Tag Archives: facebook

January 13, 2015 · 12:54 pm

There’s an election in four months

cameron-lol

There’s an election in four months.

Those are the only words you need to remember. Whenever a politician drivels before an invited audience of heart-eyed acolytes or assembly line workers glazing over on company time for half an hour, just remember: there’s an election in four months (or three, or two, or one…).

Yesterday renowned cryptographer David Cameron said there should be no “means of communication” which “we cannot read”. This has been interpreted by technically literate commentators, mostly through the medium of boggle-eyed laughter, as expressing a desire to ban encryption or enforce the addition of backdoors. I’ve seen many, many tweets setting out the stupidity of such a move, and I have no desire or need to rehash them here.

Because the only thing you need to know is: there’s an election in four months.

Cameron is talking about monitoring the internet because there is an election in four months. He wants people to vote for him. He understands — better, sadly, than those tweeting about protocols and key escrow and men-in-the-middle and laptops left in taxis — that none of all that matters. He’s not talking to that audience, the tiny audience that groks the detail and the implications. He’s talking to the other 99%, who saw the attacks in Paris last week and think (thanks to Be Vigilant And Report Darkies posters) that we’re next.

Let’s imagine Cameron is elected in May with a majority. What would he actually do? We have no idea. There’s no manifesto yet, and manifestos can’t be trusted anyway. On past experience — we have almost five years of it now — his words don’t much match his deeds. I expect there’d be a series of meetings, possibly involving token techies invited as a sop to industry, and the End Terrorism Forever Bill 2017 (probably) that would emerge would contain no clauses capable of achieving any such thing.

There’s an election in four months. That’s all Cameron is worried about.

The opposition parties (I include the Lib Dems in that category for election purposes) have the same phrase in their heads. If they want to oppose Cameron on this issue — and I’m not entirely sure the Labour party does, for fear of being labelled soft on terrorism — then there is absolutely no point in talking technology. That’s preaching to the choir.

To oppose this policy they need to do two things: pursue, with great vigour and purpose, the support of younger people (beneficial side-effect: these are least likely to be slack-jawed kippers); and tell them in specific terms which apps and services Cameron thinks they shouldn’t be permitted to use without being snooped on.

Snapchat, WhatsApp, iMessage, FaceTime, Yik Yak, Rooms, Skype, etc, etc — and also Facebook and Twitter and plain old email, of course, but with less emphasis since younger people don’t use those so much. Ignore the likes of HTTPS, Tor, and all that: too confusing for the audience you’re trying to reach.

Keep it simple. Non-technical. Personal.

Avoid greyfaces and clumping hooves of rhetoric: all an utter turn-off for the audience. Don’t make it an official party video at all. You want Cassetteboy, not Saatchi, and if you don’t know who Cassetteboy is, fire yourself.

Here’s an idea off the top of my head: take one (or more) of those ubiquitous thirty-second promo videos from an app vendor’s website — you know the ones, with the indie guitar solos and the Californian hipster voiceovers — and every time a toothy blond communicates with another toothy blond, intercut video of Cameron sitting at a computer screen.

It doesn’t matter that it’s inaccurate or simplistic: so is what he’s claiming to propose.

There’s an election in four months.

3 Comments

Filed under Random

Tagged as , , , , , , , , , , , , , , , , , , , , , , , ,

April 29, 2013 · 10:58 pm

The fuzziness business

I have a mild obsession with metadata. I imagine there’s a twelve-step programme to sort me out, hopefully tagged metadata, obsession and self-referential gag so I can locate it.

My current instance of the obsession concerns dates on photographs. Not the most radical of concepts. A problem, you might think, solved long ago when the elders plonked multiple competing metadata chunks inside an unsuspecting jpeg. One minute it’s slowly interlacing its way through a lady’s bosoms, the next it’s been given a sick bag of acronyms like EXIF, IPTC and XMP. It’s drowning in dates, unlike me.

