Sipping a Shanghai surprise

This ad appeared several times on websites I visited (e.g. xe.com) over a period of several days earlier this month. It illustrates a phenomenon which I have also observed on AirBnB and perhaps elsewhere. The website obviously determines two different locations: one for the currency symbol ("R") and another for the actual currency (displaying "1199") so that we see the price for (say) London but with the currency symbol for (say) Johannesburg. The result is a $79 intercontinental return flight.

It could be that this is a bug in a specific script language or that the content management system somehow uses different script languages for the currency symbol and for the currency amount and that these two return different location values. Anyway, I am a bit surprised that this has not been fixed yet... or that Lufthansa (not so surprised) have not updated their website software.


Schneier on surveillance and (too) big data

A conversation this week with Bruce Schneier (link below) that I heard this week, confirms my feeling that companies and institutions might in future store much less data about us than we suppose. The main reason is liability for losing it as well as the fact that much of it is fairly invasive but rather useless. For example, on my way to the office on a quiet suburban road, there is an stretch after an intersection which Google Maps tends to show as having moderate traffic. This is nonsense: it is merely a downhill bit where the single drivers they are tracking presumably slam on the brakes a bit. Google Maps is obviously inferring the "moderate traffic" by gathering the data of a single vehicle.

I believe that companies and institutions should store as much data as possible in hashed or encrypted form in order to reduce the potential liability arising from data breaches. If it's really necessary, they might want to consider offline storage!

A Reckoning for Big Data – http://www.cato.org/multimedia/daily-podcast/reckoning-big-data


Die soms lastige onbepaalde lidwoord in Afrikaans en Unicode

Sedert woordverwerkers soos MS-Word begin het met sogenaamde outokorreksie van aanhalingstekens (as "smart quotes"), sukkel ons baie met die Afrikaanse onbepaalde lidwoord [’n] wat sommer net die lelike en lastige [‘n] word. Verskeie platforms (soos Facebook) het in die laaste paar jaar oorgeslaan na die spesiale Unicode-enkele karakter U+0149 ("LATIN SMALL LETTER N PRECEDED BY APOSTROPHE") om dié probleem, vermoedelik, te vermy. Dít stel die hele lidwoord as één karakter voor. 

Unicode is 'n stelsel wat die meeste van die wêreld se skryfstelsels kan behartig maar die probleem met hierdie benadering sien ons vanoggend in Rapport (soos hiernaas). Die webblaaier (Safari op OSX) het nie die ietwat eksotiese U+0149 in die lettertipe wat aangewys word nie en vervang dit met U+0149 uit 'n ander lettertipe en dit lyk nogal lelik! Sover ek weet is Afrikaans die enigste taal met 'n enkelkarakter-lidwoord in Unicode en dít het amper per ongeluk gebeur. Die relatief algemene gebruik daarvan is nogal verrassend en vermaaklik.


Data breaches illustrate that we need privacy

The US Office of Personnel Management (OPM) has been the victim of a dramatic data breach over the past several months, in which personal details of millions of people have been lost to, presumably, an unfriendly foreign power. This includes advanced security clearance forms of 127 pages (each) in which people divulge very personal information about their past and present bongs, liaisons etc. It also provides details of family relationships of people working in very sensitive (or secret) capacities.

That this happened at one of the most well resourced organisations on earth is striking but anyone even casually familiar with IT policies and procedures at large institutions might not be very surprised. If our personal data are stored by a large number of institutions, some or most of them careless and incompetent, sooner or later breaches will take place. I have a strong feeling that eventually the public will realize that it (and the general order) are under threat from this.

My first response was to think that personal data should simply not be available online and that businesses and governments should store their data offline, preferably in files in long hand. However, the genie is out of the bottle and any information that BigCorp or GovDept have can relatively easily be digitised and placed online. The better solution is to simply stop providing superfluous data to all the entities that currently ask for it. We need to provide far less data than we did 30 years ago!

Take your gym, for example. Their operational need is really for a proof that you paid the fee and some way (e.g. a photo) to ensure that two or more people do not use the same membership. At most, an emergency contact number might be included. They actually do not even need to know (and store) your name. I am looking forward to the start of a revolt against the personal information form...

