Classifying Choice

Two things have been slightly dominating my thoughts in the last day or so. At first glance they have nothing to do with each other: but then again…

The first is is the problem of classifying all the posts on this site. There’s a wealth of material but we increasingly feel that people aren’t aware of everything that there is. We create sometimes 20 new posts a week (not including events) and most have some kind of long-term interest: the recipes, reviews, interviews, wildlife notes and information about the companies we promote, for instance.

Each post can be given a tag: or – and here’s the problem – more than one. A post could be about a dozen, or five dozen, different things, depending on your point of view and how specific we want to be in describing it. Some things are obvious: a post might be mainly aimed at children, for instance. If it involves sports, should the sports be listed? If football is one, is it useful to have, at the end, a prompt for an article that mentions Swindon Town FC’s recent relegation? For some people it might be great; for others, it would be a tedious irrelevance.

Many of the people who visit this site do so because they’re prompted to click on a link in the newsletters or on social media. Once they’ve read that, what might they be in the mood for next? Something similar, something complementary, something opposite or something unrelated? One obvious technique is, at the foot of each post, to draw attention to others that might be of interest. Should a recipe involving rhubarb, for instance, prompt people to visit other recipes for desserts, or other recipes using similarly seasonal ingredients, or an article about growing rhubarb, or an article about a pub that specialises in English puddings? Or do we assume that people have now had enough of rhubarb and would rather do a quiz, read a literary parody, watch a video of a cat falling off a table or find out about music classes? How many other posts should we prompt? Four? Forty? There’s no way of knowing.

There needs to be a logical structure to the process of applying tags. It’s impossible, though, to create a list of such tags and say ‘that’s it’. New ones will always need to be added. One might have ‘Christmas’ but have forgotten ‘Easter’. Adding ‘Easter’ to the list of tags is easy: harder is then to think back over every post and make sure that it’s applied as consistently as ‘Christmas’ was. The addition of each new tag casts the whole previous classification system into doubt. As more than one person works on this, there will be different ideas about what is and is not a useful tag to add to a particular article. The application is subjective in any case. So too are the ways the people searching might use a term. ‘Christmas’ and ‘Easter’ are quite easy. What about ‘self-improvement’, ‘water’ or ‘humour’? Are these things that might need to be grouped or flagged, assuming it’s possible to be sure which posts covered these themes? If so, what else? Where would it all end? To pick one example, how should this article, which covers a number of things including classification, be classified?

Another possibility is that people might want to search using negative criteria. “Give me a list of events in December that have nothing to do with Christmas,’ for example, or ‘show me all classes or courses except art classes which don’t meet on a Thursday’. Maybe there are even more recondite ways one can search for things. What might these be? How to we display them? Again, where does it end? The more connections we provide the more we risk going insane in trying to keep them consistent. Maybe I’m over-thinking this but there seems no easy answer that satisfies all these conditions.

If you can give us any insights into either how you would like to find and cross-reference posts on this site or if you have a practical way of making the process better and more automated, please let us know. if we can crack this, I suspect we’ll all become billionaires.

The second thing on my mind is my contribution to an online discussion with about eight of my university friends about arranging a get-together in London later this month. The date has been fixed: the problem is the venue. There’s a huge email chain about this but so far no decision.

On closer examination, without someone stepping in and making a unilateral decree no decision will ever be reached for the simple reason that everyone’s suggestions are based on radically different criteria. Some are (or appear to be) negative, others are (or appear to be) the reverse. Some appear straightforward but may conceal a different agenda. Others have meaning only in terms of their relationship with suggestions made by others.

A couple of examples. One early suggestion was ‘somewhere near Kings Cross’. I think this could be translated as something like ‘I’m working in Camden that day (or have to get onto the Victoria Line afterwards) so Kings Cross would be convenient for me as well as probably practical for most other people.’ Self-interested altruism, perhaps. However, if these were his needs then other areas would do as well and perhaps better.

