Mobile is the Magical Measurement Machine - Role of Mobile in Big Data Future: now cameraphone used in anything from measuring bra size to counting trees

I saw a couple of interesting Tweets last week that reminded me, that I should do an update of how mobile is in one of its magical abilities, as a measurement tool. As I teach in my workshops around the world, among its many amazing powers, Mobile is a Magical Measurement Machine. And yes, we now have fresh evidence that mobile can achieve cool and weird measurement achievements, from faster and more accurate measurement of trees, to accurate measurement of bra sizes for women.

AUDIENCE MEASUREMENT

Obviously mobile being digital and interactive, it can be used to measure its own consumption. So we can measure the amount of downloads of our app or ringtone, the response rate to our message or the click-through rate of a given banner ad etc. That yes, mobile’s own content can be measured and some of it is collected to absurd level of accuracy, automatically by the system. The value of a single domestic phone call in our network in our home town, that runs a minute, can cost say 10 cents in the Industrialized World or something like 3 cents in Emerging World. Yet the billing system will often measure that call’s duration by second. So the increment of detail is collected up to the value of one sixth of one penny in say Europe and America, and down to value of one twentieth of a penny in many Emerging World countries. And for that ridiculously low-value activity we collect the full phone number of the caller, including country called from, network called from, the physical location of the call typically to the accuracy of a neighborhood in the city, then the duration of the call, and then the called number, that person’s network and country, and if it was within our network, also the neighborhood of the city where that caller was when he or she received the call. Plus plenty more details are collected in case its a specialized call type like a premium or free call, etc. And all this data is collected for every one of us 7.6 Billion mobile phone accounts on the planet of 7.2 Billion people, for every one of our 3.3 calls on average placed per day, automatically. And yes, if its a short local call, the value of that call is less than the cost of one paperclip.

We do, yes this industry, collect enormous value of data on items up to that level of meaningless value, as the cost of a paperclip. Even paperclip makers do not collect that kind of detail, they don’t bother to track their business to less than boxes of paperclips. The telecoms billing system for mobile is the most complex and detailed and powerful data-collection machine ever devised, at whole orders of magnitude more detailed and powerful than the data collection systems at banks, credit cards, loyalty card programs etc. And they also collect incredible amount of detail about our behavior. But nothing comes close to the data that the 600+ mobile operators/carriers in the world do, automatically and in their normal daily business. But thats not what this blog is about. That is just a comment on the fundamentals. There is nothing as powerful as the mobile telecoms billing systems to collect consumer data. It is not by any means ‘optimized’ to collect the most useful data, most operators/carriers are woefully inept at capitalizing on the data they can collect, and they are utterly hopeless at monetizing that ability. But lets not kid ourselves into thinking that some online web service can hope to collect the same level of detail as accurately and thoroughly as mobile telecoms can. Not even Facebook or Amazon (while they can collect enormously valuable information far better than anything that existed before the internet).

I led the project that created the world’s largest multi-operator billing system early in my career when I still worked for the carrier/operator side of this business in Finland in the 1990s. So I got very deeply indoctrinated into what all data we can collect into our ‘CDR’s (Call Detail Record)s and explored early ways to expand that ability such as adding the number of data items that could be collected into the same CDR string as we as an industry back then were going from numeric to hexadecimal etc. (the same two digit ‘code’ if using only numbers gives you 100 choices, or 99 choices and ‘none’ ie zero, while hexadecimal uses the same two digit space but expands the data items collected to 256 choices ie 255 actual items plus zero). So yeah, I rolled up my sleeves and got my hands dirty doing really that level of project, a multi-operator (multi-carrier) billing system across systems from multiple vendors and produced the world’s largest multi-operator billing at the time, a world record. And later when I was leading Nokia’s Global 3G Consulting Department, that knowledge was very valuable in what kind of data a carrier/operator could capture - about its own services or those services it co-developed with partners.

USE MOBILE TO MEASURE OTHER MEDIA

I kind of forgot about the billing for a while, as the mobile internet and related mobile data capabilities brought us vast new technical abilities. But then about ten years ago, I got back into looking at data collection, when it became obvious to me, that mobile could be used to measure other media. Some of the early uses were for example from New Zealand, where a TV station ran a contest to let viewers win nice prizes including cars. What viewers had to do, was to send in an SMS text message whenever they spotted a given logo on the TV screen. Now, the consumers had to pay regular consumer-rate SMS costs for these messages (they were not like votes on American Idol where we pay extra, ie ‘premium SMS’) but still, as the TV station flashed those codes several times per day on its two channels, a consumer might quickly run up surprising SMS costs on the phone bill. So to compensate the TV station also sent by return an equivalent value in digital content, so if you sent say 10 messages, you had earned enough credit that you could then buy a ringing tone etc. This meant, that the consumer felt the service was essentially ‘neutral’ in its costs.

