Evidently Beattie Communications, which does public relations work for the UK distributor of Powerizers, has very good blog monitoring software. I briefly mentioned the stilt/pogo-stick things in this blog yesterday (spelling them with an “s” rather than a “z”) and today Beattie has been on the case offering me a demo pair to try out.

When I was a child I had both a pogo-stick and a pair of stilts, so I figure I will be a natural. How hard can it be?

Hmm. I’ll let you know how I get on...

Today I heard for the first time about OpenStreetMap.org, a project that aims to do for maps what Wikipedia has done for encyclopaedias. The aim is lofty - to create a complete map of every highway, railway, road, river, canal, path, cycle-lane and track on the planet. This is, of course, a task well suited to the crowdsourcing approach, even if collecting the data is not without its hazards.

Parts of the map are currently quite sparse, but others - no doubt in the vicinity of the project’s 14,500 current contributors - already boast a level of detail and precision beyond that offered by commercial rivals such as Google Maps.

Although the mapping effort is well under way, alas the software is not really ready for prime time. Searching - even for roads evidently on the map - is a little hit and miss, and editing can be an exercise in bafflement. In the tradition of Wikipedia anyone can edit anything, but casual users are invited to do so using a particularly unintuitive Flash application called Potlatch. And dangerously, the editing process seems to lack a rollback function. As the army of Wikipedia change junkies have proven, being able to reinstate a known good state is essential to quickly annul the work of vandals or the clumsy. Updates are also only committed to the master map once a week, thereby failing to offer Wikipedia’s addictive instant-gratification hit.

The small mews in which I live was missing from the map, and it took me a little head-scratching to work out how to add it. Rendered overconfident by this effort, I attempted to add a nearby footpath, but in trying to correct a mistake I inadvertently erased a nearby road. With no rollback option I had to manually fix my blunder - hopefully successfully. This was enough to dissuade me from trying to add anything further until the software has been made a little more idiot proof. I’m not what you’d call a natural cartographer, it seems.

Late last month I was invited to attend a one-day workshop on the future of satellite navigation and mapping, organised by IT trade body Intellect, the Location & Timing Knowledge Transfer Network, ITS UK and the Royal Institute of Navigation. The event is due to be held at the Intellect Conference Suite in London on Wednesday 17 October 2007.

The workshop will examine some pressing issues, ranging from public over-reliance on the current generation of satnav devices and their crude route optimisations, through to the problems caused by the Ordnance Survey's management of the UK’s mapping data. At similar events in the past I've heard industry insiders mutter that the OS has more in common with a 19th Century book publisher than a modern distributor of digital data. “This workshop will look at how to cut the 18 months it can take to incorporate changes in the road network,” the organisers note, pointedly.

Invited parties include the Department for Transport, local authorities, mapping companies, device manufacturers, retailers and communications providers, so attendees will be assured of a lively debate

Having invited me, Intellect then evidently realised I was a journalist and said I couldn’t come, saying, “As this is a workshop I'm afraid it is not appropriate nor relevant for the press to attend.” I pointed out that these days anyone can blog, and after a couple of days the organisers relented, albeit with the condition that the meeting will be held under “more than Chatham House rules”.

The Chatham House rule means I can use information from the meeting but can’t attribute it to anyone in particular, or to the organisations they represent. I’m not sure what “more than” that rule means, but no doubt the chairperson of the meeting will explain - assuming the rules apply to everyone, not just to me...

Why the paranoia? Probably because there’s a big mess where a government policy on mapping should be. Local authorities, the National Land and Property Gazetteer (NPLG), the OS and Royal Mail each have a stake in mapping and tend to pull in different directions. For example, while OS handles publicly owned data it is supposed to be self-funding, so it charges for its services, while local authorities are required to give information to the OS free of charge. As another insider told me, there is huge resentment from the councils who are forced to provide information about roads and properties free of charge, but then have to pay when they want to use the same data to help fire engines and ambulances find their way. “If you want to see adults act like children put NLPG, Ordnance Survey and Royal Mail in a room together,” my informant tells me.

I’ll keep you posted. Or not, as the case may be.

Almost two years ago I signed up as a member of PipelineCard.org - an interesting experiment in online collaboration.

