Thursday, August 21, 2014

Reasonably priced voice and data rental SIMs finally come to Japan: B-Mobile PAYG

SIM card rental in Japan is exorbitantly expensive because law prohibits non-residents from obtaining cellular voice contracts, blocking them out of the (relatively) reasonably priced market. The only options are rental services.

The rental market predates 3G, and Japan never used GSM. In the past, visitors had no choice but to rent hardware, which created an environment that (somewhat) justified premium pricing. There was no significant decrease in price following the start of SIM card rental after introduction of 3G networks, and once smartphones proliferated, mobile data options were added data but at laughable prices.

Not anymore.

Inexpensive data with a voice rental SIM

The B-Mobile line of SIM cards from Japan Communications, Inc. (JCI) got a new product today, the prepaid PAYG rental SIM, which provides 60 minutes of outgoing voice calls (incoming is free) and 3GB of 3G/LTE data for ¥9,980. The SIM is active for 7 days, after which any remaining data and minutes become invalid. The specifics are:

B-Mobile PAYG Rental Voice and Data SIM
  • ¥9,980 for 7 days
  • 60 outgoing domestic/international minutes
  • 3 GB 3G/LTE data
  • 5/75 seconds deducted for domestic/international SMS
  • Free tethering
  • incoming calls and SMS free
  • Prepaid (no worry of unexpected, additional charges)
  • Nano, micro, standard SIM card sizes
The SIM card can be purchased in stores (Yodobashi Camera) or preordered. Unfortunately, complying with Japanese law complicates the activation process. If you preorder, you will first set an activation date and upload a scan or image of the passport page with your photo. Then, following activation but within 24 hours of arrival in Japan, you will have to also upload pictures of 1) your entry stamps and 2) the image of the same passport page originally uploaded. If you buy in a store, then you will need to upload all this together as soon as possible and wait for activation.

In a press release, JCI indicated they are working with the government to try and smooth the activation process. By the way, if Tokyo police have their way, this will be required for ALL SIMs sold to non-residents, even data-only SIMs, which means that most MVNOs won't bother to go through the effort and just start requiring residency for data-only SIMs, just like for voice SIMs.

Traditional data costs with a rental voice SIM

At ¥110/day and ¥110 per minute (with a ¥315 service fee), the current SoftBank rental SIM would cost ¥7,685 over a 7 day period if a comparable 60 minutes were used – a better value for just voice. However, if you want to have minimal background data usage for mail, maps, etc., the SoftBank rental costs increase quickly.

Softbank rental data fees are ¥0.32/packet (128 bytes). Doing the math yields the following:
  • 1 KB: ¥3
  • 1 MB: ¥2,621
  • 2 MB: ¥5,243
  • 250 MB: ¥655,360 ($6,500US)
  • 1 GB: ¥2,684,355 ($27,000US)
  • 3 GB: ¥8,053,064 ($80,500US)
Yes, current Japanese rental SIMs value 3 GB of mobile data at 80,000 US dollars. As I said, laughable. Of course no one would ever use this much, and the daily maximums are capped I believe, so it wouldn't be possible anyway. However, because the traditional rentals are postpaid with a credit card, there will always be the concern of incurring additional charges.

A more realistic comparison

Long time readers will recall that I once did an experiment over about 8 months with a b-mobile FAIR data-only SIM. (There is an interactive chart showing how I consumed data during those 240 days at the bottom of this post.) The FAIR is wide open with no proxy or speed restrictions (and accordingly priced). 1 GB of data costs about ¥8,000 and is good for four months. This translates to 250 MB for ¥2,000.

All the other MVNO SIMs, especially the other b-mobile products, are speed restricted and relatively high latency. Because of this, I think it would be near impossible to use 3 GB with the PAYG SIM in 7 days (or even in an entire month). So comparing prices based on high data consumption is completely unrealistic.

Rather, I will compare prices based on my experience with the FAIR SIM because I think my behavior regarding data usage with a 250 MB monthly quota is a good analog for the way people use traditional Japanese rental SIMs. That is, I thought about the cost of everything I did with the phone. On light-usage days, I would consume around 2 MB, which is the equivalent of uploading one or two photos.

Let's compare costs by taking about a quarter of my light-day usage, 500 KB. Seven days with a Softbank rental SIM, using 500 KB per day and 60 minutes total, would cost ¥16,645. To bring the cost down to the PAYG SIM price over 7 days, total data usage over the entire week must be limited to just under 1 MB. One. Megabyte.