What did the TV station learn? Gosh, this was far more powerful than anything that Nielsen boxes could ever hope to capture. Now they knew not just if the TV was on, but was the consumer actually watching the TV show. Because we don’t share our phones with others, not even our spouces, they got perfect viewer data on whether one family member watched the whole time or one left, and another came in to watch. They could flash the logo at the beginning of a TV show and the end of it, to discover if the same person watched the show from beginning to end. They could run the logo on the ad break, and see - measure - accurately - how many of the viewers of a given TV show actually stay to watch the advertising. And as TV shows program sequential programming, ie they hope that we stay after a popular TV show to watch the new show that comes right after it, they could also measure that. In other words, of all who watched the TV show from 8 to 9, how many stayed on to watch the next show that starts at 9. And the sample audience was enormous, far far bigger than any Nielsen or other TV audience survey ever before in history. Massively bigger meaning the sampling errors were reduced to infinitessimally small.

That was when I became convinced that mobile is the most accurate measurement tool for any media. This can just as well be done for radio or newspapers or magazines or even billboards and cinema. Even the internet can use mobile to measure its audiences. And about this time, Tony Fish's company AMF Ventures also came to the same conclusion, when they compared the accuracy of media measurement. They said that on TV about 1% of the audience can be measured accurately. On the internet about 10% of the audience can be measured accurately (we use multiple accounts, delete our cookies, sit behind firewalls, use multiple devices, access from many locations, and many people share the same PC especially for home PCs in families with kids). But on mobile AMF Ventures found that 90% of the audience can be measured accurately. No, mobile is not perfect but it is awefully close. And nothing else is anywhere near as good. Mobile is a whole order of magnitude more accurate than the internet, two orders of magnitude better than television. And let me just mention here Tony's brilliant book My Digital Footprint if you want to read the best book on what mobile data is, and what it can do by someone who really understands this industry, not by some misguided American 'big data' dude who is clueless about the true power of mobile.

CONSUMER MEASUREMENT

But it then went beyond mass media, and my next astonishment came from Japan, Lars Cosh-Ishii of Wireless Watch Japan told me about a Marlboro campaign, about 8 years ago. Marlboro which sponsored Formula 1 and Ferrari at the time (this was when Michael Schumacher was still driving for Ferrari) had a very special competition, every time you opened a pack of Marlboro cigarettes, under the lid of the box, there was unique code which you needed to send to Marlboro via mobile messaging. The codes were unique per cigarette package of 20 cigarettes. And every time you sent another code, increased your chance to win a prize. The grand prize was a visit to Maranello Italy as Ferrari’s guest, a factory tour, of course seats at the Monza Grand Prix, and meeting Michael Schumacher.

Cool contest, yeah. But can you see what Marlboro was doing? It never knew its end-users personally, nor how they consumed the cigarettes. Now as they got per mobile phone number, the unique codes per box, they knew what type of cigarettes you smoked. They knew how much you smoked. But they also knew if you bought Marlboro by the Carton (you personally sending in all codes of the packs of cigarettes in one carton) or bought the packs individually. They could then further tabulate data on whether you tended to buy your cigarettes from a vending machine or a store, and how many differrent stores you might use, when they cross-tabulated user data by store-data of where the given cartons of Marlboro had been shipped. Etc etc etc etc etc. The most thorough and valuable and accurate consumer data ever collected in the cigarette industry.

Today we see this all the time. That is why Pepsi asks us to scan the QR code on the bottle or their advertisement, why there are SMS codes under the bottle top, etc. Its a variation of that same idea, but yes, mobile is a magical measurement machine. One of the ways we can measure, is if we incentivize our audience to participate in some interactive campaign or contest and we can collect amazing data from them on their consumption. And what comes next? This is what Omo the detergent from Unilever did in South Africa in 2013. They ran a similar ‘text to win’ campaign with every package of Omo detergent purchased. As they knew the size of the detergent package and the frequency of that specific housewife’s purchase pattern, they started to send reminder messages a little before the current package ran out. Was it successful? Try this: 60% increase in repurchase rates!!!!! That is why Winston Wang, the Director of Innovation at Anheuser Busch InBev (the world’s largest beer brewer) says “Mobile is measurable like nothing before.” And it is what for example author Peggy Anne Salz now writes in her latest book, The Everything Guide to Mobile Apps that app developers should use SMS (and MMS) to engage with their users.

