptor for my iPad (that also worked with my iPhone 4), which was useful for watching shows on hotel room and holiday rental TVs (though I’d note that the HDMI adaptor for newer iPads that use the Lightning connector aren’t really adaptors, but rather mini network attached streaming devices – that actually reduce the quality of the output). I’ve yet to get a Slimport adaptor for the Nexus 7, and I probably won’t bother getting the slightly different sort of adaptor the S4 mini needs.

What I’m not missing

A whole bunch of crappy games that my kids downloaded.

A whole bunch more games that I downloaded.

How my world is better

Swype – probably the best 69p I ever spent. It totally baffles me that Apple blocks this kind of awesome UI improvement from its environment.

I don’t have to use iTunes any more (e.g. to get videos onto my tablet before a trip).

I have enough storage on my phone for all of my MP3s and everything else.

I can share links to stuff on pinboard.in without having to jump through hoops (because of an open plugin architecture for such things).

Conclusion

Leaving behind the iOS world for Android has involved little pain, and a good amount of joy. Maybe I’d be just as happy with newer Apple gear, but I’d be somewhat poorer and I’d be missing out on a bunch of stuff.

Lastly, I should answer Fred Wilson’s question ‘What Are People Doing On All Those Cheap Tablets?‘ – for me it’s exactly the same as what I was doing with my expensive tablet – Feed.ly, Twitter, watching videos, Kindle, Facebook and some casual gaming. Perhaps more on that in another post – I think people who do a lot with their tablets are willing to spend more on them, but if a heavy user like me can be happy with a cheaper device then the landscape is changing.


…at least not in an embedded environment.

There’s a commonly held myth in modern software development that compilers are smarter than people at optimising code for its eventual runtime environment. By extension there’s no point in writing efficient code, because your idea of efficient code might not actually be all that efficient, and any time spent on doing that is wasted because the compiler will know better. This post is about one of those times when that’s not true. Functionally identical code can end up compiling into very different binaries, and that can lead to significant differences in resource usage on the target environment. That might not matter on a PC with a multi core multi GHz CPU and many GB or RAM, but it can make the difference between working and not working in an embedded environment – the kind of environment that the Internet of Things is being built out of.

The back story

A couple of years ago I did a simple project with my kids to make little boards that would flash their names in morse code. It used the cheap ($4.30) and popular TI MSP430 LaunchPad development kit. The code to do the morse bits was fugly but it worked.

Some time later my friend Andy Bennett asked me why I’d coded the project the way I had. The honest answer was that I’d been pretty lazy and cut’n’pasted the whole thing together, but for good measure I threw in a comment that I thought it would be easy for the compiler to digest and turn into an efficient binary for the highly constrained target environment of a low end MSP430 chip.

I’m still not quite sure whether Andy set out to prove me right or wrong, but I’m impressed by the effort and rigour that he put into his subsequent investigations. Some time later he presented his findings at OSHUG #25 (pdf slides).

The (more elegant) alternatives

Andy created three additional versions of the morse code flasher program (I’ve put all four into this Github repo – EmbeddedMorse). I think anybody with the slightest sense of smell around code would argue that all three are better than my original (#2). #1 and #3 make much better use of C data structures, #3 has some clever bit mapping of morse characters, and #4 uses a switch statement instead of many extra functions.

If the myth of compiler omnipotence held true then all four examples would have compiled down into the same binary for the MSP430.

They didn’t – it turns out that the compiler can be led astray.

A quick detour into a limited memory map

The MSP430G2211 device that I used has 2kB of Flash and 128 Bytes of RAM – so anything that’s going to change at runtime (variables) needs to fit into that 128 Bytes and everything else can be accommodated in the more generous 2048 Bytes of Flash. The actual memory map for an MSP430 device (or any other microcontroller for that matter) is a bit more complex, due to memory mapped peripherals etc., but that matters less when it comes to squeezing some code in.

Memory sections of compiled programs

A compiler will generate binary code to fit into various different sections such as .text (machine instructions), .data (static data such as constants) and .bss (space for variables).

For the compiled program to fit onto the device then .bss must go inside 128 Bytes and .text and .data combined can’t exceed 2048 Bytes.

Here’s what the compiler actually produced

Example .text .data .rodata .bss
1 391 832 156 0
2 1122 0 0 0
3 422 26 44 0
4 844 0 796 0

The first notable thing is that they’re all different – the compiler hasn’t spotted that the code is functionally identical, and it’s compiled it into different shaped binaries.

The second notable thing is that they all should work – none of the code uses any .bss, so that tiny 128 Bytes of RAM shouldn’t be an issue.

