Monday, May 22, 2017

Economics of a singularity

So, lately I saw this video on "The Dark Side of the Singularity".

For those not inclined to watch (though, I encourage it, as it covers some interesting points), to summarize:

In the video, the presenter, Joe Scott, draws a parallel with the infrastructure of New York city being 99% dependent upon horses at the turn of the century and 99% driven by cars. The relevance with today's society being the rise of automation (AKA robots) being predicted to shelve 30% of jobs within 20 years. With artificial intelligence and robotics advancing continually, the video argues that most jobs will become redundant.

It makes sense, right? If you own a company, do you want to pay someone to do a task that a robot could do quicker, without lunch breaks and welfare or even a wage? This is essentially the argument that Joe Scott puts forward.

There are some very important issues covered: the prevalence of jobs in the transport industry versus the fact that automated driving is essentially already available, in terms of parking assist, lane assist, etc... I'd agree at this point that things look really bad for those who drive for a living.

If left unchecked as the video suggests, then I'd agree that the resulting massive unemployment would be catastrophic, socially. However, considering random people on the internet are bright enough to work out that -potentially- this is going to cripple the social structure, I'm pretty sure the powers that be are unlikely to allow it to happen.


So, why am I so skeptical? These are market forces, right? The natural inclination of capitalism, which has the majority of the world's population under its grasp. Right?

Enter the spectacularly over-simplified chart of relationships between business, government, and the populace. Essentially, it tells you what you already know: you pay taxes, you work for a company (or own one), you have the option to elect your government. If you're unlucky, you're unemployed, if you're slightly more lucky, like in the UK, then the government actually pays you. Additionally, businesses pay the government money too, and are regulated by government, to some extent.

Great. So what happens if companies start automating? First off, they earn more money, and so pay more corporation tax. Next, they will expand to push out other businesses; those businesses that choose not to automate will be out-competed by those that do, and jobs will be lost as a result.

I've read some hypothesizing that essentially ends the story here by saying that essentially everyone that doesn't own a company becomes unemployed, and discontent fuels rebellion and then anarchy. In some scenarios even ending with the robots being at the top of the food chain. (Hell, in an anarchy situation, my money would be on the robots too...)

But, there are some elephants in the room that this theory forgets: (1) Supply and demand, and (2) government, and (3) the flow of money.

(1) Supply and demand is the easiest to figure out: if noone has a job, and noone is making substantial money, noone can afford to buy anything. If noone can buy your product, it doesn't matter if its being made by machines or people, your company isn't going to profit from it.

(2) Governments always get the pointy end of the stick of criticism, but -economically speaking- preventing major crises like the above is actually what government is for. Market forces and personal gain are exceptionally disruptive influences on populations, and businesses have indeed been responsible for some really terrible things. Companies do not provide law enforcement, they have little incentive to provide welfare, they are not generally interested in national security.

However, these are things that governments are really quite adept at. This is why they spend so much time making laws. Sometimes it certainly doesn't seem like this is government is for; sometimes it seems like they are striving for the opposite (I'm looking at you, Brexit), but a job is a job, and the government is in the job of order.

Having large sections of the population laid off is really bad for order, and the government will act to stop it. It will probably act late, but -in the end- the government is elected by the populace, and when the populace collectively decides it wants something, the government will start re-orientating itself to act accordingly.

(3) The flow of money is an important issue, because point (2) isn't terribly convincing given how government is portrayed. Governments are corrupted by business interests, etc., right?

Perhaps. But how important is business to the government as a whole? Well, financially, not so important as you might think. The government gets most of its power from the people, in the form of taxes. This will differ by country, but some figures I found put the proportion of taxes obtained from the population at 80%, versus 11% for corporations.

In this light, the population is far more important to the government than business, in the general sense. In a way, business is just a way of getting your people to earn more money to pay you in taxes.

So what happens if the above doomsday scenario occurs? Enter the spectacularly over-simplified chart of the impact of automation on the relationships between business, government, and the populace.

Assuming there are still enough people earning money to buy products, the companies might earn more, and thus pay more taxes. However, the government will make a net loss, because all those people that were laid off no longer paying taxes. If you're country is benevolent enough to provide unemployment welfare, then the government is essentially screwed on the path to self-destruct.