MEASURE IDENTITY

Well, we all know how much I love SMS and mobile messaging but we can do far more with mobile as the magical measurement machine than just ask our audiences to interact with us. Obviously we can spy on them with apps. I am not going to bother with that here now. But let me show some more magical things we can do. I was just in Sweden earlier this year to watch the Stockholm Marathon (to support my nephew running in it) and stayed at one of the Clarion Hotels right in downtown Stockholm. That was one of the first hotels where they let you use you smartphone (if it has NFC like now finally Apple is also bringing to iPhone 6) for operating the lock on the door to your room. No need for plastic key card. That is one level of measurement, we can assign digitally (and even turn on and off remotely) the access rights of a key to a hotel room (or office or rental car or our home). Assa Abloy the world’s largest lock-maker has been offering NFC phone operated locking equipment for many years now.

And then really, seriously, mobile is powerful at very simple levels too. It need not all be Sci Fi iPhone 9 time-travel and teleportation stuff. We don’t need to go for NFC or an app. Somethings can be so very simple. Like the pharmacist chain in the USA, Rite Aid. they noticed that all of their customers had mobile phones. So they created a parallel number for their loyalty card. You could register your cellphone number as the alternate number to the loyalty card, and then stop carrying the plastic loyalty card. As there are no identical mobile phone numbers in use on the planet - yes the mobile phone number is the nearest thing to a universal identity number - Rite Aid lets you now just quote your mobile phone number at the cash register, to collect your loyalty card benefits. The flip-side is, that Rite Aid now has a direct channel to your pocket, and will for example send you reminders when your prescription is running out. So understand how powerful mobile can be, and that we don’t need to always make it Augmented Reality or NFC or QR Code or something fancy. Even just the basic mobile phone number has incredible value and can help identify us. In other words, to help measure the loyal customer behavior...

And what can we do with this? Gosh, if the phone is that accurate it can identify us, it can be used for example as legally-binding signature for contracts, as they do in Spain and Turkey for example. Yes sign a contract by sending an SMS text message! How efficient is that? How much faster than faxes or couriers. Then in Norway you can file your taxes by SMS! Again, how much better is that than any alternative (as long as your tax return is not very comples with international holdings and various trusts and corporations etc haha). And the ultimate was in Estonia. The first national election where you could vote by SMS! Every eligible voter in Estonia when registering, could opt in to register one mobile phone number as their official voting phone, and then use that to do the vote. Of course this is safer easier faster and more accurate than any other paper or machine we have for voting. Why isn’t every country doing this?

TRAFFIC SPEED

So lets go back to Sweden, I was there a couple of years ago delivering a keynote to the Location-Based Services industry event in Malmo and learned of another interesting measurement. In Sweden the traffic authorities noticed that they can get extremely accurate traffic data by monitoring the speed of handset movement from one cell site to the next. Blackberry was one of the first to deploy a commercial system that generated this type of data for municipalities. So rather than install radar guns at various points on the roads, they could just take grouped data of say 10 mobile phone numbers at a time, take the average of what speed are they moving from one cell site to the next. Any which moved at walking speed or say bicycle speed are removed, and the remaining numbers are averaged. We get essentially perfect average speeds on highways in current real-time data. With a little bit of tweaking, this kind of measurement can get essentially a perfect traffic picture for any city (with obviously the help of at least one major operator/carrier which would need to provide access to this type of data).

Note we are not spying on the consumers, nobody’s privacy is invaded, nothing is pushed or sold at the consumers. We don’t need to know even the real phone numbers of the people, so the carrier can create virtual phone numbers or virtual identities, just so, that the one same phone number is tracked on one journey, and when it is moving clearly on highways, we measure that movement. If that one phone suddenly slows down while most others on the same route continue at normal speeds, we know that one driver now pulled off to a petrol station or something else, so we stop counting that phone for hte highway average speed. But if there is an accident, then all cars on those lanes suddenly stop, that is cause for traffic monitors to find out what happened and send a police helicopter to go take a look etc. There are plenty of commercial applications of this type of technology, for example many car insurance companies are now starting to offer ‘pay as you drive’ insurance plans that then monitor the actual driving behavior (with GPS and cellular data obviously and with sensors both in the car, and synchronized to the driver’s mobile phone number).

CREDIT SCORE

That all is still very obvious and not that ‘magical’. Now lets get to the magic. The first really cool or weird, depending on your viewpoint, way to use mobile telecoms data for amazing measurement is with an American data analytics company called Cignifi. They say they can calculate a credit score based on our mobile phone behavior. Note, they are not trying to spy on our calls or messages by reading them or listening to us. They really do it by the communication pattern. So, a person with a healthy credit status will behave in a certain way, in terms of incoming calls, outgoing calls, returning calls, etc. A person who is on the brink of bankruptcy behaves very differently, clearly avoiding certain calls, not returning certain calls, and calling some numbers frequently etc.