The goal was simple: to see if a large and loosely organised body of people, garnered mostly through online word of mouth (or should that be word of mouse?) could use their collective buying power to gain a discount from a major fuel retailer.

To date, half a million people have signed up. And the message from the scheme’s founders is, erm, not positive. A loosely organised body of 500,000 people can’t manage to secure a discount from a major fuel retailer, it seems.

“I won't bore you with the long list of broken promises, false hopes, and corporate inactivity,” wrote founder Ben Scammell. “The short version is that our original ‘partner’ let us down, we have spoken to several others and the industry has clearly closed ranks against us. What a surprise!”

It is indeed not much of a surprise. After all, what Scammell was proposing was a loyalty card scheme where the retailer suffered most of the losses associated with such things - having to give a discount - while failing to garner the big gain: ownership over members’ details and therefore opportunities to data mine or to upsell associated products such as motor insurance or loans. This leaves the main benefit being simple customer loyalty - a card member is more likely to buy repeatedly at forecourts that give a discount to members. But again, the retailer will see a dubious benefit as they will not own the scheme and could see that loyalty shift to a rival at a future date.

“We have learnt a lot along the way and one of the things we have learnt - the hard way - is that 500,000 members aren’t enough,” Scammell wrote. “We are now about to launch a new publicity campaign and we aim to double that figure within six weeks. We are very confident that one of the petrol retailers will break ranks - but only if we can get a lot more members.”

Will a million members be enough for the single benefit of loyalty to start outweighing the drawbacks from a retailer perspective? Maybe.

More likely is that Scammell and Co’s efforts have simply poked the sleeping giants - and the fuel retailers will put more effort into marketing their own loyalty schemes. For example, in May this year Shell replaced its existing PlusPoints scheme with a new Drivers’ Club offering discounts, Airmiles, and donations to charity or to CO2 reduction schemes.

I certainly admire the founders of Pipeline Card, but I fear their mission is a bit too Quixotic to succeed.

In the IT industry we’re all so used to hearing discussions about web 2.0 - like or loathe the label we all know at least vaguely what it implies - it’s easy to forget that lots of organisations still have a bit of trouble with web 1.0.

Yesterday I booked a hotel room, and was able to narrow down my search by looking at the enormous range of web resources that seem to be devoted to this topic. Not finding any suitable last-minute offers via the likes of LastMinute.com, I ended up trawling through the web sites of individual hotels, found via Google Maps. I’m going to a wedding reception and want a hotel that I can find when plastered, so proximity to the venue was paramount.

Most of the sites I looked at had online booking facilities, allowing me to narrow down my search to those that actually had suitable, vacant rooms at a reasonable price.

Eventually I got down to one hotel - good location, available rooms, affordable, and which wasn’t afraid to put detailed pictures of its rooms online. Visitor feedback on the various hotel-promoting web portals checked out.

At this point I would have happily clicked my way through the online booking form, credit card in hand, but for the fact that I wanted a room on the top floor (don’t ask). I couldn’t specify this online. So I dusted off my phone receiver and called the hotel.

I spoke to a very helpful receptionist who put me down for exactly the kind of room I would have booked online, only on the right floor. The surprise came at the end of the call. The rate I was quoted was much cheaper than online. A full 20% cheaper. I didn’t ask for a discount, I just got it.

Someone, somewhere, hasn’t got the hang of this web 1.0 thing. If I do the admin work - checking availability and filling in my credit card number, and don’t take up the time of a human being on the end of the phone, it’s supposed to cost me less. It’s not supposed to cost me more.

I’m going to have to think twice about this whole buying-things-online lark. 

Anyone familiar with the travelling salesman problem will know that computers are not generally very good at route optimisation. This is not for want of processing horsepower, but because routing problems can be challenging to solve algorithmically.

A to B route-finding is a lot simpler, as the existence of my TomTom satnav system proves. However, the algorithms and heuristics employed in this kind of A-to-B planning are not very good at providing useful alternatives to the optimal path, for those occasions when the optimal path happens to have been dug up by the council.