Perhaps I have too much faith in government, but the idea that a nation would let itself walk blindly into this situation seems absurd. Either the government will step in to regulate companies, or it will have to provide some kind of universal welfare.

Both will probably happen. Joe Scott (from the aforementioned video) states that the mindset in the US is not one that embraces something-for-nothing welfare; however, there are countries that do provide for their citizens quite readily. The only problem with this, of course, is that a country that automates will become much more productive, but this can be offset in countries less inclined to automate by introducing import charges to prevent foreign goods becoming too cheap.

In my opinion the biggest problems we have to deal with in the event of "singularity" style mass automation are:

(1) Putting (a) welfare or (b) regulatory systems in place in advance of unemployment becoming a massive problem.
(a. cont.) Trying to relocate the lost jobs into other sectors and, if this is impossible*, finding something for the unemployed to do**
(b. cont.) Becoming too insulated by taxing imported goods, and falling behind in productivity to less-regulated countries

That said, I've evidently simplified considerably with the above, and it seems quite possible that automation could lead to crisis within any given country. Since governments tend to be fairly slow to respond to paradigm shifts, I think its fairly likely that crises will happen. However, I don't believe that the situation is completely stark; there are actions that can be taken to mitigate against automation, and perhaps even integrate it into society. Hopefully, we can come out the other side better off overall.



*Eventually, it should be expected to become impossible, because if we invent AI that is as clever as ourselves, or more so, then essentially there are no jobs. It would be handy at this point to make ourselves smarter so that we can out-think machines and thus still be useful for something. As Elon Musk points out, this also mitigates against our robots taking over, if that happens to be your concern.

**Sure, not having to work sounds OK, but the whole human experience is oriented around having tasks and providing a future for our children. Humans have not evolved to cope with having nothing to work towards***.

***Hell, even the ancients went so far as to invent things to work towards in terms of the afterlife! Though, people quite happily retire. Perhaps we should view this as the human race's collective retirement?

Saturday, April 1, 2017

[FIX] 7zip and Trados 2015 Mojibake / characters displaying as boxes

So, this is a simple fix (or more accurately, a workaround) for a couple of tools I use a lot in translation, Trados 2015 and 7zip.

Symptoms

In Trados 2015, you may find that everything displays fine when you insert the source text into the translation field, but when you actually start typing over the translation, a seemingly random set of characters in that sentence are replaced with boxes (e.g. something like に対して becomes に口して)

In 7zip, you may find that filenames inside zip files are completely illegible: for example, "<原稿>英訳用%u3000申請リリース案"  becomes something like "üâî┤ìeüäëpû¾ùpü@". (This is particularly problematic if you want to send the files back to the customer with the same filenames as you received them!)

Fix/workaround

A simple workaround is to download and install locale emulator, which lets you right-click the shortcuts to 7zip or Trados and select "Run in Japanese", which fixes the above problems and -so far as I can tell- has no other significant impact on the performance of the programs at all.


One problem on Windows 10 (and probably anything 8 upwards) is that you can't use the RMB to access the locale emulator from the start menu; however, locale emulator lets you make a shortcut that essentially performs the above automatically:

First, right-click your shortcut (to Trados or 7Zip, or whatever else you are trying to fix) and click "Open file location" (for the win 10 start menu, right-click, then select "More"->"Open file location". Typically for windows, this will only display the shortcut-containing folder.. You will need to right-click that shortcut and again select "Open file location").

Next, in a new window, find the locale emulator folder (the folder you unzipped from the earlier download) drag the application file onto locale emulator; this gives you an option "Create shortcut". By doing this, you make a shortcut (on the desktop) that will open the application in your locale of choice (for me, Japanese), even better, this shortcut can sit in the start menu without any problems (RMB on shortcut, "pin to start").

What's the deal with this problem?

The problem evidently is one of the OS locale, given that the fix is a locale emulator. Consequently, you should theoretically be able to fix the problem by changing your OS locale; however, there are a number of reasons you might not want to do this. Personally, I use Windows 10's Cortana assistant, and I like to do that in my mother tongue. Having system menus in English also helps speed maintenance issues.