Now, I have no knowledge of how accurate Cignifi is in reality, but they claim they can calculate an accurate credit rating by analyzing just the last month of phone records data. The Cignifi system is already being used in Brazil and Tanzania. But imagine specifically that problem for a credit card company or a bank. In the Emerging World there is considerable need for credit, there is plenty of new affluence but there is very little ability to prove credit-worthiness. Issues of identity, address, employment, taxes, etc are all murky at best. So where we in the West can rather easily see how an adult person will have collected ‘credit score’ data like changing jobs, making payments on some credit instruments or having history with salaries, bank accounts etc that kind of info is very rare in the Emerging World and very many truly credit-worthy potential customers can not be accurately assessed. Until mobile. Mobile is a Magical Measurement Machine.

CAMERA AS MEASUREMENT INSTRUMENT

Again, I could go on and on about this, but lets move to the less obvious sensors. We can of course imagine using the network data, the location-data, the telephone and messaging history data to create inputs for our measurement. But what about the other sensors on the mobile? We can use the microphone (there are security systems built around voiceprint). Shazam was the first company to use the mobile phone microphone to listen to music and then compare that to their database of recorded music, to identify the song that is playing. And so forth and so forth. So lets move to the camera. Now we really do get magical stuff.

My first fave camera-as-measurement example came from Japan a few years ago, when NTT DoCoMo introduced its cameraphone calorie-counter. Their engineers had created an artificial intelligence based system that identified the 1,000 most popular dishes eaten in Japan ranging from sushi to a Big Mac at McDonalds. And the system very accurately identified what was on your plate just by snapping a picture of the food you were eating. Then the system automatically calculated the rough amount of calories in that dish you were eating. Wow, this is cool. This is magical. You can just take a picture of what you eat, and the system collects the calorie-count for you. Mobile is the Magical Measurement Machine.

BUY WHAT SHE IS WEARING

So then even more Japanese cameraphone magic. Our Alan Moore (co-founder of this blog and my co-author to the signature book to this blog, Communities Dominate Brands) writes about Tokyo Girl (aka Girlswalker) the youth teen mobile magazine in Japan, in his latest book No Straight Lines. The service is just amazing on every level but let me just skip to the cameraphone magic. They run twice a year the fashion show in one of Japan’s largest indoor sports arenas, packed with screaming teenager girls all crazy about the latest fashions. So, one of the gimmicks is, that when the fashion model walks on the catwalk, you can snap a picture of the dress with your cameraphone, and click-to-buy the item, in your size, in the color you want. How magical is that? If you like something the fashion model is wearing, just point your phone at it, and buy (obviously these girls have the Osaifu Keitai mobile wallet on their phone, as I explained in my blog about mobile money around the world. This is what the iPhone 6 is bringing now to America, Japan has had it for literally 10 years).

But yes, how is that possible? Teens at the back of the stadium pointing a cameraphone at the stage will only see specs, how can it possibly identify the model on the catwalk and get us the right clothing? Its a trick! They don’t use the camera, it only seems like it. They are synchronized to time. The show producers know which model, wearing what clothes, is on the stage at any given point. If you click then, you are served the link to the clothes of who is scheduled to be on right now. They don’t need to ‘see’ who is on the stage. But teenage girls don’t need to know this. They just know, take out your phone and be ready to snap a picture when nice clothing is on the stage, you can buy it on the spot.. Isn’t that cool? We don’t necessarily even ‘need’ accurate camera images to use the camera as a sensor haha.

MEASURING FOR BRA

Which brings me to my two newest fave stories. First lets go to naked women... Apparently (I don’t know as I’m a man) 80% of women wear ill-fitting bras and they mostly do it because measuring for the right bra is a cumbersome procedure with strangers measuring you very intimately. So cameraphones to the rescue! An American company called ThirdLove has developed an app that requires taking a couple of pictures of the chest of the lady (naked, in private, via selfie, without discomfort, at home) after which it then calculates the exact proper bra type. The app was built by an ex NASA engineer and the company includes some ex Google staff, it seems pretty sold. Apparently women hate bra shopping so much they’d rather have their eyebrows waxed haha... (Again, I can’t comment, we men don’t have either of those problems). So. Measuring by the camera sensor? That is magical. We can measure food, we can measure ladyparts and if you think thats too touchy-feely, you want something robust to measure. How about industrial forestry?

MEASURING TREES

Finland is one of the big forestry, pulp and paper producer nations in the world (70% of Finland’s surface area is forested). At one point in the 1980s, we said in Finland that the Times of London, the New York Times of the USA and Pravda of Russia were all printed on Finnish paper. If your tennis racket or hockey stick is made of wood, there’s a good chance that wood came from Finnish woodmills. Nokia was once selling toilet paper (among its many product lines before it decided to focus on telecoms). Some of the world’s largest paper and woodmill companies are Finnish with operations from Canada to Indonesia. Now trees are particularly ‘non digital’ goods. Most things that come out of factories can be marked in some way that can include bar codes or NFC chips or something but trees, gosh, they just grow in the forest. So if you are a forest manager and need to run your business professionally, you do need to know about your tree inventory and their age, when to cut the trees etc. And measurement in the past meant sending experts into the forest to count. A tedious manual process.

