Where do all the comments go?

I don’t get many comments on this blog: fortunately I do it for the opportunity to write amid the torrent of administrivia rather than for the affirmation. But I do get more than’s readily apparent, which tells us something about social media and opens up an opportunity. Blogging is up there with Twitter and Facebook as the epitome of social media: a way to share parts of your life and mind digitally with more-or-less unselected parts of the world. You could argue that it’s also the epitome of egotism to think that anyone else might want to share those parts of your life and mind that you foist upon them, but that’s another story. The social media space has consolidated somewhat to yield several large services with a surrounding cloud of also-rans, which is pretty much what you expect in a market that places a premium on networking and connectedness: by definition one large network is more valuable (in many senses) than several smaller ones. But it shows no signs of consolidating into a single hegemony that captures all the aspects of social media we’ve come to find: Twitter and Facebook are distinct but linked, and likely to remain that way. The same is true for blogging platforms. The reasons for this are complicated but relate to the ability of one company to climb several steep learning curves simultaneously. It’s more efficient to focus on innovating in one area — climbing the single curve quickly — rather than trying to provide a unified experience that involves multiple areas of innovation. This is Adam Smith’s pin factory for the knowledge economy. This economically attractive choice has consequences, though. Since there isn’t a single network, one needs to access multiple networks to reach the widest community. Many people — including me — do this by tying their services together. My Facebook status is updated from my Twitter feed, for example: I almost never update it directly. The same is true for my LinkedIn status. My web site has the Twitter stream and the blog side by side, and if I post to the blog a notification goes out of on Twitter. Actually, if you’re reading this at all, you probably know at least some of this by virtue of how you got here :-). This is the socialisation of social media, in effect, but it means that what one thinks of as a single conversation is actually going on in several places simultaneously, and that’s where we need to ask where all the comments go. I get very few direct comments on the blog, despite some posts getting quite a lot of readers: the conversion rate of readers to commenters is significantly less than 1 comment per 500 distinct page views. However, that’s not the whole story, because I also get comments by the other social media channels: as comments on my Facebook page, as replies and retweets of the announcements, as email from people who see these, and so on. That makes it hard to measure the impact of a post as measured by the comments it generates. The popularity of a post isn’t what’s important to me (although they’re always nice to get, of course), but it does mean that a significant benefit of social media — the ability to engage in a conversation over people’s opinions — is impeded by the fractured nature of the comment stream, even when it’s being aided by wider distribution. So I think there’s an opportunity here for someone: how can we aggregate comments? We aggregate blogs themselves using RSS feeds: can we do the same for the comments, and feed them back to a single point (the blog itself, perhaps) so that they’re available for further comment regardless of the channel on which they were issued? It’s not a trivial task, because you need to identify which comments relate to which blog posting, but it’s certainly not insurmountable. Might make a good final-year undergraduate computer science project. Actually this sort of technology is useful more broadly. Any web site or book that attracts comment is likely to attract them in different fora: how do you find them? How do you bring them to one place? This has both intellectual and commercial implications: intellectually it increases the network of the social media comment space and so should make it more efficient; commercially it improves the means for assessing the popularity of a web site which is often what drives advertising revenues. Overall it would let the increasing specialisation of social media continue without this irretrievably fragmenting the very things that make it most valuable: the two-way flow of ideas.

Who invented meringue?