Good luck with that.

Tuesday, August 19, 2014

Thin bezels are thin. Heavy bulkware is heavy.

An article by Ishikawa Tsutsumu discussing the effects of Docomo releasing the iPhone late last year made brief mention of a Sharp Android phone for both Softbank and Sprint.
Softbank is hoping that Sharp’s expertise in LCD technology will help Sprint to differentiate itself from its competitors in the US market. The company plans to develop a single Android device for both countries.
Based on that, we knew this phone's defining characteristic would be it's screen.

While Mr. Son knows first hand what an exclusive, must-have phone can do for a carrier, he also knows that phones with just as thin or thinner bezels will be released by every maker soon. In a country with the population spread thinly over a large area, network coverage and quality (as opposed to phones) become much more important factors when choosing a carrier. Unlike Japan, where network coverage – and congestion – is pretty much equal in all major metropolitan areas, there are absolutely stunning differences between the four major US carriers across even highly populated areas. (I used Sprint for a week in San Francisco, and it was the most frustrating mobile experience in recent memory.)

It seems the US-based tech sphere is enamored with this phone, but I wonder how they'll feel after experiencing the quirks of a galasuma. In particular, domestic makers have been pretty bad at removing expected functionality. Some American consumers could be in for a shock with their first Japanese smartphone.

Having said that, however, Sharp is one of the domestic makers to survive the initial release of Android, which means that their firmware didn't suck as bad as NEC and Panasonic et al's. That one of our contributors loves Sharp phones also shows that Sharp is doing something right.


Preorders started today for the Aquos Crystal, and it goes on sale at the end of this month. Really, the only thing this phone has going for it is the screen. No one seg, no osaifu-keitai, and NO WATER RESISTANCE. Wut?! I thought the whole "aquos" brand was marketed around waterproofing. Silly me, I forget that brand predates mobile.

The Crystal X is the much better phone, but it still LACKS WATER RESISTANCE. It does have one-seg, osaifu-keitai, and VoLTE, though, so if I were to opt for one, it would be the Crystal X. However, it won't be available until the end of the year, perhaps even next year (12月以降), at which point I bet we'll already see similar thin bezels from other makers.

Monday, August 4, 2014

SwiftKey Japanese beta (updated)

UPDATE2: I like it.

Though all in all, I find the stock Android keyboard and the Google Japanese keyboard better. However, I am very impressed with SwiftKey so far, and it does a very good job to combine a full-featured English keyboard with a proper Japanese keyboard.

There are some things it does much better than Google's Japanese keyboard, such as correction suggestions and prediction for words with "n". Google's keyboard won't always offer suggestions after you enter an "n" until you press the key a second time to convert it to ん or hit a vowel convert it to a な etc. I find this annoying to no end. SwiftKey on the other hand never enters a romaji "n" at all; ん comes out after one press, and then な after hitting an a. This means one less tap to enter a word like わかんない. Tapping "n" twice still inputs ん, keeping with existing conventions. Nicely done.

Now, if they implement godan, it will be just about perfect.

Actually, most of my gripes are with the English input. I find Flow input to be much less good at prediction than either Swype or the stock Android keyboard. Often times I'll end up with some totally random word and am forced to explicitly type out the exact word that I want. I rarely have to do that with other keyboards that support swiping to input English. Another annoyance is with long pressing the period (.) key bring up an alternate punctuation mark. Swype is also guilty of this, which is default to inserting... a period. Why would I want to do that? The stock Android keyboard does the correct thing and suggests punctuation that is not a period (like a comma) after a long press.

UPDATE: As pointed out in the comments, the SwiftKey Japanese beta does in fact have a QWERTY option for Japanese input. I'm not sure how I missed it – probably because I was attempting to change the input layout in the section called "layout". So I'll give this keyboard a second look and update this post in a week or so.

Even after finding the setting, it's still not immediately obvious that the layout can be changed because the they are oddly named ひらがな and ローマ字, which are not appropriate ways to describe keyboard layouts. For comparison, Google's Japanese keyboard uses the terms ケイタイ配列 and QWERTY. (With system language set to English, the keypad layout becomes "12 key".)