So enter the cameraphone. Finnish company Trestima has built an app that uses the camera to take pictures of the forest to be measured. What normally would take a whole day of inventory work can now be done by 40 minutes of work taking pictures. The system is smart enough not only to detect the types of trees and their ages and their density, but also far more accurately. Accuracy is more than doubled compared with traditional methods. Mobile is faster, easier and more accurate. By using the camera. Mobile IS the Magical Measurement Machine.

As we head to the information age and we approach ‘big data’ etc, there will be lots of ways we can collect data and many sources for it. But mobile is the most prevalent sensor network today (3 times more mobile accounts than total internet users globally). Facebook is nice but SMS is 4 times larger by active users than Facebook. Cameraphones have ever more powerful sensors the Nokia 808 Pureview and Lumia 1020 have their monster 41 megapixel sensor if you wanted really sharp detail like say magnifying tiny print on a page. This is called ‘megapixel microscopy’ ie using the mobile phone camera as a magnifying glass. Similarly now the Samsung Galaxy K Zoom and for example Sony’s accessory lenses for the Xperia, offer 10x optical zoom - thats a genuine ‘binoculars’ effect in your pocket, every day, everywhere you go. There are infrared and microscope attachments to smartphones already.

If current cameras can do this kind of magic, where are we in a few years. The number of sensors on the mobile phone make it an incredibly powerful instrument for collecting data. The connectedness allows it far more utility and power. We are very early on that journey to what all the mobile phone can measure. And as management discovers the power of mobile as a measurement machine, it will embrace it, because anything that can be measured can also be managed.

Comments

[Henry Sinn]

Tomi,

Great to see blogs of other mobile happenings again....

Seems you've been overly distracted by Elop and other iThings for a few years.

Awesome stuff. Keep it coming

[virgil]

" And the ultimate was in Estonia. The first national election where you could vote by SMS! Every eligible voter in Estonia when registering, could opt in to register one mobile phone number as their official voting phone, and then use that to do the vote. Of course this is safer easier faster and more accurate than any other paper or machine we have for voting. Why isn’t every country doing this? "

Well, I don't know about other countries, but mine has in the constitution written down that the vote must be secret and non-identifiable (i.e. the government can't know who I voted for, only that I voted). As you say, an SMS is very much identifying, the guys operation the service would know what I voted, which would be illegal. And for a good reason.

[Tomi T Ahonen]

Hi Henry and virgil

Henry - yeah, you know it was partly by design. When we approached year 2010, it became obvious to me that the smartphone space would be hot - as its been the past 3.5 years - and as I had some insights I could share to help our readers navigate that difficult global market, I felt it a worthwhile investment of my time. I can tell you it was at times quite difficult to find the time to blog about the latest news while on crazy travel in my consulting work. But I think it was worth it. As you know, that crazy time is now past and most of the smartphone wars are now settled.

As to Nokia and Elop, well, that was just Elop. If he'd only made one mistake, I'd have written a couple of blogs at best. But as you saw, he managed 33 truly strategic and immensely damaging management mistakes in his 35 month tenure as Nokia CEO, so the reason I had to write so much about Elop was simply that he kept digging Nokia further into that hole he blasted with his idiotic memo, haha.. I wasn't repeating most of his mistakes as the cause of any new blog, while I would often mention his history, most of those blogs were caused by yet another new mistake. Luckily finally the Nokia Board saw how bad he was and removed Elop from office a year ago and I could stop that series of blogs, with only the wrap-up blogs about the final count of the damage haha...

But yeah, its 'good to be back' talking about the other things, there is far more to mobile than the smartphones. And you know also Henry that on the blog I have been writing every so often about Africa's mobile scene or developments in mobile advertising or mobile messaging etc, so its not only been smartphones the past 3-4 years haha... But yeah, I have more of these items still pending, items partly drafted or just waiting to be blogged. And I'll still finish this year of the Bloodbath series but probably next year I won't bother with it.

virgil - haha interesting viewpoint. Actually I do believe that almost whatever is your country you probably have some form of absentee voting too. Voting by mail or voting at the embassy if you're abroad etc. Those would be quite suspect to capturing voter data and actual vote (or else be suspect to voter fraud). But specifically SMS voting, its actually rather easy technically to do what you wanted. Its similar to how most polling places work in the real world too. One part of the polling place verifies your identity, you have to provide your ID or your voting document whatever it is. Then separate from that, is a secret voting booth where you do your voting by paper or machine.