What the invention of complicated foods tells us about discovery, innovation, and university research funding. Over lunch earlier this week we got talking about how different foods get discovered — or invented, whichever’s the most appropriate model. The point of the discussion was how unlikely a lot of foods are to have actually been created in the first place. The lineage of some quite complicated foods is fairly easy to discern, of course. Bread: leave out some wet flour overnight and watch it rise to form sourdough. Do the same for malt and you get beer (actually the kind that of beer that in Flanders is called lambic). Put milk into a barrel, load it onto the back of a donkey and transport it to the next town, and you’ll have naturally-churned butter. It’s fairly easy to see how someone with an interest in food would refine the technique and diversify it, once they knew that the basic operation worked in some way and to some degree. But for other foods, it’s exactly this initial step that’s so problematic. I think the best example is meringue. Consider the steps you need to go through to discover that meringues exist. First, you have to separate an egg — which is obvious now, but not so obvious if you don’t know that there’s a point to it. Then you need to beat the white for a long time, in just the right way to introduce air into it. If you get this wrong, or don’t do it for long enough, or do it too enthusiastically (or not enthusiastically enough) you just get slightly whiter egg white: it’s only if you do it properly that you get the phase change you need. Of course you’re probably doing this with a wholly inappropriate instrument — like a spoon — rather than a fork or a balloon whisk (which you don’t have, because nobody knows there are things that need air beating into them yet). Then you need to determine, counter-intuitively, that making the egg white heavier (with sugar) will improve the final result when cooked. Then you have to work out that cooking this liquid — which has actually to be a process of drying, not cooking — is actually quite a good idea despite appearances. It’s hard enough to make a decent meringue now we know they exist: I find it hard to imagine how one would do it if one didn’t even know they existed, and furthermore didn’t know that beating egg whites in a particular way will generate the phase change from liquid to foam. (Or even know that there are things called “phase changes” at all for that matter.) Thinking a little harder, I actually can imagine how meringues got invented. In the Middle Ages a lot of very rich aristocrats competed with their peers either by knocking each other off horses at a joust or by exhibiting ever-more-complex dishes at feasts. These dishes — called subtleties — were intended to demonstrate the artistry of the chef and hence the wealth and taste of his patron, the aristocrat. Pies filled with birds, exact scale models of castles, working water-wheels made out of pastry, that kind of thing. In order to do this sort of thing you need both a high degree of cooking skill and a lot of unusual food-based materials to work in. You can find these as part of your normal cooking, but it’s probably also worth some experimentation to find new and unusual effects that will advance this calorific arms race a little in your favour. So maybe meringue was invented by some medieval cook just doing random things with foodstuffs to see what happens. The time spent on things that don’t work — leaving pork fat outside to see if it ferments into vodka, perhaps? — will be amortised out by the discovery of something that’s really useful in making really state-of-the-art food. Contrary to popular belief the Middle Ages was a time of enormous technological advance, and it’s easy to think of this happening in food too. So food evolves under the combined effects of random chance operations shaped by survival pressures. Which is exactly what happens in biology. A new combination gets tried by chance, without any anticipation of any particular result, and the combinations that happen to lead to decent outcomes get maintained. At that point the biological analogy breaks down somewhat, because the decent outcomes are then subjected to teleological refinement by intelligent beings — cooks — with a goal in mind. It’s no longer random. But the initial undirected exploration is absolutely essential to the process of discovery. Bizarrely enough, this tells us something more general about the processes of discovery and innovation. They can’t be goal-directed: or, more precisely, they can’t be goal-directed until we’ve established that there’s a nugget of promise in a particular technique, and that initial discovery will only be performed because of someone’s curiosity and desire to solve a larger problem. “Blue-skies” research is the starting point, and you by definition can’t know — or ever expect to know — what benefits it might confer. You have to kiss an awful lot of frogs to have a reasonable expectation of finding a prince, and blue-skies, curiosity-driven research is the process of identifying these proto-princes amongst the horde of equally unattractive alternatives. But someone’s got to do it.

Sensor and sense-ability

I’m talking at the BCS Edinburgh branch tonight about sensing and sensor-driven systems.

It has been an old maxim in computing that incorrect inputs can acceptably give rise to unacceptable outputs: “garbage in, garbage out”. This is ceasing to be true, and many classes of systems must behave predictably even in the face of inputs containing substantial garbage — although researchers delicately use terms like “imprecise” or “unstructured” instead of “garbage”. In this talk we discuss some approaches to managing the problem of imprecise, inaccurate, untimely and partial inputs in the context of pervasive and sensor-driven systems, and suggest that we need to re-think radically the way we build software and represent decision-making in these environments.
Details of the venue are available here.