With regards to 7Zip I'm not expert as to what restrictions exist with filenames etc.; however, for Trados I'm a little more skeptical. By its premise, Trados is a multilingual tool, and yet it handles the issue of OS locale quite poorly compared to non-specialist tools such as Microsoft Word (which admittedly has the benefit of being designed by the same company as the OS itself), or most web browsers.

Another workaround

This leads me to the other workaround I tried, which is a translation tool called Felix that works right inside MS Word. Because Word doesn't have any problem with locales, Felix doesn't suffer from the above problems. Felix isn't without its problems of course, and using Felix in a workflow that requires the use of Trados package files will add an extra layer of frustration. Nonetheless, Felix is probably my favorite tool after Trados for translating (and it very nearly tops it due to its ease of use).

Wednesday, June 22, 2016

Turn a smartphone into a microscope for free?

A while back I found a neat idea for turning a smartphone into a microscope. The idea was you 3D-print a clip to house a minute glass bead (around 1-3 mm across). This essentially acts as an objective lens, considerably magnifying the object in front of the lens (supposedly up to 1000-fold, though I remain sceptical).

Problem is... who has glass beads lying around?

As a nice hack, you can get a little way towards recreating the effect with just a drop of water or oil right in front of the lens (best do this on an old phone, people).



1) Kyocera KC01 (budget JP smartphone) with a blob of water over its lens (circled red to give an idea of the extent of the water droplet). The droplet is a little too large and slightly off centre, but illustrates the general idea.


The problem with water is that it evaporates very quickly, and so the magnification strength is lost very quickly. Vegetable oil was much better; I made one application and it lasted for several dozen minutes.

So, demonstration photos:
2) Photograph of thyme leaf broken at the petiole (leaf stem), showing detail of bristles. Taken using a Aquos Crystal with a droplet of vegetable oil approx. 1 mm across. Backlit using a single 100-yen-shop LED light.



 
3) Photograph of thyme leaf, showing detail of spots on the leaf. Taken using a Aquos Crystal with a droplet of vegetable oil approx. 1 mm across. Backlit using a single 100-yen-shop LED light.
For (2) and (3), Much of the blur around the edges is due to the droplet not filling the entire field of view. The clarity isn't exactly great elsewhere, possible because it was a real pain trying to keep my hand still.

The thing I like most about (3) is that I didn't notice those spots until I looked at the magnified image, so even these relatively poor images do help bring your awareness down to the micro level.

4) Photograph of human hair against paper. Taken using a Aquos Crystal with a droplet of vegetable oil approx. 1 mm across.

5) Photograph of Japanese bill at minimum focal distance of Aquos Crystal without modifications. The letters in the white area are around 2 mm across.
6) Photograph of Japanese bill taken using a Aquos Crystal with a droplet of vegetable oil approx. 1 mm across. Backlit using room light (approx. 3000 lux, according to lux meter)

I took (6) to get a rough idea of the magnification. On my computer screen the character extends 7 cm across my display without too much blurring (though, this may be more of a problem of my hand shaking, rather than the optics), which would suggest a magnification of around 70 mm / 2 mm = 35 fold. However, I should also account for the fact the unmodified camera produces characters of about 1 cm across on my screen, which in itself corresponds to a magnification of 10 mm / 2 mm = 5 fold. So in fact, the oil droplet was only 7 times more powerful than the unaided device.

Nonetheless, the patchiness of the printing is a feature that is almost impossible to detect with the naked eye, and yet became very prominent on using the oil droplet.

But, at only ~7 times better than the standard smartphone lens, I guess dishing out for a cheapo magnifying glass would give better results.

Unsurprisingly, seems people have already thought of this approach, here. Seems that by using polymers instead, you can bake them into permanent lenses. Cool!

I'd be thrilled to hear of any improvements to this. My first thought is to attempt producing a bead by melting clear plastic to flow along a wire into some cold water, which should hopefully solidify the bead in a... bead shape. I don't expect it to work, but it would be better than buying a box of 15000 beads and having 14999 spare beads...