So we just need to do the same via technology. If we have the national registration of the mobile phone number as the individual registration for voting, its rather easy then. We install a 'white list' of approved voter phone numbers. That will act as the gate-keeper just like at the polling place someone inspects your ID. Then if your mobile phone number is approved, then the vote is passed through to the counting system but without the phone number. Meanwhile the number is marked as 'having voted' so you can't vote twice, and you can't go vote in a physical location etc. But the vote count is now separated from the mobile phone number so only your vote counts without identifying your name (or mobile phone number or any other identifier) with the vote. Its technically rather easy to build, we did similar things for example when we did some consumer behavior data from telecoms data, where we didn't want to allow the individual to be targeted but wanted aggregated behavior data. So technically this is rather easy to build to satisfy that need your constitution has. Obviously we then have to trust that nobody like the CIA gets into that system to do some data-hacking or spying at the stage when the SMS arrives before the phone number is stripped from the vote, haha...

An SMS vote is at least as secure as a mail-in vote, and far more easy, accurate, fast and less costly to process than a mail-in vote.

Tomi Ahonen :-)

[virgil]

Hi Tomi,

Well, if we go into security, I have to disagree :)
The problem with SMS is that it carries option+identity in the same message, and you have to trust someone (who can read both) to record the identity, transmit the option further, and then forget the identity.
With mail-in, you can actually have an envelope within the envelope. The outer envelope is the identity (who voted); the inner one is the actual vote. Whoever handles the votes can mark you as "has voted" and put the inner envelope in an urn with the rest of the mail-in votes. Yes they could theoretically open the inner envelope, but that would make it rather obvious that they have broken the law.

BTW, my country doesn't have mail-in votes (or any other 'absentee voting' (you have to vote at one of the official polling stations - embassy if you're abroad). I've had a rather long discussion with "my" senator on the subject, he feels that the critical problem that cannot be worked around with absentee voting (and the reason why we don't have it/ he doesn't push a bill on e-voting) is that you can't guarantee that the votes are 'free' - i.e. votes are not bought (or otherwise constrained).

I disagree with him on this, I think there are ways to make sure that the vote is free even outside the official polling stations; but that would be a long discussion. My point was that you give away some non-trivial liberties/rights with these "conveniences", we shouldn't always hail them as "great idea, why doesn't everybody implement this?"

[Jonas Lind]

I agree with virgil. A voting system has to be both anonymous and tamper-proof. Any e-voting system is susceptible to software hacking and/or corruption. There is no way to guarantee that a touch screen vote at the polling station is not manipulated by some software exploit (or corrupt officials or evoting machine vendors). The only way to ensure that no voting fraud has taken place is to have a paper trail where the votes can be recounted if there is doubt about the result. This is particularly important to prevent conspiracy theories forming among the losing side in an election battle.
The anonymous vote is important to prevent voters from selling their votes to the highest bidder, and to protect for example women with abusive husbands from forcing them to vote for the party he wants her to vote for.

[Gonzo]

Some very good ideas here for Samsung to catch up, as expected, at least by some of us here, the market share strategy failed and today samsung announced an amazing 60% profit drop in Q3, a stunning financial decline so fast that is hard to digest, and this is the 3rd or 4th Quarter in a row. It was so clear that racing to the bottom and selling cheap phones eventually leads to problems.

Well, but maybe this is what samsung needs, maybe they will learn a lesson, and focus on creating value for consumers instead of going after a pointless market share strategy selling cheat phones in a make no money ecosystem, well…. except Microsoft, An Android cross-licensing contract with Samsung had been yielding Microsoft $1 billion each year in payments.

[Paul Ionescu]

Another book about Elop's adventure at Nokia called "Operation Elop":
http://www.hs.fi/talous/a1412646148850

[chithanh]

@virgil
Do you assume that the vote will be sent via SMS in plain text? The problems that you mentioned have long been solved and SMS based voting needs not be less secure than paper based.

The voting message would ideally consist of three parts: Voter identification, encrypted vote and a unique vote ID which is generated by the voter (and kept secret).

The "gate-keeper" will only be able to verify the identification. It forwards the encrypted vote and vote ID to the vote database, where it is decrypted and counted.

The database of votes and associated vote IDs will be made public. So every voter can download it and verify through the presence of their vote ID that the vote was counted properly.

@Gonzo
There is no other strategy for Samsung. They will continue to see falling profits because Chinese manufacturers can provide almost the same thing at a lower price. Their only hope for margins is their brand and keeping costs down through economies of scale.

[John Phamlore]

@Tomi,

You in fact may know an analogy to what women face shopping bras if you get your suits hand-tailored. For one of the questions such a tailor will ask is, "Which side do you dress."