At a Royal Institute of Navigation seminar this week I heard Alan Jones of a startup called Camvit (Cambridge Vehicle Information Technology) present his firm’s work on this real-world problem. Camvit has developed software capable of finding a range of good routes between A and B, rather than one optimised route.

As Jones explained, humans planning a journey - with an old-fashioned paper map - will tend to look first at the available trunk routes, selecting a few options and ignoring the fine detail until the major routes have been selected. They will thus tend to chose from a range of pretty-good routes, favouring known roads and avoiding known problem areas.

Jones cited the example of travelling from Cambridge to Manchester, where there are five distinct routes with very similar travel times, variously involving the A1, M1, M6, M62 or A1(M).

Satnav software, the AA’s online service or the likes of Google Maps will typically provide the user with just one route. Coarse adjustments are then typically possible, allowing the user to choose a route that goes via a particular point, or avoids particular roads, or goes via the shortest route, or avoids motorways. But these approaches tend to be very poor at providing a spectrum of routes with reasonable travel times.

Other adjustments tend not to help either. For a long journey, displaying the second, third, fourth and fifth best routes will typically create variation only in the fine detail of the route, at the local roads level near to the ends of the journey.

Altering the weighting factors (such as favouring distance over time) will also not necessarily throw up alternate routes. Jones noted that where a northern route is, say, 300km long, and takes exactly three hours to drive, an adjustment to the weighting will never find a southern route that is 301km long and takes three hours and one minute, even if there is a nice pub on the way.

After a lot of work, Camvit has come up with a clever approach called Choice Routes that it says is quite good at finding the kind of routes humans would spot.

It works by pre-calculating a vast set of conventional, optimised routes between points on the map. This process creates a unique, many-branched tree - sprouting out in all directions - for every point of interest on the map.

To calculate routes between two particular points such as Cambridge and Manchester, the software first compares the pre-calculated Cambridge tree with the prepared Manchester tree, and removes all routes that the two trees don’t have in common. Given that the branches of each tree will tend to be roughly radial, with the town of interest at the centre, typically there will be very few branches of any length that the two trees have in common.

“A surprising thing happens,” Jones said. “We find that there are a few long chains of marked roads, and the rest is noise.” These few points of obvious overlap are then used as seeds, forming the basis of the detailed routes that the software goes on to calculate. Play the above video to see how it works.

This turns out to create human-like solutions using a method that no human could employ - kind of like the exhaustive way in which computers compete with grand masters at chess.

There’s more to the choice of several good routes than meets the eye. Knowing that a range of routes is available also helps after you’ve set off on one particular route, because different routes will tend to have pieces in common. All five good routes from Cambridge to Manchester, for example, depart along the A14. The Choice Routes software is able to determine, therefore, these decision points: junctions at which the driver has the option to switch from one good route to another.

The driver can thus be alerted when these points crop up, choosing the route that seems best at the time. Or, with software connected to real-time traffic information, the Choice Routes software might give updated estimates of the best route in a much smarter way than current systems.

I also think there’s a big safety implication. From my own experience with a TomTom unit, I know that route switching is the time when the driver is most tempted to try to interact with the satnav while still driving. When leaving a motorway due to congestion, for example, TomTom will point you back onto the motorway until you can find a way to tell it not to, and there are seldom convenient places to stop. So anything that automates the business of realising that a driver wants to switch routes looks like a giant leap forward in safety to me.

Camvit aims to sell its system to the big navigation software suppliers. “If the techniques prove popular, we would hope to see Choice Routes become available as standard, or at least as a ‘give me choices’ button, in most journey planners,” Jones told me.

“The technology is protected by a patent application, and we are in advanced talks with several companies about their markets and pricing models,” Jones added. “We are always keen to hear from other companies that think the technique might be useful to them. Although our initial focus is on road journey planning for the individual driver, we are also keen to look at applications to public service dispatch of multiple vehicles, fleet management, traffic and travel planning of all sorts, and even systems as diverse as internet routing, pipework planning or integrated circuit layout.”

As any IT project manager will happily confirm, if you take an information system designed for one purpose and then wrap it up and try to use it for another, unrelated purpose, you will soon find out that life is not as simple as you had hoped. There will always be small, non-obvious exceptions to the rules that you blithely assumed would apply...

