Saturday, July 19, 2008

British 10k

Two weeks ago I managed my first real run since March 2007 when I trashed the meniscus cartilage in my left knee. It felt good to be lined up in what I was counting as a race. I arrived a little late (who'd have thought the roads would be closed even to bicycles!) and so crossed the start line pretty much dead last. It was by no means a fast run at 52m 56s, but that was a good 7 minutes faster than I'd managed 10k in a training run the week before.


The organisation was strong as it has been when I've run this race before. One small improvement that I'd like them to make is to publish a proper map of the course before you run it. Despite checking the pamphlet they sent and all sorts of websites, all I could find was a curious isometric perspective map that didn't really allow you to trace the route. With my GPS tracking 10.35km at the end (and I can't remember taking a wrong turn) I wasn't absolutely sure when the finish line was going to show up - there are a couple of double-backs (westminster bridge, parliament square and whitehall in the last couple of km). It would be nice to see it marked clearly on a map.

Is this the start of the Road To 3:30?

I am ... I want to be ...

Sometime in early 2003 I gave a presentation to the DWP where I talked a lot about my usual topic of too many government websites. I'd taken a look at DWP sites in the run up to the conference and become very confused by all the different sites (all with different branding, different layouts, over-lapping information and so on) - this wasn't an uncommon problem then, either in the UK or globally, and, whilst I don't look too often, I suspect it's still pretty common.

This is one of the slides I put up:


I played around with the "I am a" and "I want to be a" concept - maybe people come to websites as unemployed and want to be employed, maybe they come as young unmarried folks and want to know what happens when you get married, maybe they come as employed and want to know what happens when they're retired. I pictured it as a set of dropdown lists that you could select from. There would have been all sorts of problems implementing it that way, but the essence of the point was that people come to a site wearing one hat but want to know about something else - and the path from one to the other might be what they're interested in - and we'd need to think hard about how to present that in a simple way, and it might need some data (that we'd need to keep hold of to improve the experience next time)

I hadn't seen any implementation of this at any level, until today when I happened on this:


It's on the Civil Service's own website and it makes perfect sense. You may visit because you're a civil servant and you want the latest news, or you may want a job - in which case, sadly, they send you to a new site built with a different engine, with a different layout, a different search engine etc.

Plainly they weren't thinking what I was thinking, but nonetheless, it raised a wry smile this morning amidst the gloom.

Sunday, July 13, 2008

O Day


Zero day exploits are written about all of the time - there are 2.93 million hits on google for "zero day" and 314,000 for zero+day+exploit. Saturday of this weekend was the UK's first O day (and the ones exploited were early iPhone adopters):

- O2's servers fell over under the load of iPhone 3G activation. An eminently predictable event - one that should have seen months of planning - resulted in a Black Swan day for O2. The probability of high take up of the new iPhone was known, but the consequences were not well predicted. Nassim Nicholas Taleb can add this case study to his next book. O2 can perhaps be known just as "O" for the next few days.

- Oyster cards across London fell into disuse as the central unit that operates them fell over. I had no idea these were run by some central server - I always figured that the card only talked to the local reader during the transaction and that perhaps there was some kind of bulk upload of transactions periodically or even once a day. Every system has a bad day - and Oyster's seem few and far between - but for a multi-hour outage to occur? (Don't worry folks, David Frost would have been unaffected - Freedom passes are still paper-based)

Olympics organisers had better watch out. These are good examples of how not to demonstrate the capability of our infrastructure to respond to high load events.

Wednesday, July 02, 2008

More on trending e-government

There were a couple of comments on the last post asking about the google trends data I presented (one asked if it was fair to compare with and perhaps i should have used a state portal; the other asked what the 180k visitors might actually be doing on the site). Here's some more data to perhaps cloud the picture showing average daily traffic over the last year for and


This graph comes from, who - to help answer the "what are the people doing when they visit" question - also say that these are the keywords that get visitors to the site:


Some definitions: share is the percentage of the total, engagement is the ratio of time spent on the site having used that keyword (so the word that generates the most stickiness is 100, in this case that's "access to work") and effectiveness is a kind of visitors * time spent combination so that you can determine the most effective keyword.

There's a BIG caveat with this data as far as I can tell. And it comes from this text on the site "Compete ranks the top one million websites in the U.S. based on the number of People the domain attracts each month" - now I don't see why anyone from the USA would be visiting these sites (ok, some folks might, if they were considering visiting the UK or moving here perhaps with their employer, hence "access to work"). And in the FAQs, this is said

Compete estimates site traffic and engagement metrics based on the daily browsing activity of over 2,000,000 U.S. Internet users. Compete applies a rigorous normalization methodology, leveraging scientific multi-dimensional scaling (by age, income, gender and geography) to ensure metrics are representative of the U.S. Internet population. Compete members are recruited through multiple sources, including ISPs, the Compete Toolbar and additional opt-in panels to ensure a diverse distribution of user types and to facilitate de-biasing across the data sources.

So maybe there are a bunch of UK people who have installed the compete toolbar widget? The idea that Americans might be searching a UK site for "disability living allowance" seems far fetched.

As another angle, here's some data from Alexa - I don't know that this helps or hinders (it measures "reach" which is apparently a combined view of page views and unique users, again using a toolbar that you download and that collects data on sites you visit - they don't say how many people have downloaded said toolbar). This appears pretty consistent with the Google Trends data in the last post.


On the "Is it a fair comparison?", I don't know - I'm game to try other ones. suggest a few sites that I should compare and I'll post the results here.
And, to help answer "What are they doing" ... this is what google says are the top few search terms for - none of them are "access to work" you'll see. The plain conclusion from these google results is that Road Tax is by far the most attractive offer for

200807021808.jpg Some other stats that I have - from a very short sample period in January 2008 - show the following as the top 20 terms that people used to get to from external search terms. The sample period is only a few days so the numbers aren't very relevant ... but they do show that "Road Tax" isn't the biggest thing, although there's no question it's a big driver (ha!). The sum of these visits is just over 100,000 and about 20% of those are car related (including theory test etc). What is interesting is that people use google to search for "direct gov" and "" (and there are appearances lower down the list of and even - and I suspect that this somewhat discredits the view of another commenter that the "e" in e-government stands for elitist! There are also plenty of people who searched for and found themselves at

200807021822.jpg 200807021834.jpg

Finally, to help the "What are they doing?" point,next to the external search engine figures I've put the top 20 search results from the internal search engine - i.e. what people searched for once they arrived at the site (from the same period as the figures above). Again, car related topics feature often.

Does any of that help?


I went to a new building the other day ... the meeting room where we sat was called "John Lennon" ... Afterwards I went looking for Paul, Ringo or George ... and found only Heathrow, Gatwick and City. Not even a JFK or Da Vinci.

There wasn't even a seance in the dark.