From this we see what the real problem is that mobile purports to be solving. First there is obtaining customized data, for which no doubt the camera on a smartphone provides an excellent solution. But more importantly there is the translation of such data into economically producing and transporting a physical object, in this case, an article of clothing.

It's not like the smartphone is the only solution to the data collection problem. Not too long ago in my memory women were advised to go to the nearest high class clothing store where an experienced older woman would be working who could take decent measurements and offer advice. From this example we are starting to see what the real problem is, particularly in the case of the United States. So here's something for your files about how terrible and inefficient the United States can be. What really changed is that older more experienced workers have been laid off to be replaced with either younger less experienced ones or no workers at all. So now women are left to fend for themselves.

Now for the second part, the production and transportation of physical objects: Where I live, a wealthy area, it is apparently quite popular for shoppers to go to a mass discount store and attempt to find clothing among stacks of clothing that are simply piled on tables, with no changing rooms in sight to even try clothing on. Any thought of customization or service is thrown out the window. Why would people put up with this? The obvious answer is price, and the reason price can be such a factor is the production of clothing can be moved around the world to seek the absolute cheapest area and workers. But there is of course other prices to be paid and one of them is that this works well only if one's size of clothing is on sale.

Note that clothing such as bras are especially where the idea of offering just a few sizes completely breaks down for satisfying large numbers of potential customers.

But there is another factor, a psychological one, that may explain some of mobile's appeal but also show its potential limits. One can shop for clothing in piles in relative anonymity, not having to interact with anyone other than the clerk taking one's payment, and even that human interaction is starting to disappear. More on this later on the real secret and possible limitation of mobile ...

[eduardom]

off topic: book on Elop at Nokia

http://yle.fi/uutiset/new_finnish_book_microsofts_elop_was_no_trojan_horse_just_a_terrible_boss/7515214

[Christoph]

I completely agree with my co-commenters on the voting by SMS issue. Not only that the election secrecy can be easily violated but also it opens the door to election fraud as there is no paper which can be re-counted. And thousands of votes can be easily falsified if you only need to hack one computer system. Modifying thousands of papers is much more difficult.

Concerning measuring for bra: "naked, in private, via selfie, ...". You may search the web for the word "fappening" in case you think your smartphone pictures are really private. I know, you can disable upload to the cloud, but I assume like these "stars" (I hardly recognize any name of the affected women) most people are not really aware of this option.

Concerning credit score: I heard about a similar project which wanted to use your Facebook activities and contacts to determine your credit score. One of the data points was if you have contact to other people with financial problems. If people are aware of these scoring practices, they will most likely adapt their behavior to appear less "suspicious", in this case they could abandon their contact to their friend in financial trouble, stigmatizing these people even more.

Expanding this idea to other fields: Will you fix a date with your doctor for an AIDS test if you know that your health insurance could rise your fee due to "unsafe behavior"? If you're in financial trouble (but not bankrupt yet) will you call a debt counseling if you know that your bank could cancel your mortgage as a reaction? I'm afraid that this kind of data analysis could lead to self-censorship which in the end does more harm to society than good.

[Per "wertigon" Ekström]

I too am highly sceptical of e-voting, for three things;

1. Ultimately you need to trust your government (or other official body administring elections) not to be corrupt and keep records of voting. One can theoreticly make something like, say, a block-chain based ledger keeping track of who votes and who does not. But ultimately, you need to ensure that what you vote does not get registered, and the only entity that can do that is the election administration body.

2. Provided we come up with a way to 100% anonymize voting, which is a very hard problem to solve, we today do not have a 100% proof that votes have not been tampered with. In ballot voting, it is easy to find out if someone tampered with votes - just count them again. You have a paper trail, use it. In e-voting, no such confirmation can be made.

3. And finally, even if we solve problem #1 and #2, there is problem #3; what if the person voting is voting under threat of violence or other forms of intense pressure? What if Alice wish to cast a vote on Republicans, but Bob is holding a gun to her head and threatens to kill her if she does not vote Democrats? Yes, the example is overly dramatic, but it is one that has to be considered. Unless voting takes place in a public place like, say, a library, then no such control can be made.

These are three very, very complex problems that need to be solved before I would consider e-voting as a viable choice. Not ruling it out; just saying these three problems need to be solved first.