Unless I can think of a use for 14999 glass beads...

EDIT: So, some obvious improvements:
1) You can buy a screen protector -they are clear and don't interrupt with the light getting to the lens too much-, which means you can use oil without worrying about damaging the lens.
2) You can apply PVA glue to this screen protector. When it dries, it becomes transparent, and you can indeed get a lensing effect from it! Decidedly more of a permanent solution that water and oil..

Saturday, January 16, 2016

Swiftkey vs Google vs Swype for Japanese input

Summary:

SwiftKey
  +Fastest for Japanese input
  +Switching languages is easy
Google Japanese input
  +Almost as fast
  +Excellent dictionary
  -English support is poor - but you can switch input methods
Swype
  +Still excellent for handwriting input - possibly useful for beginners to Japanese
  -Considerably slower for Japanese

So, every once and a while, I look through the Google play store and try and find a better way of typing Japanese on my mobile, because frankly, I do a lot of that, and a couple of minutes investigating which method is fastest will pay off in the long run.

Previously, I compared the Swype, Google, and Wnn keyboards. Since then, I've changed phones twice (once to an Aquos Crystal, then to a Kyocera KC01). The increased processing power of these more modern devices makes switching keyboard methods easier, so for a long time I was switching between the Google methods, which still provide excellent predictions for both languages*.

More recently I have been using the Swiftkey keyboard, and despite the fact they have done away with the 12-key keyboard (which works really well with Japanese input), I have increasingly found that it is quicker than Google Japanese Input.

To be sure, I conducted a small test using the following text:

"17日は前線を伴って低気圧が九州に近づくでしょう。九州は昼頃から雨が降り始める見込みです。中国、四国も午前は広く雨で、夜は近畿でも雨が降り出すでしょう。東海、北陸、関東も夜はところどころで雨や雪が降る見込みです。東北と北海道は日本海側で雪のところがありますが、晴れ間の出る所もあるでしょう。18日月曜日の朝は関東の内陸部でも雪の積もるところがありそうです。"

I should probably have used a shorter text... My thumb aches... But this should hopefully make differences between the different methods more apparent.

So, this is how long it took to type with each method:
6:06 with SwiftKey
6:20 with Google Japanese Input in 12-key format
7:49 with Swype keyboard in qwerty format, tapping, not swyping.

These results tally fairly well with my more qualitative observations; Swiftkey is only slightly better at Japanese than Google, but both of these methods are superior to Swype. I would point out that swyping is still not particularly usable for Japanese, but the word prediction is just about sufficient to make tapping viable.

As well as being the fastest method for inputting Japanese, switching to English in SwiftKey is faster than changing input methods with Google (again, footnote*), and the English prediction for SwiftKey is fairly capable. Nonetheless, for words with unusual characters, my gut feeling is that Google performs much better. Additionally, my main gripe with Swiftkey is that it is very easy to mis-press the vowel extension "ー" key instead of "a", and SwiftKey's programming isn't intelligent enough to reinterperet the extension as an "a" depending upon the rest of the word. Then of course, Swype has an excellent integrated handwriting recognition system, which might be worth it for those who need to input characters they don't know the pronounciation for (I imagine, however, that this is mostly going to be in the context of looking words up in a dictionary, for which aedict offers a fairly good method that isn't keyboard-dependent. Plus, free handwriting alternatives now exist).

The bottom line, however, is that SwiftKey is faster and more convenient when switching languages, and for me, these are the deciding factors.

*As I've mentioned before, despite being able to type English using "Google Japanese Input", the prediction is awful, and so I rely on changing input methods.

Sunday, January 10, 2016

The Tokyo Train Tech-crusher, and lightening the load for presentations with the Amazon Fire TV Stick

Sometimes I have to make presentations. Back in the UK, this was never a huge problem; even on those occasions when I had to take the train, the trains were always roomy enough that I could find a safe spot for my bag.

