Cnut, Jenkins and the law of Brownian motion

I wonder if they still do the experiment with smoke at schools to observe Brownian motion. Essentially you peer down a microscope at tiny particles suspened in air and track the complex paths that these particles follow. In doing this you are following in the footsteps of scientist Robert Brown who gave his name to these paths. What causes the complex shifts and changes in direction is the result of many tiny collisions at the microscopic level. Imagine a huge gang of people surrounding and running together towards a giant beachball. The ball will move but you won't be able to predict the direction.

The same effect can be observed in the media coverage of the MP's expenses scandal. Journalists are pushing in one direction, MPs another, the public have joined in, lobbyists and activists are running to the scene. There's an election so minor parties, PPCs, and councillors all want their piece of the action. The police, the speaker and the CPS have all been sucked in. The story is bouncing, not at random but in complex and unpredictable directions. 

So it is impressive that the electoral reform lobby managed to judge the moment and move the story in their direction. This is only possible because so much work has been done with the political class over the past ten years to embed the possibility of reform. Once it was mentioned, many people clutched at the issue as a way to move the story on. This helps journalists who are becoming bored at yet more duck islands and it helps MPs who want to talk about something, anything else. So there was a lurch in that direction but everyone's caught up now and so the Brownian jumps start all over again. 

Cnut isn't everyone's choice of poster child for public relations but he is well worth hanging above your desk. You will recall that he proved the limits of his wordly power by ordering the tide to stop. To no effect. Everyone caught up in the current maelstrom should remember that we can ride and harness the tides but we still cannot control them. 

David Cameron hopes that he can reap the whirlwind of reform without having to deliver proportional representation. Historically the Tories have been able to swat away the PR lobby with a quick comparison to Italy but Tony Blair and Roy Jenkins have provided a viable model. Alan Johnson has already called for a referendum on this model. It falls well short of what the Lib Dems seek but that wouldn't stop them supporting some reform rather than none.

None of us can really see where this is going. Most of us don't have to make calculations based on the unpredicatable path of an out of control public narrative but the leaders of our political parties do. Let's hope they remember Cnut and Brown as well as Jenkins.

Image is Cnut the Great and is public domain

We need a fair electoral college

Last night, you very possibly, like me, took part in the most important expression of European, if not global, democracy. Across time zones and across borders of nationhood and identity, all eyes were turned towards a single purpose: voting for the greatest song in Europe. It was an amazing spectacle, certainly, and managed to involve citizens in an active discussion and debate about their future (or at least about the costumes worn by many of the participants). 

Even so, I cannot have been alone in worrying about the distorting nature of the voting system. Can I?

Eurovision uses a simple electoral college. Each country has 58 points to allocate and they allocate these on the basis of a popular vote combined with a jury system. The jury system is an unusual democratic innovation but one that I will gloss over in this post. What I want to focus my attention on is the electoral college and how it can be reformed. 

The problem with the electoral college is that every country has the same number of points irrespective of their population. So Andora (population 67,000) carries the same clout as Russia (population 142,000,000). This is a massive disenfranchisement of the population of Russia at a time when we should be encouraging democratic values in that state.

The simplest solution is to award points in the electoral college on the basis of population size. There are 2,436 votes available. If these are allocated to countries based on population size (relative to the whole Eurovision population of 758,000,000) then Andorra receives 0.22 points in the electoral college and Russia 417. Had the votes been weighted in this way last night, then Norway would still have won but Azerbaijan would have gained second place, Turkey third, UK fourth, pushing Iceland (last night's runners up) into fifth. 

There would still be an unfairness in this system because countries cannot vote for themselves. This would disadvantage larger nations because they could award more points than they could receive. 

I have modelled a simple correction for this bias. Take the largest nation (Russia) and the next largest (Germany) and identify how many points each would award in place of today's 12 points (94.44 for Russia and 54.7 for Germany). The difference is the bias against Russia, so this amount is deleted from Russia's electoral college points and distributed equally between the remaining 41 countries. Then the next two countries are considered (Germany and Turkey) and the difference distributed between the remaining 40 countries and so on. 

Once this correction has been established, Andorra now has 3.22 votes in the Electoral College compared to Russia's 416.7 votes. If these corrected votes are applied to last nights scoring then, sadly, it makes no difference at all to the top five places.

Of course this is really an imperfect way to correct for population biases. A more satisfactory model is provided by the Qualified Majority Voting once widely used in the European Council (though being edged out because of the power of veto it can provide to smaller countries). Under QMV each country has a vote as a country, and a number of votes proportional to its population. To win a majority a vote has to have the support of more then 50% of the countries and more than 50% of the population. Eurovision does not lend itself to this system but I have adapted the model into what I choose to call Modified Single Transferable Qualified Majority Vote.

Under this system each country votes as at present, awarding 1,2 3, 4,5 ,6 ,7 8, 10 or 12 points. However, these points are not totalled, they are treated as a ranking system and considered in stages. 

First we look at all the countries awarded 12 points and award votes for these countries based on the population of the country awarding them 12 points. Then we add all of these points up and see if any country has the support of 50% of the voting public of Europe.

In last night's contest, Greece was awarded 12 points by only one country: Albania. In stage one Greece would be awarded 3,751,000, which is the population of Albania (adjusted using the method detailed above to correct for bias against large populations). 

Once all the 12 point votes were added, no country had passed the 50% mark, so the 10 point votes were considered. Once these are totalled and added to the 12 point results it is clear that Norway has the support of over 50% of the viewing public. 