For my part, I have always had a deep loathing for trendy bars where you can’t hear yourself think, which serve fancy cocktails you can’t pronounce or warm bottles of lager with a chunk of lime shoved in the top, where the waiting staff are bored or incompetent, where the furniture is gleaming but uncomfortable and where everything comes in units of £10. To express my preference in this way would seem too negative so I contented myself with ‘how about an old-school pub, if any still exist?’ This doesn’t quite describe my wishes either: I’d be as happy with an Italian restaurant, for instance.

Someone else suggested a particular bar ‘or somewhere like it’ and gave a web link. All photos of pubs on websites make them look identical so it’s hard to tell much. In any case, ‘like it’ in what way? Postcode? Typeface of the logo? Pricing? We’re no further advanced.

Then someone else suggested another bar, quite different from the first,  which seemed to be a cool music venue. From the website it was impossible to tell quite what kind of music it specialised in. I was left with the faint impression that everyone sat round with headphones on. I suppose anything’s possible in Kings Cross these days.

Another idea was ‘somewhere where we can book a table’. One normally books a table to eat. Did this mean that she wanted a meal? If so, what kind of meal? Food hadn’t been suggested up til then and risked introducing a different dimension of choice that would make any decision completely out of the question. Or did she just want to be able to sit down? If so, a pub that wasn’t going to be packed might be what she meant. But did it need to serve food as well? I didn’t know.

Another place was recommended because ‘it wouldn’t induce hallucinations’. This seemed on the face of it an easy enough condition to meet – as easy ‘it doesn’t contain a large number of leopards’ – but could have been a criticism of one or other of the previous suggestions. If so it wasn’t clear which one. Another place was rejected on the grounds that the place looked ‘shiny’. Again, I sort of knew what was meant but the word might have been specific to some past drinking trauma and have nothing to do with the furnishings.

Then someone else popped up and said, sorry, she’d been away, but would abide by the majority decision. Jeez – we had to have a majority decision here? Just getting two people to agree on anything, even the basis on which we should be making our choice, was hard enough. A clear majority would be impossible.

After all this, one could draw up the following list. In order to be acceptable the venue had to satisfy an uncertain number of the following conditions:

  • It had to be in Kings Cross, or perhaps somewhere like, or near, Kings Cross, or perhaps on some of the same tube lines as Kings Cross
  • It had to be one of the places some people had suggested, or like these places in some unspecified way
  • It had to be a music pub, or a talking pub, or an eating pub but probably not all three
  • It had to be somewhere where we could book a table or, by implication, eat or perhaps just sit down
  • It had to be a place that would not induce hallucinations
  • It had to be a place that wasn’t shiny
  • It had to be a place where the majority of us wanted to go

One of us – a Professor of Computer Science at Cambridge and so well used to spotting odd patterns in the algorithms of life – suggested that our decision-making displayed the same contorted, overlapping and mutually incompatible logic as did The Celestial Emporium of Benevolent Knowledge, of which I’d never heard. This turned out to be a parody by Jorge Luis Borges of a philosophical idea he found absurd. It took the form of a list (written by him but purportedly a translation from an ancient Chinese encyclopedia) showing how all the animals of the world could be classified into 14 categories. It’s one of the best lists I’ve ever seen. The animals are divided as follows:

  • Animals that belong to the emperor
  • Embalmed animals
  • Animals that are trained
  • Suckling pigs
  • Mermaids (or Sirens)
  • Fabulous animals
  • Stray dogs
  • Animals that are included in this classification
  • Animals that tremble as if they were mad
  • Innumerable animals
  • Animals drawn with a very fine camel hair brush
  • Et cetera
  • Animals that have just broken a flower vase
  • Animals that, at a distance, resemble flies

What is so utterly wonderful about this is that each implies a clear if at times subjective true/false dichotomy but are followed by another which has no connection to it whatsoever. This makes it so much like our decision-making about the venue and, for that matter, about  the problem of what criteria to apply to searches or groupings on a website; or, by extension, about the way so many conversations and discussions are conducted, with everyone rattling along on their own train of logic or thought regardless of what if any relevance it bears to reality or what anyone else is talking about.