Having started work in Tokyo, the main problem I have with giving a presentation is the crush of rush hour. Usually, its at least tolerable; however, on occasion, often due to a delay of some sort*, people will cram themselves into the train until the passengers become compacted into a solid wall, making it impossible to add further bodies. Personal space is limited to whatever clothes separate you from your neighbour... Then again, the jostling, elbows and corners of bags poking into me almost makes it feel as though my personal space is limited to my head, which tends to rise at least a little way into the little breathing space that is left at the top of the carriage**.

The first few times I was caught in this situation, I found my bag pulled off my shoulder as people flowed around me from the door like sand filling every gap at the bottom of an hourglass. Another time, the zip of my bag got itself caught on the belt straps of some poor salaryman's trousers. Some simple solutions to this is to have a backpack (usually worn at the front, presumably to keep the nearest person in front at least a bag's width out of one's face), or to press your shoulder-bag hard into your front. There are luggage racks, but count yourself lucky if the Brownian motion of people into the carriage happens to afford you a space near one (and lucky again if you find a space to put your arms once you've finished stowing your bag).

If you have to take something fragile with you, you certainly have a few options, of course. Firstly, you can try leaving earlier, and simply wait for a carriage that doesn't happen to be packed. This is particularly good if you don't need to be somewhere during the rush hour; jams during the day generally are due to accidents etc., and tend to clear up before too long. If that isn't an option, you might reserve a seat (if there are green cars on the train). Otherwise, you could just bring a reinforced case for that laptop.

The latter option strikes me as the one that offers the greatest flexibility and convenience; however, laptops are heavy enough, without buying bulky casing, and the increased size of the baggage is also bound to be another headache. So, instead of thinking bigger, stronger, better, I decided to go the opposite direction.

The Amazon Fire TV Stick is essentially an Android computer that plugs into an HDMI port. I was at first put off by its name, which sounded less functional to me than the "Chromecast" by Google; however, the Fire Stick has one very potent advantage for what I had in mind: onboard apps. While chromecast would probably be my main choice if I intended to stream media from one of my smaller devices to a bigger screen, I don't own (or desire) a TV***, and given my experience with wifi connections, I suspect that if the system functions better in isolation, it will be more practical for applications on-the-go.

Of course, my second-hand monitor has no HDMI port, so I had to buy an HDMI-VGA adapter to actually ensure the presentation setup works beforehand (the adapter works surprisingly well, and has audio output too, just in case I want to use the Stick for recreation). On the bright side, this will come in handy if I happen to present somewhere without HDMI.

The next obstacle to abusing the device into doing work, rather than play, was the lack of any decent office software on the Fire Stick's app store. Luckily, there is plenty of information on how to install standard Android software, WPS office being my suite of choice.

Excellent. And WPS office plugs straight into Dropbox, letting me pull down presentations in no time. The problem, however, is that WPS office requires a touch screen or mouse. I managed to get as far as linking WPS with dropbox, downloading, and opening a presentation file, but the Fire Stick's remote just wouldn't let me select the "play" button. There are a number of mobile-to-TV apps that I could have used to solve this problem; however, I wanted to be sure that the device will function even if my phone batteries die and the wifi has exploded. This is where Mouse Toggle for Fire TV comes in. It lets you double press the play button on the remote to summon a cursor that can be controlled by the up/down/left/right keys on the remote. Not exactly elegant, but better than buying a bluetooth mouse (after all, I won't actually be producing any content on the Fire Stick*4), and the remote will double as a slide-changer which I can use without being tied to a laptop.

So, I found myself in a position where I can actually give a presentation. Lovely. Especially considering the entire kit could fit inside my pocket. The only problem is that the Fire Stick wont show non-native apps in its main menu. Every time you want to open the app you have to go to Settings>Applications>Manage Applications>[Application name]>Launch. Ye gads. The Fire Stick isn't the sprightliest of devices, and add in that you might need to enter a pin to use the Manage Applications menu (depending on your security settings), this takes too much time; what if I need to switch applications, perhaps to open a paper I reference in the presentation, in response to a comment?

Thankfully, there is a friendly group of arsonists programmers that have developed "Firestarter", which is basically an app tray that loads automatically, leaving the original system intact, but doing away with its annoyances. Now, I can switch between apps at a reasonable pace, and set up in a much more reasonable order of time (you can download it here).

