The “elevator pitch”: not so uplifting?


Peter J. Denning is something of a computing hero to me. He formulated the concept of the working set, around which I based my MSc project and he has been personally very kind to me in reading and commenting on that report.

Elevator Pitch for Katie

Elevator Pitch for Katie (Photo credit: Marco Wessel)

So, the article has written (for the ACM), with Nicholas Dew, in which they debunk the idea of the “elevator pitch” as a key business tool is doubly interesting, as, in my “real” job, elevator pitches are regarded as essential tools of the trade:

To make this work, you need to re-interpret the pitch. It is not a transmission of information but an offer to havea conversation. It is often much easier to ask someone to join you in a conversation than it is to present a polished, sticky, commercial-grade presentation. A conversational pitch will get you closer to your idea being adopted.

Well worth reading the whole thing – it’s not long, though not quite of elevator pitch length either!

Big data to save the planet? A Question To Which The Answer Might Not Be No


British readers of a certain age may remember a groundbreaking TV series from the autumn of 1979 – The Mighty Micro – in which Christopher Evans discussed the impact of the coming microchip revolution. (The series was broadcast after Evans had died, aged just 48).

In many ways the programmes – from what I can remember (and I was an avid viewer) – rather underestimated the impact of what was to follow. But the last programme did – and still does – stick in the memory because of what seemed, and seems, like a hyper-optimistic prediction: that microchips could save us from war.

Essentially Evans’s view was that by hugely increasing computational power, micro-powered computing would allow us to accurately predict the outcome of military conflict and so prevent it (why start a war when you know you are bound to lose or if you do win you, and your domestic critics, know it will devastate your society).

There are a lot of flaws in this argument. One only has to think of the jihadist claim to “love death” to recognise that the certainty of defeat might not be deterent enough and the 2008 financial crisis also demonstrates that increased computing power might just create new ways to mess things up, not to solve them.

But, but, but… maybe there is something to it after all. This week’s New Scientist reports on the release of the “Global Data on Events, Location and Tone” (GDELT) data set and the way it has been successfully used by Jay Yonamine, then a PhD student at Penn State, to model the spread of conflict in Afghanistan.

Yonamine was able to successfully model how the conflict would spread through Afgahnistan using GDELT, which geolocates major news stories and uses natural language processing to store a very short summary of them.

Modelling how the conflict spread is not the same as predicting where the next jihadist inspired conflict will take place though, of course, but it may be the first step on being able to draw out undercurrents of news stories and issue early warnings. The key question is whether it can be an effective leading indicator.

Maybe the idea has promise. At the very beginning of my memories of the world are the events of August 1969 – when the British Army was drafted on to Northern Ireland’s streets to avoid a bloodbath. Just six months before no one would have predicted that would have happened – even if the tempo of civil disputation had been increasing and certainly no one expected them to stay on the streets, as they did, for the next 30 years. And more importantly, perhaps, nobody – beyond some zealots on either side – would have wanted either outcome.

Again, think of the 2007 – 2008 financial crisis. Could it have been foreseen as early as 2004? Certainly some politicians claim that it could – but how could you tell whether they were any good at prediction: Mitt Romney,a  pretty serious person after all, really believed that would be president even on the night of election day – does that mean everything he says is nonsense or just some things?

Uisng Google Trends to investBig data might help sort some of that out too. Another piece of research highlighted by New Scientist, and undertaken by Tobias Preis at the University of Warwick, Helen Susannah Moat and  H. Eugene Stanley of Boston University suggests that an investment strategy based on analysis of Google Trends could have made substantial sums over the 2004 – 2011 period (see graph).

Their abstract states:

Crises in financial markets affect humans worldwide. Detailed market data on trading decisions reflect some of the complex human behavior that has led to these crises. We suggest that massive new data sources resulting from human interaction with the Internet may offer a new perspective on the behavior of market participants in periods of large market movements. By analyzing changes in Google query volumes for search terms related to finance, we find patterns that may be interpreted as “early warning signs” of stock market moves. Our results illustrate the potential that combining extensive behavioral data sets offers for a better understanding of collective human behavior.