[virgil]

@chithanh - that's not the system described by Tomi. He describes "voting by SMS", by humans - and not a machine that uses the SMS protocol/delivery system in order to transmit data.
I don't think your average voter is able to encrypt his vote - not even with a simple XOR encryption :D

[chithanh]

@virgil and Christoph

Nobody made the absurd suggestion that the voters would encrypt their vote by hand. This is taken from the e-voting website of Estonia:

> If mobile-ID is used, the voting procedure goes like this:
> 1. The voter opens the webpage for voting.
> 2. The voter enters his/her mobile number into the computer. After that a control code is sent to voter's mobile phone by SMS.
> 3. The voter identifies himself/herself by entering the PIN1 code into the mobile phone.
> 4. The voter is shown the candidate list of the appropriate electoral district on the computer screen.
> 5. The voter makes his/her voting decision, which is encrypted. A control code is again sent to the voter's mobile phone by SMS.
> 6. The voter confirms his/her choice with a digital signature by entering the PIN2-code into the mobile phone.
> 7. The voter receives a notice on the computer screen that the vote has been accepted.

Seems reasonably secure and anonymous to me, as long as the receiving SMS gateway is not able to see the decrypted vote.

It could be made even safer by not using a website, but a J2ME or smartphone app to encrypt the vote, and let the voter add a self-chosen vote ID. But I challenge the idea that this system is inherently less safe and more susceptible to manipulation than paper based voting.

[RottenApple]

@chithanh:

"But I challenge the idea that this system is inherently less safe and more susceptible to manipulation than paper based voting."

As long as there is even one place in the entire chain where the vote can be assigned to some data point that makes user identification even remotely possible it's inherently less anonymous, and for something as sensitive as voting, is unacceptable.

There's good reasons why, despite having personal computers for 30 years now, in many countries, voting is still strictly being done on paper and computers only coming into play when the actual numbers have already been verified.

Of course, even with paper voting you have to make sure that no single group has total control over parts of the voting process - otherwise a rigged vote is almost ensured.

[chithanh]

@RottenApple
"As long as there is even one place in the entire chain where the vote can be assigned to some data point that makes user identification even remotely possible it's inherently less anonymous, and for something as sensitive as voting, is unacceptable."

But this is true for paper based voting by mail too. The voter identification and vote are sent in the same envelope. So the postal service can intercept all the votes. Same for going to a vote booth: The people who check your ID and hand you the ballot can secretly mark it.

If done properly, e-voting can be made fully secure, fraud-proof and anonymous. Those who think like Christoph that the Estonian SMS voting system's "election secrecy can be easily violated but also it opens the door to election fraud" are welcome to explain their findings in more detail. (Well maybe not on this blog if Tomi thinks it belongs elsewhere)

"There's good reasons why, despite having personal computers for 30 years now, in many countries, voting is still strictly being done on paper and computers only coming into play when the actual numbers have already been verified."

The reason why voting computers are not used in more widespread fashion is that the advantages that they bring are only marginal in the current voting tradition.
In the future, they could allow Condorcet methods (e.g. CSSD) that are often infeasible to count manually.

[RottenApple]

@chithanh:

"But this is true for paper based voting by mail too. The voter identification and vote are sent in the same envelope"

In which country? If you ask me, that's the fundamentally wrong way to do it. The voter's identity should be checked when requesting the papers, not when sending it. That way voting by mail can indeed be completely anonymous. But for SMS that's fundamentally impossible because the SMS will have a sender ID attached to it by design.

"The reason why voting computers are not used in more widespread fashion is that the advantages that they bring are only marginal in the current voting tradition."

That may be a reason for careless countries but in most countries the reason not to use them is that any automation is a security risk. Wanna bet that even in 50 years there'll still be lots of places in the world where voting will still be done the old fashioned way without any electronic help allowed?

[chithanh]

@RottenApple

"In which country?"

This is common practice in many countries. I rather ask you to tell me one where this is not the case.

http://en.wikipedia.org/wiki/Postal_voting

In the US for example: "To vote by mail, an individual marks the ballot for their choice of the candidates (or writes in their name), places the ballot in a secrecy envelope, seals it, places it in the provided mailing envelope, seals it and signs and dates the back of the mailing envelope."
The mailing envelope contains the name and signature of the voter, and the secret ballot itself. Anybody who gets physically hold of it can learn the vote of that person.

"But for SMS that's fundamentally impossible because the SMS will have a sender ID attached to it by design."

That SMS have a sender ID is correct, but not different from mail envelopes which have a return address.

"That may be a reason for careless countries but in most countries the reason not to use them is that any automation is a security risk. Wanna bet that even in 50 years there'll still be lots of places in the world where voting will still be done the old fashioned way without any electronic help allowed?"

Of course there are legitimate security concerns but these can be addressed. If done properly, electronic voting is not less secure than paper based.

[David Bailey]

Great to read that Estonia was mentioned - you can do your taxes and vote online now in Estonia! Every person has two unique PIN codes, a 4 digit and 5 digits one - you use the 4 digit one to login to vote and then the 5 digit one to confirm your vote - thus it is very secure.