Of course, I can control the device using my mobile, if needs be, and it also supports miracasting; however, in my opinion its ability to run as a standalone device is its greatest strength. Even if it needs a little persuading.

On the day of the presentation, I was happy to have a lighter load on the 1-hour, standing train journey, and not have to worry about the device imploding (the box it comes in is fairly sturdy). Moreover, set-up was simple using the seminar room's HDMI input, and the device was responsive enough for a smooth presentation.


*Japanese trains are generally pretty punctual, but I think the massive throughput of the Tokyo railway system means even a small delay can cause a pretty jam. Longer delays are nightmarish.

**I regularly complain about how annoying it is being tall, and its rare I get through a day in the city without knocking my head on a hand-strap; however, when the big squeeze occurs, I'm only too happy to have my head as far away from the seething mass of shoulders and elbows below. I can only imagine what that must be like for a shorter person...

*** Yeah, Amazon Fire TV Stick... Ironic, I know. I should also point out that I don't think the Chromecast is bad; for a home set-up or anywhere you have good control over possible show-stopping factors, the Chromecast would be an excellent (and marginally more economical) choice.

*4 That's what proper computers are for... Though I find it ironic that I have a beast of a machine in the laboratory, and giving a presentation just a few feet away requires a far less functional device.

Tuesday, December 29, 2015

Comfort Women and Justice? And revisionism!

From what little time I have for the news, I gather that, for some time now, the Japanese government has been asserting that it will clear up the issue of comfort women with Korea by the end of the year. Personally, I applaud this effort to try and reconcile differences between Japan and Korea, and I hope this marks a greater shift towards a deeper understanding between Japan and its neighbouring countries.

However, I have a couple of issues with how this reconciliation is being approached. The short version is that the approach of both governments is fundamentally flawed.

1) The necessity for financial compensation.

It makes sense from a legal perspective that victims be compensated for the traumas (in this case, quite horrific ones) incurred. However, who should be paying that compensation? The issue is such an old one now that it is no longer possible to round up the responsible people and impart justice.

From the agreements reached so far it looks like the Japanese government will be paying compensation, which superficially seems fair: it was the Japanese government that occupied Korea, right? Well, I have problems with this.

Firstly, the Japanese government now is nothing like the organization that rampaged around East Asia during the war. To my mind, it is a separate entity, established largely by the US when they demolished Imperial Japan and led them by the hand to a truer democracy*. Since then, it has evolved yet more. As such, I think it odd that modern Japan be held directly responsible for the compensation of the comfort women.

Secondly, where does this money come from? The government gets its money from taxes, which are paid by working individuals in Japanese society that are entirely innocent. Sure, the figure per head is going to be infinitesimal** (7.8 yen, something like 4p), but to my mind this is foundationally flawed.

2)Japanese revisionism

Just so we're clear, it's not just the Korean side I have issue with here, the Japanese side is equally flawed. It is plain to see on the news, Abe wants to "put an end to the issue before the year is out". Why the rush? Is this really the best premise for discussion? He may as well announce "Let's just get this over with quickly so that when the new year comes we can forget all about it". Anyway, part of the initial terms drawn up by Japan was the removal of a statue erected outside the Japanese embassy. This statue symbolizes the pains inflicted by Imperial Japan and a defiance, which I applaud, towards Japanese indifference to the issue.

I think part of the issue here is that the statue hurts Japanese honour. But let's be honest, it should do. The actions of Imperial Japan are a shame to the name of Japan, and it is by overcoming that shame that Japan should move forward; the Japanese government should be working towards building a society where past lessons are learned and can never be repeated. Instead, Japan has focussed on this shame as something that needs removing, rather than a harsh lesson.

I feel slightly alleviated by the fact that Korea haven't agreed to this term yet; however, I am somewhat disappointed that revisionism isn't the main thrust of Korea's discussions. If you're going to hold modern Japan to account, at least do so for its own wrongdoings***...



*The barbarism the US displayed (admittedly blood was on everyone's hands back then) is at least partially offset by the spirit of cooperation the US managed to forge with their vanquished.