My grasp of logic and taxonomy now all but destroyed by these various problems and lists, I found myself thinking about other possible classifications. Three occurred to me:

Colours

  • The colour blue
  • Blue-ish colours
  • Colours which can be made using yellow paint
  • Colours that women can see but men can’t
  • Colours that change depending on the angle of view
  • Colours which are invisible or which don’t exist
  • The class of all colours (except those which are invisible or which don’t exist)
  • Colours not used on national flags
  • Unsympathetic colours
  • Colours which don’t look good next to other colours

Capital letters in English

  • Letters which are written using a non-prime number of strokes
  • Letters which never appear in words starting with a vowel and ending with a T
  • Letters which look like other letters
  • The letter S
  • Necessary letters
  • Letters which are also used for other purposes
  • Letters which, by any criteria, comprise less than a fifth of the class of all letters
  • Letters whose existence can be inferred from any of the letters so far specified
  • Letters which fail to satisfy three of the above conditions

Numbers

  • Numbers whose integers add up to 13 or a multiple of 13
  • Numbers which appear to be smaller than they actually are
  • Numbers which have never been written down
  • Very small numbers
  • Numbers which have been used in sonnets, villanelles or odes
  • Odd numbers which are almost even
  • Even numbers which contain no odd integers
  • Numbers which can reasonably be confused with other numbers
  • Numbers which serve no useful purpose
  • Numbers which are by any definition, rational or otherwise, interesting
  • Numbers which have mutated into other numbers
  • Numbers that look like prime numbers but which are not prime numbers
  • All other numbers except for numbers containing a 7

As you can see, this has driven me slightly out of my mind. You can help, though, in three ways.

  • You can tell me how we organise the tags and key words in this website
  • You can tell me where I am going to meet my friends next week
  • You can contribute your own lists of classifications

Choice? Classification? Who needs them? Sometimes, I just want to be told what’s what. I await your instructions.

Brian Quinn
• For further articles, please click here
• For rants and musings set to music, please click here

 

 

 

Share on facebook
Facebook
Share on twitter
Twitter
Share on pinterest
Pinterest
Share on email
Email
Share on print
Print

One Response

  1. OK, I’ll bite. Especially since my brain is rebelling at the task my employer has mandated I should perform. Just about anything is preferable. Soooo…

    2) You will be going to the Star of Kings (or is it the King of Stars?)

    1) Forget tagging and use rule-based searching instead. Each user has a set of search rules – some private, some public. New rules can be added using a simple interface (rather like setting up email rules for moving emails into different folders automatically). They can be simple (search for a keyword) or complex (cf your examples). A user can then browse the set of private rules, or the set of public rules, and apply whichever is most appropriate to the moment. The result might be a set of “hits”, which could be refined further by applying other rules to that set. You could perhaps set “meta-rules” such as “I’m hungry – run all the hungry-related rules”, and of course there would be a simple way to add new meta-rules. At the bottom of pages being presented on the screen there could be a selection of follow-on links in three sets – “related”, “anti-related” and “random”. The first would be the top two hits from running another the rule that is most closely related to the rule you just ran to find the page; the second would be the top two hits from running another rule that is the most distantly related to the rule you just ran; and the final would be the top two hits from a randomly-selected rule.

    3) Search rules taxonomy:

    1. rules that are intrinsically beautiful
    2. rules written after 11pm
    3. rules that would cause the maximum embarassment if performed on the entire web
    4. rules containing the letter z
    5. rules whose results contain only the letter z
    6. rules written in esperanto
    7. rules that would, but for 2 above, betray a sense of unwordliness excepting only the provision that such rule might, if such were possible, be the product of a “grey moment”
    8. rules of which Penny Post approves
    9. rules producing a countably infinite number of hits
    10. rules that a cat would produce if permitted to stray across the keyboard

    And now back to work (grumble) …

Leave a Reply

Your email address will not be published. Required fields are marked *

Sign up to the free weekly

Penny Post

e-newsletter 

for local, positive news, events, jobs, recipes, recommendations & more.

Covering: Hungerford, Marlborough, Wantage,   Lambourn, Newbury, Thatcham & Theale