The only one that does work is #2, and it’s pretty obvious why that is – the entire compiled output is .text, which can go into flash.

It’s less clear why the other three examples don’t work:

#1 – if .data goes into RAM rather than flash then it’s clearly going to break past the 128 Byte limit

#4 – if .rodata goes into RAM rather than flash (which it totally doesn’t need to) then it will also break the 128 Byte limit

#3 – even if the .data and .rodata go into RAM then it’s only 70 Bytes, so there are 58 Bytes left. In theory that should be plenty, but in practice… some further analysis needed to figure out why it breaks.

Conclusion

Compilers aren’t as smart as people make out. Functionally identical source code can lead to very different binaries, and that can matter a lot in a constrained environment.

I didn’t know any of this when I wrote my initial implementation, and it’s just good luck that I picked an approach that worked first time. If it hadn’t then I might have done battle with the compiler for some time, or I might have given up and found an easier platform to get it working on.


I’ve been a big fan of Koss Spark Plug earbuds for many years. They’re cheap, comfortable and sound great. I got my first set after some very expensive Sony noise cancelling earbuds failed on my (just after the warranty ran out), and it was obvious straight away that a good snug fit was an excellent substitute for active noise cancelling.

As I find myself using Skype and Google Voice from my laptop (and my new phone) a lot for calls these days I’ve found that the built in microphone picks up too much noise, so I got the iSpark as an alternative. It’s basically the same headphones I’ve come to love with the Spark Plugs, but with the addition of a microphone and push switch. The microphone seems to work very well for calls, and the push switch is handy for pausing shows I’m watching on my iPad.

Although they’re marketed to go with the iPhone these things seem to work with anything that has a 3.5mm jack for combined headphones and microphone. They also come with a handy little carry pouch, so I’m more likely now to just pop them in my pocket rather than having them sat in the bottom of my bag.


TL;DR

The S4 mini is a fantastic piece of kit. The size is right, the screen is vibrant, the battery lasts, it works fine as a phone. The *only* thing I might ask for as an improvement would be a higher resolution screen. Dual SIM is very nice too.

Background

2 years is too long for a phone upgrade cycle, and I’d let it go even longer than that with my iPhone 4. It was frankly always a pretty useless phone, but it’s been a very nice pocket computer. I’ve started to need a phone that actually works recently, and ended up switching to my old San Francisco (ZTE Blade) as my primary phone – swapping the data only SIM I had for it over to the iPhone. The Blade was pretty satisfactory, but a three year old low end phone was really only good for emails and voice – I needed something newer for apps. I backed the Ubuntu Edge, but ultimately $32m was too much of an ask for the community – so that’s not happening. I waited for the iPhone 5s and 5c to launch, and decided I wanted neither (or perhaps I just wasn’t prepared to pay *that* much of a premium for an iPod touch with a modem in it).

Goldilocks screen size

I went for an S4 mini rather than the full S4 (or a Nexus 4) because my pockets just aren’t big enough. I generally have a tablet or two within reach if I want more screen real estate, so I want a phone rather than a phablet.

Enough performance

The S4 mini has a dual core 1.7Ghz CPU, which is a lot less than its big brother’s quad core 1.9GHz processor. It doesn’t matter – it’s plenty fast enough for anything I’ve tried to do with it. Everything feels snappy.

Relatively unmolested Android

If there was a Nexus version of the S4 mini Duos then I’d have tried to get it – I’ve yet to see anything that a manufacturer (or worse still an operator) has tried to add to stock Android that’s actually made it better rather than worse. Luckily Samsung don’t seem to have messed with Android too badly (or at least not badly enough to make me want to switch straight away to CyanogenMod).

Dual SIM

The ‘Duos’ naming means that the GT-I9192 can take two (micro) SIM cards. I’m finding this pretty useful for a couple of reasons:

  1. The most obvious is when travelling. I’m in the US right now, and have my T-Mobile PAYG SIM card in the phone so that I’m paying 10c/min rather than £1/min.
  2. Back at home I’m running a GiffGaff SIM in the second slot. Pretty much my whole family now use GiffGaff, and it means that they can call and text me for free (whilst I wait for my Vodafone contract to expire[1]).

The software makes it pretty easy to choose which SIM is used. One can be selected as default, but the phone dialer and text messaging app offer two buttons for dial/send so the right SIM can be used.

The one trade off with this model is that it doesn’t have LTE. I’m not that bothered as I don’t use any services that support yet.