**1 billion yen divided by 127 million people. 7.8 yen doesn't sound like much, but 1 billion yen is. A lot of good could be done with that money; lives could be saved with it, but as it stands a lot of elderly people and their families are going to get large sums of money. Yeah, they deserve a break, but so do starving children in the slums of the places we like to forget about.

*** My greatest fear here is that, by focussing on the money, the solution will only be a temporary one. I'm sure there is still plenty of anger in Korea regarding this issue, and while Japan keeps avoiding the issue, the Japanese will be less informed as to why such a sentiment exists. This in turn reinforces Japanese hostility. In such a situation, one can only hope that time passing will counteract this.

Tuesday, October 27, 2015

Review of the Ewin Bluetooth keyboard for 7-inch tablets

While I wait for my more "competent"* computer to compute the Mann-Whitney-Wilcoxon test on a 2x4000 dataset , I thought I'd do a first-impressions review of the keyboard I recently got for my Nexus 7 (2012 version).

I had previously been using one of these (yes, it's a Japanese link, but they were definitely available in the UK, which is where I bought it), and to be honest, that was great; it fit snugly around the Nexus, protected the screen from harm, and was ultra-portable. The only real complaints I had about it were the lack of a TAB key, and the fact the letters eventually wore off**.

There are new versions of the keyboard that have the TAB key, and I would have gotten one; however, I was seduced by the idea of having a nexus case that enveloped the whole device, not just the screen-side.

Consequently, I bought this little gadget (not the pink one, the black one!). The device itself was (like any other keyboard I've come across) simple to set up; just turn on and have the tablet search for bluetooth devices. I've not had any problems with connection, and the keys are very responsive. Indeed, the keys have much more depth to them than my old device***, making it easy to feel whether you've hit a button sufficiently. However, I did have the backspace pop off the keyboard a couple of times at first (perhaps due to a bumpy ride here?). I've not had the problem since, but I'm inclined to be gentle with the device for the time being.

The keyboard itself is only marginally larger than my previous one; however, the extra vertical spacing between the buttons makes the keyboard feel decidedly less cramped, and I think I'm making far fewer typos as a result. The layout is a bit different to what I'm used to, but I know from experience that I'll adapt in no time.

The biggest change in this keyboard is the introduction of function keys. This is great as it allows for a more spacious layout and it also allows me access a number of the tablets functionality without lifting my hands from the keyboard****. However, because this is a generic keyboard, it has buttons that cater only to iProducts (a cmd button that appears to have no use, and non-functional brightness keys). Oddly, some of the typographical keys are also relegated to fn keys: the apostrophe, inverted commas, square brackets, and parentheses. If you like your apostrophes, this will take some getting used to. Luckily, being acclimatized to Japanese keyboards and their odd apostrophe placement (shift-7), I suspect that I'll adapt to this keyboard soon enough too.

Of course, for just over 2000 yen, I'm not surpised that the faux leather is of uninspiring quality; however, the product doesn't look ugly, and it is functional. One advantage that the case has is its ability to fold back and form a stand that actually positions the screen at a natural distance from the user. Even better, you can detach the keyboard from the case entirely for completely free placement of the device. I found the inability to do this with my previous keyboard decidedly awkward.

So, all in all, I think the keyboard is well worth the money. Despite some initial mishaps, it feels like a sturdy product and, having used it to type this article, I feel it functions better than I need it to.


*the lab workhorses could manage this within a few seconds...

** This ended up being the downfall of the keyboard; in an effort to maintain its usability, I resorted to using correction fluid to write the letters back on it, which worked well enough, but I didn't fancy going into work and using it (it looked decidedly DIY, and not in a good way). So, I bought some letter stickers from the local 100-yen store, and used nail varnish remover to (after much effort) get rid of the now-ingrained correction fluid. Unfortunately, I got somewhat carried away in this cleaning, and at some point the keyboard stopped working. All I can say is that keyboards and acetone don't mix...

***I hear that some people enjoy the sound of keyboards clacking. I'm indifferent, but the keyboard might satisfy such people.

****My particular favourite being the unlock button, as the lock button on my Nexus 7 has been temperamental since I bought it.