OPM head blames old security, lax practices for cybersecurity breaches

OPM's archaic IT infrastructure opened door for massive data breach

Massive data breach followed 'long history' of failed IT systems at OPM

Sex, lies and debt potentially exposed by major data hack


The Economist on Kindle is not what it used to be

Reading magazines like The Economist in the Kindle app, with full pages (as in the printed version) in high resolution on its beautiful screen, is one of the reasons I even have an iPad. One of the disadvantages of this is the huge weekly download (slow on a South African connection at home) but I have always found it worth the trouble. The page images are no longer a pleasure to read however, as below.

Further, the plain text version (that flows on the screen and is nice to use on a smaller screen as well) now suffers from frequent bad formatting, as below. For me, "upsetits" is really confusing...

Finally, for some reason the date at the top of the page is now always "January 01, 1970" and although I can well imagine this to the fault of Amazon or of Apple, it is exceedingly silly!

If this does not improve soon, I shall seriously consider unsubscribing.  The Discover panel of the new Opera browser is really nice anyway and, for me, a Google News killer.


Does Uber have a payment problem?

A few weeks ago, friends and I used Uber to get a car from the station to a rugby game and were then stuck at the stadium in the dark because Uber had decided to suspend my account because of a (yet to be specified) payment problem. I wrote to them and was asked to send all manner of documents, which I did. Nevertheless, I was put off until Uber seemed like a really convenient option once more on a weekend. Nothing! My original card (copies of which had been sent after the first incident) did not work and neither did two others, on one of whom Uber nevertheless managed to make a R10,00 verification charge.

In total, I have wasted at least to hours trying to get Uber to accept my cards and I have now finally given up. In South Africa, Uber does not accept PayPal so there really is no way forward. Is it any wonder that Uber is apparently starting to accept cash payments in Hyderabad?

This leads to the following obvious question: do companies have any idea how many customers they lose through miserably bad services? How would they (and investors) know how many people just walk (actually, navigate) away and are probably lost forever? Is this problem worse for the Internet economy than for brick-and-mortar commerce?


Pretty good Internet – airport 2015 edition

Speedtest.net at domestic Wimpy in Johannesburg International Airport
I have finally found a place in South Africa where you can simply pay for instant fast Internet access – the airport! The AlwaysOn network (with some complementary access) which covers many facilities in South Africa delivered the excellent access stats pictured above while I was having breakfast at Wimpy. Pricing is volumetric, of course, and I do know that in half the world, free coffee shop Internet delivers the same... However, this is a champion south of the Limpopo.


Eskom-kragmageddon baie naby?

Dis maklik om te verstaan dat beurtkrag net van 'n breukdeel van die totale kragaanvraag kan ontslae raak. Ons doen immers steeds ons wasgoed, kook en verhit water, verkoel en vries ons kos, sny die gras en maak die swembad skoon, net op ander tye. Beurtkrag (gesien oor 'n hele dag, byvoorbeeld) verwyder net die las van daardie kragverbruik wat nie verskuif kan word na 'n ander tyd toe nie. Daar is dus 'n perk op die krag wat Eskom uit die vergelyking kan haal met onderbrekings. Wanneer dít die dag te min is, kan ons ineenstorting van elektrisiteitsvoorsiening soos ons dit geken het, verwag.

Dink só daaraan, van die totale aanvrag van A per dag, kan 'n breukdeel p deur beurtkrag gespaar word. Die maksimale besparing per dag is dan pA (p vermenigvuldig met A). Gestel Eskom skakel deurgaans 'n breukdeel x van die netwerk af. Per dag word dan xpA gespaar deur beurtkrag. Indien die totale opwekkingskapasiteit B (per dag) is dan kan ons oorleef mits

B  ≥ A - xpA

met 'n x tussen 0 en 1. Só 'n x bestaan mits die ongelykheid

A - pA ≤ B

bevredig word. Wanneer B so klein word dat dié ongelykheid nie meer geldig is nie, gaan beurtkrag nie meer help nie. Dan moet 'n hele gedeelte van die land (of die hele land) permanent afgeskakel word. Indien A groter word en/of B kleiner, is daar geen manier om dit te vermy nie.

Gegee die probleme met B is al hoop om p groter te maak (nie in die kort termyn nie, dink ek!) of om A kleiner te maak. Daarvoor sal 'n drastiese verhoging in prys voldoende wees. Myns insiens, hoe gouer hoe beter.

Vrywaring: hierdie is nie 'n ingenieursmodel nie en hoort nie aldus geïnterpreteer te word nie. Die vereenvoudigings het na my mening geen beduidende uitwerking nie.