So, risking the wrath of John Rentoul, this could be A Question To Which The Answer Might Not Be No.

British government to mandate IPv6?


An illustration of an example IPv6 address

An illustration of an example IPv6 address (Photo credit: Wikipedia)

I did not vote for either party in the current coalition government in Britain and I doubt I ever will. But credit where credit is due – they have done quite a good job at beginning to fix (most) government IT procurement and actually made the claim of the previous Labour government that they were open to using free and open source software (FOSS) real.

Still they have not fixed all IT procurement – the politically driven “Universal Credit” project looks to me like it will make all the failed big procurements of the Labour years look like well-thought-through successes. The lesson is that when IT policy passes out of the hands of Francis Maude at the Cabinet Office – who has done so much to drive the politics on FOSS – then the government heads for disaster.

The latest, and possibly the biggest yet (as it might wreck the mobile data market if pushed too far) potential disaster seems to have been signalled in the “Queen’s Speech” – the formal outlining of the government’s legislative programme for the next year. In it the Queen, on behalf of the government, pledged:

In relation to the problem of matching internet protocol addresses, my government will bring forward proposals to enable the protection of the public and the investigation of crime in cyberspace.

This is to replace the so-called “snoopers’ charter” – a proposal, much like that from the previous Labour government (which the Tories scrapped within days of coming to office only to attempt to revive two years later), to force service providers to maintain records of all internet browsing and emails (records in the form of which computer interacted with which rather than the content of the communications) so that these might be accessed by the police and the domestic security service, MI5.

The revived proposal was squashed by the Liberal Democrats, the coalition’s junior partner, on “civil liberties” grounds. But now it seems that the Lib Dems have also been persuaded something needs to be done and so have backed an idea – agreed with the Tories – to give every device that connects to the internet its own IP address.

Great idea! Except – well, it just won’t work with the bulk of today’s internet.

It is quite difficult to know exactly what the government are thinking of, as the whole idea seems too cracked, but let us give an (entirely fictitious) example. Mr A. Docstudent has been accused of smuggling quails’ eggs into the UK but when the police raid his home they find nothing, they still suspect he’s been in touch with Mr B. I. G. Smuggler, the quails’ eggs kingpin, and that he has been sending emails from various university networks using the fake identity Ms N. O. Clu.

The problem is, without a record of when Docstudent authenticated his devices against the international Eduroam network they cannot even prove Docstudent was on a university campus at the time. But if everyone of Docstudent’s devices had a unique internet address then they could simply point to that and say “you used your Psion 3 to send that one, you used your Sinclair ZXMobile to send that one” and so on.

So an easy solution is to give every device a unique IP address – after all all your devices already have a hardware identification through the so called “MAC address” which is unique to your machine. Force retailers to log who has which unique address (which can be based on the MAC address) and you do have something of a nightmare of a register, but it’s simpler than the “snoopers’ charter”. Or we could just ignore that and just go for raids and test the MAC addresses of all seized devices against the unique identification.

And we even have a fully worked out way of converting MAC addresses into unique network addresses – via version 6 of the Internet Protocol (IPv6).

And this is where it all falls apart because no one, or hardly anyone, uses IPv6, no one, or hardly anyone, knows how to set IPv6 up and no one, absolutely no one, has shown any willingness to pay the costs of converting today’s IPv4 networks into IPv6 networks.

For good technical reasons – IPv4 has run out of addresses – there have been excellent reasons to convert to IPv6 for years but the fact that it has not happened tells its own story. If the government mandated that all new network connections had to be IPv6 then most ISPs would likely go out of business and goodness knows what would happen to the mobile phone network (where providers currently treat their networks as walled IPv4 gardens and use the private nature of their networks to keep bad things out and to contain contagions inside their own networks – having a unique, internet visible IPv6 address would likely shatter that).