Cost

The handset was £295 on eBay. Most of them seem to be coming from Singapore, but I paid a local UK supplier to avoid hassles with shipping and customs. Delivery was advertised as being 10 days, but I had it in my hands 2 days later :)

I’ve ordered a top of the range 64GB microSD card to supplement the 8GB internal memory (and 16GB microSD that I had lying around). It cost £34.99, which seems a whole lot more reasonable than the £160 premium for a 64GB iPhone 5s.

Leaving the iOS ecosystem

I’ve had feet in both camps (iOS and Android) for a few years now, so the new phone has been more of a transition than a switch. The only thing that I’ve left behind is TomTom. I bought the navigation apps for W Europe and USA & Canada a few years ago to use on my iPod Touch, and it was great to keep using them when I got an iPhone (along with a handy windscreen mounting widget from eBay). I may yet get TomTom for Android – I’m just not in a desperate hurry to re-buy £50+ of apps (and I’m also waiting for that 64GB microSD card so I have plenty of space for my music and apps).

Overall

I’m still in the honeymoon period, but so far I’m loving the S4 mini Duos.

Notes

[1] It was a mistake to sign up for another year with Vodafone just to get reasonable roaming rates in Europe. I’d almost certainly have been better off switching sooner to GiffGaff and putting up with whatever their deal is for roaming.


Now that OpenELEC 3.2 is out the dev team are moving to 3.3 (which will eventually become stable as 3.4).

The dev team have asked me to wait until things are more stable (for around 4 weeks) before doing more builds from the master branch.

The dev team have also made a decision to remove .bz2 compression. Let me know if you’d rather stick with tar.bz2, or go with the flow to straight .tar files (which are about 10MB bigger)?


Yesterday I wrote about how Google Wallet is even worse than PayPal. A quick reminder:

  1. When I buy stuff on PayPal I don’t have to pretend to have a billing address in a different country. They’re quite happy for me to use my UK issued card for payments to US suppliers.
  2. I don’t recall ever having a PayPal transaction cause an issue with Amex.
  3. The PayPal resolution process only required me to submit one form of ID.
  4. The PayPal resolution process completed same day.
  5. I’m not dependent on PayPal for Google services (like renewing Google Apps domains or buying apps or content on my Android phone/tablet).

The saga continues…. The good news is that it didn’t take 3-5 days to get a response. The bad news is that they’ve rejected my proof of ID as being inadequate.

They asked for a scan of my credit card statement with the first 12 numbers blanked out:

Wallet_first_12

This I did. As it’s an Amex card, which has 15 digits, this left 3 digits showing.

Now they email saying that they want the last 4 numbers visible.

Wallet_last_4

Make you minds up guys. If you don’t give me clear resolution instructions, then your response makes you look like dicks.

For extra irony I should point out that my full credit card number without any digits blacked out was at the bottom of the scan – where’s the harm – they have the whole number anyway.

There’s another possible twist here… As I used a supplementary card for the recent transaction I can’t possibly send a statement with that number. It just doesn’t exist – the statements only carry the number of the main card. (I’ve written before on the topic of additional cardholders being a weeping sore in the security model for online payments).

I’ll try a utility bill.

Congratulations Google – you’re the topic for my 300th blog post.

PS I should have added to my post yesterday that when I bought an iPad online from the US Apple Store and had it shipped to my friend in Virginia I had no trouble whatsoever (and Apple didn’t make me use iTunes to pay for it ).

Update 13 Sep 2013

Once again Google have replied within a day, but once again it’s not good news. They’re insisting on a statement showing the additional card number, which simply doesn’t exist:

wallet_must_have

I’ve (once again) resubmitted my documents, this time with the following comment:

The last transaction on my account was made using an additional card. The numbers for additional cards are not shown on my statement, and therefore I have no way of supplying the information you’re now insisting on.

I’ve done my best to comply with this process, but you’re now asking for stuff that I (or anybody else ever using an additional card) cannot provide. So we meet an impasse.

I have proven my identity. I have proven my address. Amex approved the payment (second time around after I’d been through their fraud prevention false positive hoops).

What now? Do I have to abandon my Google Apps and stop using Android phones because I can’t pay for anything with Wallet any more?

I have a Google apps domain that’s due for renewal next Tuesday (17th) and I have no way to pay for that. Suggestions?

I predict this won’t end well.

My plans to get a Nexus 7 LTE (and a Galaxy S4 mini) are now on hold – what’s the point in having Google devices if they shut down the means to pay for anything?

Update 18 Sep

No word from Google – the daily back and forth seems to have stopped.

It seems that I’m not alone having this kind of problem when buying Nexus products. It also seems that Google is systematically unable to do anything about such issues.