These acronyms mean I can sprinkle an image file with hundreds and thousands of items of metadata —  dates, and keywords, and copyright statements, and model release IDs, and Photoshop thumbnails, and GPS coordinates, and camera data, and Wake up! Wake up!

Unfortunately for me, these nuggets of metadata all have one thing in common: certainty. Attempt to poison the well with a droplet of uncertainty, and you’re out of luck.

Consider this:

Billy, Art and Leslie

This photo from my family’s archive was taken during World War Two, not far from the Suez Canal. It shows three friends — my great-uncle, his brother-in-law’s brother-in-law, and his half-brother — who pretty much bumped into each other while Montgomery was shaking his fist at Rommel in the desert.

Here’s the problem: I don’t have a precise date for this photo. Neither do I have a precise location. But I do have some very important and useful imprecise information: 1942 and the Suez Canal.

How do I encode this within the photo’s metadata?

Keywords, you say? Well, yes, but that’s the everything is a nail solution. I’m a fan of keywords, some of my best friends are keywords, but I don’t think they’re the solution here, at least not shovelled in as plain 1942 and Suez Canal, with nothing to identify them as approximate dates and locations.

Are there any other possibilities? On closer inspection it turns out that, at least for some date properties in the metadata ocean, you can reduce precision. Instead of the full YYYY-MM-DD HH:MM:SS.SSS+ZZ:ZZ monty you can say just YYYY-MM-DD, or even just YYYY. Whether or not any given vendor supports this is a different matter entirely – Flickr has a “chop the tail off” feature for dates, but I don’t know whether it automatically spots chopped-off dates embedded in files you upload.

But anyway: problem solved! I’ll put the date as 1942 and move on.

Except that’s not a very good general solution to the problem. And in any case, I lied earlier: I don’t know that the photo was taken in 1942. It might be 1943 — and I can’t encode “either 1942 or 1943” in one date field. Some beard-oriented committee’s idea of imprecision simply doesn’t match the real world.

And I can think of other, similar imprecisions, perhaps most importantly entities portrayed (“this part of this image shows either person/thing A or person/thing B”) and location (“this image shows either fuzzy place A or fuzzy place B”).

Is it too much to want to preserve this type of metadata in the file? It’s certainly valid metadata, and more useful in the real world to real humans than Did the flash gun fire? But it seems I can’t do this. Despite having apparently defined “date of creation” several times over in some perverse kind of Groundhog Metaday, nobody appears to have found the time to specify how to encode fuzziness.

This matters because more and more images that weren’t created by modern digital cameras pointing at actual, current events are making their way onto Flickr and Facebook etc. Let me just dangle the word genealogy here to emphasise the point. (And please note, vendors, there is a very significant difference between “date of creation of file” and “date of original creation of image”. A flatbed scanner is not a time traveller.)

Lack of imprecision is a genuine problem for tagging nerds like me, and it’d be nice if someone could fix it. Handily, the XMP standard — originally defined by Adobe, now an ISO standard, and supported by common image formats like jpeg and dng — allows for extensions: in other words, it lets people define their own metadata. You just need to convince every important vendor to support the madcap scheme you come up with.

With that rather glaring caveat, here’s an idea for dates.

First, a bunch of examples of fuzzy dates people might genuinely want to encode: “sometime in 1942”, “before 1900”, “between June and July 1972”, “between noon and 3pm EST on October 30, 1980,” “either 1945 or 1947, but not 1946,” “definitely not 1903, but apart from that I don’t have a clue”, “744 BC according to the Roman Calendar of Romulus”.

I think these examples generalise to: I want to embed a file with one or more preferably non-overlapping date ranges, possibly open-ended, with each date range specifying a calendar scale (defaulting to the proleptic Gregorian).

Excitingly, the ISO 8601 standard commonly used for representing dates and times specifies how you can encode time intervals in the proleptic Gregorian, and it’s almost exactly what I’m looking for. You can say things like “2007-03-01T13:00:00Z/2008-05-11T15:30:00+0100” — meaning “between March 1st 2007 at 1pm GMT and May 11th 2008 at 3:30pm BST” — if both ends are known with precision. If not, you can leave elements out: “2007-03/04” means “March-April 2007” (elements omitted on the right of the slash, like the year in this example, are considered the same as the value on the left).