In my never ending search for an android keyboard that doesn't suck, I finally got around to trying the SwiftKey Japanese beta. I've grown to like google's "godan" input, and I never tire of the reactions I get from the uninitiated when they borrow my phone. I do however tire of the (albeit more streamlined than in the past) three step method of switching languages which is: long press the space bar, scan a list of keyboard names, tap desired keyboard.

Joining the SwiftKey open beta was a hassle. I first tried on my phone but kept getting redirected to everywhere I didn't want to be, including the play store to download tapatalk. The process starts out completely in Japanese before switching over to 100% English. Here's how I got the keyboard.

Create a SwiftKey account

From the SwiftKey Japanese page click アカウントを作成 (Create account), which will take you to a page that is completely in English. After finally getting all the math questions and captchas correct, I waited for a confirmation email that never came.

I tried again with a different email address and finally got a confirmation email after looking up the name of their iOS note taking app. (Seriously? WTF? If any Japanese person continued after hitting a wall of English, I'm pretty sure that they wouldn't get past that.) Success. Account created.

Download the Japanese beta

For the next step, I went back to do it on my phone. If you don't log in first, you'll just get dumped onto the "you don't have permission to view this page" area after tapping ベータ版ダウンロード (download beta version). The beta is kept in the discussion forums.

Honestly, after all the work to get this keyboard installed, I feel let down. First, I find the stock Android keyboard superior for typing English. I don't often use prediction because it actually slows me down on average – I find it faster to just keep tying the word rather than scan the list of suggestions for it. Granted, prediction would become better with time, though.

More importantly, I'm also not really all that impressed with the Japanese keyboard because it only has one layout, keypad. We are no longer bound by a physical alphanumeric keypad for entering Japanese. Yes, flick-style input is a significant evolution of the number pad, but why must we keep this legacy layout? Why not reinvent the layout in a completely new way now that there are no physical design restrains, which is what google did with the godan keyboard.

So after about a week of using the SwiftKey Japanese beta, I went back to doing my long-press-spacebar tango with Google's Japanese IME and the stock Android keyboard.

This is still much better for Japanese input than the crap offered up by Swype because the SwiftKey Japanese beta has a dedicated enter button and a reasonably good dictionary. If I loved this input style, I would use this as my only keyboard. If SwiftKey implemented QWERTY (or even better) godan, then I would use this as my only keyboard.

Whither the tablet?

Ars Technica had an interesting op-ed today. They look at the stagnating sales of tablets and conclude they are already much like PCs: fast enough and good enough that they only get replaced when broken. They've got to the point where the only differentiator from previous models is the price, and where people see little need to upgrade their hardware.

I'm not surprised; my Tablet Z feels as fresh and as fast as it did when I bought it over a year ago. Android and IOS are both mature so new OS updates don't feel that urgent any longer. Mobile apps are written with slower phones in mind, so there's nothing out there that even begins to tax the CPU or graphics. I doubt I'll get a new tablet until this one breaks. My wife’s iPad Mini will also likely be in use for years to come.

This spells trouble for makers that counted on tablets to replace their stalled computer sales. I'd go out on a limb and say this is especially troublesome for a company like Microsoft; they generated a lot of ill-will trying to push their customers onto the tablet/hybrid train, only to find out there was never any market growth to be had. And as Windows Phone is not exactly taking the world by storm they have few other growth segments left.

But even having a thriving phone business may only help for so long. My guess is that in another couple of years phones will be in much the same situation as tablets today. The OS and software ecosystems are pretty mature, and we've already reached the point where more speed or RAM just doesn't add meaningfully to the experience. Makers are piling on more or less useless gimmicks (curved screens! face tracking! 3D!) to entice people to buy. Some recently popular features such as water resistance, rugged design and materials, and remote disabling and locating will actually reduce the need to replace broken or lost phones over time.

So what will replace these devices as consumer growth markets? Perhaps, in the short term, nothing. These almost magical things have replaced piles of special-purpose devices. They let us do so much more with so much less. And they keep getting cheaper, better and longer lasting. Just like the agricultural revolution let us spend much less on food than we ever used to, we consumers can now spend much less money on gadgets overall, and I suspect that's where we're heading.

In the long run we could see real robotics become new must-have devices for instance, but viable, useful "home robots" are still many years away (it's a problem of hardware cost and complexity as much as software). "Internet of Things" has some serious privacy implications and still doesn't have compelling reasons why you would want any of it. Until people come up with completely new market categories (no, "wearables" is not it), life as a consumer electronics company is going to be rough.