Update 22 Sep

I’ve called Google Wallet support twice in the last couple of days. The first agent (Neysa) told me to send in a photo of my additional card. This didn’t help. The second agent (Jamie) told me that she couldn’t do anything to help as it was ‘a different department’.


I spent Saturday manning the @BrightonPi stand at Brighton Mini Maker Faire showing off various projects with Gareth James. It was great fun, and I really enjoyed talking to people about the potential of the Raspberry Pi.

There were a few questions that got asked a lot… hence this post.

The projects

I was showing off 80’s joystick with 80’s arcade game, OpenELEC, Ladder Board and a Sous Vide water bath. $son0 was adding the RPi camera to his alarm project (and generally messing around with Scratch).  Gareth brought along his timetable and secret squirrel.

1. Where can I buy a ladder board?

The Raspberry Pi ladder kit is a great introductory physical computing project, and can easily be made by kids who’ve learned to do basic soldering. It’s available online from Tandy (though sadly out of stock at the time of writing).

2. What’s that game (being projected onto the wall)?

It’s Targ, and it should be available for download from the MAME web site ROMs page (though sadly the download link seems to be broken at the moment). The classic 80’s gameplay proved too difficult for many modern gamers, and nobody made it to level 3 on the day (or set a high score). Gareth and I were amused to see a succession of kids holding the joystick upside down..

If you want to run MAME (and other classic games emulators) on your Pi then the easiest way is probably PiMAME.

3. What WiFi adaptor are you using?

I use the Edimax EW-7811UN as it’s small, inexpensive, and works easily and reliably with Raspbian and OpenELEC.

Any more?

I think that’s it for the frequently asked questions. If you have any more then please ask in the comments.


I’ve seen a ton of bad news stories about PayPal over the last few years – here are just the top three from my payments tag on pinboard.in. I’ve even run afoul of their over sensitive fraud detection myself in the last couple of months (whilst trying to buy Club Penguin subscriptions for my kids whilst travelling[1]). Today’s experience with Google Wallet is much worse than anything I’ve personally seen with PayPal.

NOFORN. Google don’t want me to buy one of these. NOFORN

I want to buy a new Nexus 7 LTE (as my original Samsung Galaxy Tab 3G is feeling very old and creaky these days, and the 2013 Nexus 7 has received rave reviews). I want the LTE version as I know from using 3G on my existing tab that I need data on the go. Having just been launched, the LTE version isn’t yet available for me at home in the UK. No problem, I’m over in the US at the end of the month, and I can get it shipped to one of my friends that I’ll be seeing. No problem perhaps with almost any retailer except Google Play. I should have known better from my last dreadful experience trying to buy Nexus 7s from the Play store.

The trouble started with ‘no supported payment method’ showing when I went to check out. Google Wallet doesn’t straight out tell you that you need a US billing address to order in the US, but it doesn’t take much searching to find that’s the case. Time for the supplementary card dodge (I’ve used this one before)… I register another card with a US address (of the company I work for) and now I can place my order.

Payment is declined, as the hair trigger of Amex’s fraud detection has been set off. This is sadly pretty much the norm, so I called up Amex, jumped through their security hoops and resubmitted the payment through Google Wallet. All looked good.

And then Google suspended my Google Wallet account and cancelled the order for my Nexus 7.

I’ve submitted scans of government issued ID and credit card statements (in a process not dissimilar to the one PayPal made me go through), and I’ll now wait the 3-5 business days for the issue to be resolved. Hopefully it will be at the early end of that scale as I need my account back in good standing for a domain renewal that’s due next Tuesday (4 business days away).

So how’s this worse than PayPal? Let’s count the ways:

  1. When I buy stuff on PayPal I don’t have to pretend to have a billing address in a different country. They’re quite happy for me to use my UK issued card for payments to US suppliers.
  2. I don’t recall ever having a PayPal transaction cause an issue with Amex.
  3. The PayPal resolution process only required me to submit one form of ID.
  4. The PayPal resolution process completed same day.
  5. I’m not dependent on PayPal for Google services (like renewing Google Apps domains).

Looks like I should just ask my friend to order my new tablet and pay him good old cash.

Notes

1. I suspect that PayPal use IP address geolocation as part of their fraud countermeasures, and that I mess things up by using a mix of US and UK IPs. It’s worth noting here that Google seem to treat Amazon EC2 IPs as being outside the US, so you can’t even try to buy things from the US Play store by using EC2 as a web proxy (but they’re fine with Google Compute Engine IPs – so they get my 2¢).