Book: Tumbledown — the old houses of Co Sligo

The west of Ireland is full of old, decrepit buildings that seem to have been left behind by the Celtic Tiger years — and just crying out to be photographed. Over summer 2010 I spent a happy few days touring around Sligo and nearby counties taking shots of the half-hidden, half-derelict houses hidden in the undergrowth. Then I discovered on-demand book publishing, so now the collection is a book.

Tumbledown by Simon Dobson
The publishing service is called Blurb, and lets you download a desktop-publishing program that can be used to generate books from a range of different templates, including some that are designed for art photography. The quality of the printing is excellent — far better than results I’ve had from other similar services. What’s even better is that Blurb hosts a bookshop from which anyone else can buy a copy of your book, so if you’d like a copy just follow the link above. I have a feeling this sort of vanity publishing could become quite addictive. I’ve thought of half a dozen other photo projects that’d make decent books — undoubtedly more popular than any of my technical writing, anyway. For art books it’s clearly much more attractive than an ebook; and while it’s more expensive per book than traditional publishing, the huge advantage is that you don’t have to pay for, stack up and distribute inventory beforehand.  It certainly reduces the barriers to entry for stroking your own ego, anyway :-)

Call for papers: Formal methods for Pervasive Systems

We are looking for papers on the application of formal methods to pervasive computing, for a workshop at FM‘11. (Full disclosure: I’m the keynote speaker.)

Formal Methods for Pervasive Systems [Pervasive@FM2011]

A workshop held as part of FORMAL METHODS 2011 DEADLINE: 20th March 2011


Logics, Process calculi, Automata, Specification languages, Probabilistic analysis, Model checking, Theorem-proving, Tools, Automated deduction FOR THE privacy, behaviour, security, reliability, interoperability, context-aware, mobility, resource requirements, temporal ASPECTS OF pervasive healthcare systems, sensor networks, e-commerce, cloud computing, MANETs/VANETs, telephony, device swarms, electronic tags, human-device interaction, etc.


Our aim is to have productive discussions and a true workshop “feel”. Thus, we invite two kinds of submission:
  1. Original research papers concerning any of the above topics; or
  2. Survey papers providing an overview of some of the above topics.
Submissions should be written in English, formatted according Springer LNCS style, and not exceed 20 pages in length. Submissions must be made via Easychair. Our aim is for an informal proceedings based on these submissions to be available during the workshop. Depending upon the success of the workshop, we intend to produce an edited book based (at least in part) upon the contributions or develop a special issue of a journal.


Submission deadline:          20th March 2011 Notification of acceptance:    1st May 2011 Pre-proceedings version due:  20th May 2011 Workshop:                     20th or 21st June, 2011


Simon Dobson (School of Computer Science, University of St. Andrews)


Michael Fisher    (University of Liverpool, UK) Brian Logan       (University of Nottingham, UK)


Natasha Alechina    (Nottingham, UK) Myrto Arapinis      (Birmingham, UK) Mohamed Bakhouya    (Belfort, FR) Doina Bucur         (INCAS3, NL) Michael Butler      (Southampton, UK) Muffy Calder        (Glasgow, UK) Antonio Coronato    (CNR, IT) Soren Debois        (Copenhagen, DK) Giuseppe De Pietro  (CNR, IT) Marina De Vos       (Bath, UK) Simon Dobson        (St Andrews, UK) Michael Fisher      (Liverpool, UK) Michael Harrison    (Newcastle, UK) Savas Konur         (Liverpool, UK) Brian Logan         (Nottingham, UK) Alessio Lomuscio    (Imperial, UK) Ka Lok Man          (XJTLU, CN) Julian Padget       (Bath, UK) Anand Ranganathan   (IBM, USA) Alessandro Russo    (Imperial, UK) Mark Ryan           (Birmingham, UK) Chris Unsworth      (Glasgow, UK) Kaiyu Wan           (XJTLU, CN)


Natasha Alechina    (University of Nottingham, UK) Muffy Calder        (University of Glasgow, UK) Michael Fisher      (University of Liverpool, UK) Brian Logan         (University of Nottingham, UK) Mark Ryan           (University of Birmingham, UK)