I was late for an appointment at The Dorchester - the posh hotel overlooking Hyde Park. It’s walking distance from my desk in Soho but Mayfair is a maze and time was tight, so I wanted to be reminded of exactly where on Park Lane it sits. I turned to the trusty Google Maps service. Which turns out not to be so trusty.

Searching for “the dorchester hotel” brings up two candidates: “A” being the famous Dorchester Hotel of Beverley Road, Hull, plus the other, evidently lesser-known Dorchester that I was after.

Clicking on “B” brought up a confirmatory picture of the correct façade, so I zoomed down to a suitable resolution to show the hotel’s locale.

And was a bit surprised to find myself looking at Rathbone Place, round the corner in Soho, rather than a prime piece of Mayfair.

I happen to know that the Google Maps arrow is actually pointing to a Royal Mail warehouse. This is probably because Google Maps gives The Dorchester’s postcode as W1A 2HJ. I think this actually functions as a sort of PO box. You’ll get the same result, incidentally, if you put IT Week’s published postcode of W1A 2HG into Google Maps. That postcode is used to divert bulky parcels away from our plate-glass entrance on Broadwick Street. The actual physical postcode for our office is W1F 8JB.

Similarly, The Dorchester’s actual postcode, helpfully retrieved from its customer-facing web site, is W1K 1QA.

I wonder how many people are being sent to Royal Mail outposts by putting too much faith in Google Maps and its flawed postcode system? The same error will of course be made by any digital map relying on postcodes - my TomTom GPS system makes the same mistake.

Forget the national ID card system, we need a proper national digital address system designed for people, not parcels. 

Munching my cornflakes this morning, I watched a discussion on BBC’s Breakfast show dissecting Richard Branson’s decision to spend Virgin Atlantic’s profits on biofuels. In lieu of an actual expert, the environment correspondent from one of the broadsheets had been ferried in to the studio, to give the anchorpeople someone other than themselves to talk to. I was half asleep, so I can’t recall who this journalist was exactly, but I do remember that she trotted out figures about where aviation would be, in terms of the number of flights, by 2050.

I’m sure she used the numbers in good faith, and got them from a reputable firm of analysts or other professional air-finger interface experts. Personally, I find it amazing that anyone might think they could forecast aviation trends 44 years into the future. Forty-four years ago, in November 1962, the French and British governments signed the deal that led to the development of Concorde. No doubt a well informed contemporary expert would have confidently predicted fleets of supersonic (if not hypersonic) jetliners of all types and sizes would be plying the stratosphere by the turn of the century.

As we now know, Concorde turned out to be an expensive albatross, unsuited to the realities and economics of late 20th Century transport. Today we cross oceans barely faster than those who travelled by Boeing 707 in the 1960s. However, we do so in vastly larger numbers and - as the three [correction, five] 707 crashes of 1962 underscore - in substantially better safety.

Bill Gates is fond of pointing out that we tend to overestimate likely progress when we look at the short term of one or two years ahead, but underestimate progress over the longer term of one or two decades. Examples such as chip clock speed or transistor density, data communication rates or the price/performance ratio of servers tend to bear out this kind of observation.

However, there’s a little more to this kind of rule of thumb than vendors of products might like to admit. We also tend to overestimate the range of possibilities in the short term, and underestimate the scope of change - the number of forks in the road - when we look far ahead.

In the case of aviation, it would have been hard to predict the death of Concorde, the growth in wide-body jets, the arrival of cattle-class travel, or the ticketless, seatless, mealless and soulless EasyJet.

Personally, I think that business travel will have declined markedly by 2050. I suspect that the networked collaboration tools of 44 years hence will provide a convincing enough facsimile of reality to satisfy the human need to look someone in the eye. Businesses will cut out travel as soon as it becomes feasible - the decision will be simple economics.

But I also expect my own prediction to be pathetically wide of the mark, and that the road ahead will fork in other, unexpected directions. Perhaps meetings will be delegated to artificially intelligent avatars. Or perhaps suborbital scramjets will let people commute halfway around the world every day.

The only thing that is clear is that we will have to wait to find out for certain. In the meantime, investing in alternative fuels seems like a sound business plan to me.