That's not the end of the story of course. We also have to look at how many countries awarded them 12 points, because they need a majority of countries voting as well. Well 16 countries awarded them 12 points and nine countries awarded them 10 points.

So, even with a fairer voting system, Norway would still have won and the UK would still have come fourth. 

Sorry Jade, I did my best.

Watch twitter mess up and shudder with horror

OK so twitter is the shiny new technology and though I'm there (follow me at twitter.com/likeaword) I'm still unpersuaded of its transformative power. 

Whether or not we all become sucked into the twitterverse, the system is currently providing a salutory lesson in how not to handle service changes. On Tuesday 12 May Biz Stone (the uber-twitterer) posted an announcement to the twitter blog about a change in service. He said:

We've updated the Notices section of Settings to better reflect how folks are using Twitter regarding replies. Based on usage patterns and feedback, we've learned most people want to see when someone they follow replies to another person they follow—it's a good way to stay in the loop. However, receiving one-sided fragments via replies sent to folks you don't follow in your timeline is undesirable. Today's update removes this undesirable and confusing option. However, receiving one-sided fragments via replies sent to folks you don't follow in your timeline is undesirable. Today's update removes this undesirable and confusing option.

The detail of this change will not interest you in the slightest unless you are already a keen twitterer and to explain it to you would throw you into a moribund state even more promptly than my usual witterings. 

What's important (in this blog at least) is to note that this is portrayed as a service enhancement. It is not based on what people want, it's based on what people need. 

So that's mistake one. People like to be consulted. And when you've asked them what they want you can say "loads of people told us that". Rather than "this will be better for you".

People were severely not chuffed with this change. And, this being social media land, started talking about how cross they were.

Mistake two is revealed by the next post added to the blog by the same Biz Stone one day later. In this post he explains:

The engineering team reminded me that there were serious technical reasons why that setting had to go or be entirely rebuilt—it wouldn't have lasted long even if we thought it was the best thing ever.

So it wasn't a service enhancement, it was a technical issue. And in 24 short hours they have annoyed a huge proportion of their userbase, generated a huge amount of ill-feeling across the twitterverse and prompted untold obscure bloggers like myself to heap our ill-informed bile upon their heads.

So the lesson is this: if your change is a technical necessity say "we're terribly sorry but the techs tell us we've got to do this, this is what we're going to do to try and fix it for you" if your change is a service enhancement, ask your service users first. And always, always decide which one it is before you announce it.

Local authority, rocket scientist or uber-cool web 2.0 business, the rules are always the same.

Image is Physics Teacher by | E = mc2 |™ released under CC-SA-2.0

Police make noise to no discernable effect

Is burglary going to go up because we are in a recession? The Police Federation certainly thinks so. They've published a research report packed with graphs and everything. On the basis of their research they conclude that property crime will rise and that the only way to tackle this surge is to put more bobbies on the beat. 

A cynic might suggest that, had the Police Federation undertaken research into potential crime levels and concluded that we needed fewer police, they would have been less likely to publish this research with a fanfare. "National organisation issues self-serving news release" is hardly enough to generate a post on this blog. 

But there is something interesting going on.

First things first. Is crime going to rise due to the recession? Last year, Tony McNulty (who was then a Home Office Minister)  said that to suggest that some crimes would rise as a result of economic slowdown

"really is a statement of the blindingly obvious"

But can this be so? If crime has an inverse relationship to economic growth then all we need to do to tackle crime is make the economy grow. Things are clearly more complex than that.

The Police Federation report models the impact of several factors, including how much people are buying and how much crime increased (or decreased) last year. They say that three quarters of the historic variation in crime can be explained by knowing the levels of these things. I'm pretty sure that they are stretching their statistical tools here but that's really a job for Ben Goldacre. I have other fish to fry.

It does seem that the police and the government are at one in agreeing that crime will increase over the next couple of years. They are not at one in agreeing what to do about it. The Government has chosen today to release its revised Crime Strategy. Actually the broad strategy is unchanged. 

The Government wants partnerships at a local level so that resources are targeted and the right people with the right skills are in place. They also want new approaches, performance management and a drive for efficiency.

The Police Federation wants more coppers.

Now don't get me wrong, I'm all for having more bobbies on the beat. But this is not going to be the way to get them. As happens so often, you have two parties in a debate arguing about different things.

The police service, broadly speaking, thinks that its primary job is to nick criminals and to reduce crime. That's what they talk about. That's what they use to scare ministers.

Sadly the government thinks that the police's primary job is to make the population feel safe.

If the Federation is serious about influencing policy rather than seizing a couple of headlines, it needs to swallow what it cares about and start talking about what the government wants to hear. 

But maybe it doesn't actually want to deliver, just to make a bit of noise.

Shropshire, devoid of Chief Executives

At the risk of being a back-seat driver I can't help noting that my local Council is prepared to pay up to £180,000 for their new Chief Exec. I don't really care about the executive pay debate but people do and the local media are going to get quite excited about this. So it would be prudent to do some publicity to explain who the Council is looking for and why they deserve that amount. I can't find any evidence that they've done that, which they may come to regret.

And while we're on the subject. Neighbouring Telford and Wrekin Borough Council is also going to be looking for a new boss. The old one is leaving. He earns £160,000. Looks like his successor might expect a little more.

And are there really two high calibre £160K plus-worth candidates desperate to come and work in lovely (though sleepy) Shropshire? I guess we'll find out.