And, in any case, I am not convinced it would work. All Docstudent would have to do is route Clu’s emails via a foreign IPv4 server and the IPv6 address would be shaved off – unless, that is, the government proposes to cut the UK off from the rest of the world!

Anyway, that is my reading of what is going on here and why it is likely to fail, leaving Home Secretary Theresa May with huge quantities of egg on her face and wasting a lot of public money if this is ever tried seriously. If someone knows a way they could do this without these problems, step right up and set me to rights!

How to frighten yourself


English: Different sites and outcomes of H1N1 ...

English: Different sites and outcomes of H1N1 versus H5N1 influenza infections. Based on Respiratory system.svg with annotations. (Photo credit: Wikipedia)

If you want to give yourself a good chilling scare then start reading articles on H7N9 – the mysterious and fairly deadly new strain of influenza that appears to be very widely spread amongst poultry in Eastern China.

No one seems very sure about how it is spreading (thankfully this appears to be just from bird to bird at the moment) or even which particular birds are the main vectors: there is some suggestion it could be city pigeons who are spreading it to poultry in markets, rather than the other way round.

But what is worse, far worse, is that if it did mutate and spread via human to human transmission we might be defenceless against it for a long time.

Now, if (or more likely, when) that happens the mutation may mean that it loses its current deadly force (as I understand it, at the moment scientists can only be sure that 20% of patients will recover – 20% have already died and the rest are ill – hopefully eventually to recover) but there is no guarantee of that. The 1918 flu did not lose its virulence on crossing into mammals and while H1N1 (swine flu) was not as bad as was once feared that was not because it was weakened on transmission.

Yet H1N1 has created a climate where politicians fear being accused of falling for scare stories and where the sort of viciously anti-science press we have in Britain would be the first to go on the attack if public money was spent to pump-prime anti-epidemic preparations. After all the Daily Mail still will not even acknowledge its despicable role in the anti-MMR scare which has caused a measles epidemic in Britain.

None of this is good.

Chemistry sets redux


As I remarked before, the main problem with today’s chemistry sets is that they are inane to the point of boredom. I bought my eldest daughter one a few years ago and we both gave up on it because it was so tedious.

But that does not justify Tesco’s claim that they are only for boys!

This storm has just begun…

The Reingold-Tilford algorithm revisited


hw5_ivmooc_rt_tree

hw5_ivmooc_rt_tree (Photo credit: marianocecowski)

A while ago, as part of early research into what became my MSc project, I wrote code to create and then draw red-black trees using C++.

To draw the trees I used the venerable Reingold-Tilford algorithm, which is more or less the standard approach. I wrote some blogs about it and pages here seem to come pretty high up in Google searches for the algorithm, so I get passing traffic regularly as a result.

But idly chasing these links has led me to a chapter from the forthcoming Handbook of Graph Drawing and Visualization edited by Roberto Tamassia which has a chapter on tree drawing by Adrian Rusu, which contains bad news for us Reingold-Tilford fan boys, as this summary from the book of an experiment comparing algorithmic performance shows (emphasis added):

• The performance of a drawing algorithm on a tree-type is not a good predictor of the performance of the same algorithm on other tree-types: some of the algorithms perform best on a tree-type, and worst on other tree-types.
Reingold-Tilford algorithm [RT81] scores worse in comparison to the other chosen algorithms for almost all ten aesthetics considered.
• The intuition that low average edge length and area go together is contradicted in only one case.
• The intuitions that average edge length and maximum edge length, uniform edge length and total edge length, and short maximum edge length and close farthest leaf go together are contradicted for unbalanced binary trees.
• With regards to area, of the four algorithms studied, three perform best on different types of trees.
• With regards to aspect ratio, of the four algorithms studied, three perform well on trees of different types and sizes.
• Not all algorithms studied perform best on complete binary trees even though they have one of the simplest tree structures.
The level-based algorithm of Reingold-Tilford [RT81] produces much worse aspect ratios than algorithms designed using other approaches.
• The path-based algorithm of Chan et al. [CGKT02] tends to construct drawings with better area at the expense of worse aspect ratio.