I don’t think there’s a standardised way to denote an open-ended range, but I think you could special-case omitted dates either side of the slash: so “/1899” would mean “before 1900”, and “2063-04-05/” would mean “after first contact between Vulcan and Earth”. A single slash “/” could be used to mean “I have no idea” (which is better than saying nothing, as it lets you record a decision — to distinguish “I have examined this image and I have no idea” from “I haven’t got round to tagging this image yet”).

A list of these date ranges then defines a bunch of non-overlapping alternatives: “either this range or this range or…”, thus allowing you to easily state “1945 or 1947”, “either January 17th 1901 or sometime before June 1896”, or whenever.

Calendar scales are probably overkill, but it doesn’t hurt to think ahead. Other groups are: for example, the VCard spec suggests IANA should maintain a canonical list of official calendar scale identifiers, and says it should allow x-name experimental/unofficial values. That seems a good idea to me. In reality, since alternate calendar scales will be rare, we should treat them as special cases and not hobble the entire idea to support them.

Now to implementation…

A quick skim through the XMP specification reveals it lets you store structured data, including lists of such structures. This means we could decompose the calendar scale and the two dates into three separate fields, but let’s keep it simple: since we have the ISO 8601 time interval, and we’ve identified some simple and very useful augmentation to it, let’s just bolt on some optional calendar stuff as a prefix. Then we’re left with a plain old list of strings. Easy.

All we need now is a little naming whitewashed over the top, some well-defined semantics, and Bob’s your imprecise uncle. Still time for Adobe to implement in Lightroom 5!

XMP extension specification: Fuzzy namespace

This namespace contains properties that provide “fuzzy” or imprecise metadata.

  • The namespace URI and field namespace URI shall be “http://ns.avaragado.org/fuzzy/1.0”
  • The preferred namespace prefix is fuzzy.

These are the properties so far defined in the fuzzy namespace:

fuzzy:depictedDateRanges

Type: unordered array of fuzzy:dateRange

Description: A set of possible date ranges that apply to the document. The actual date is unknown but believed to be within exactly one of the date ranges.

fuzzy:dateRange

Type: Text

Description: A single date range. A Unicode string containing up to five parts:

  • Optional. A prefix defining the calendar scale that applies to the date range. Allowed values are as defined by IANA. If omitted, defaults to the IANA designation for the proleptic Gregorian calendar (which I assume for example purposes would be “gregorian”).
  • Optional. A single pipe character “|” (U+007C VERTICAL LINE). Must be present if and only if the calendar scale is present.
  • Optional. A “not before” date, written in the syntax applicable to the calendar scale (for proleptic Gregorian, this is the Date type as defined in the XMP specification; the syntax for dates in other calendar scales is not defined here, but it is assumed it does not contain a pipe or a slash character). If omitted, means “the beginning of time”.
  • Optional. A single slash “/” (U+002F SOLIDUS). Must be present if both “not before” and “not after” dates are specified, or if neither of them is.
  • Optional. A “not after” date, following the same date syntax rules as the “not before” date. If omitted, means “the end of time”. Must correspond to a date no earlier than the “not before” date. Must have the same or fewer date components (year, month, etc) than the “not before” date; must not have components omitted from the “not before” date.

If only one date is supplied and there is no slash, the date is used as both the “not before” and “not after” date. This allows for concise imprecision using the XMP Date value’s ability to omit date components.

Here are some valid example values, with semantics in parentheses:

  • / (“I don’t know”)
  • /1899 (“Before 1900”)
  • 2063-04-05/ (“On or after April 5th 2063”)
  • 2003 (“Some time in 2003”, same as 2003/2003 or 2003-01-01/2003-12-31)
  • 2003-01 (“Some time in January 2003”, same as 2003-01/2003-01 or 2003-01-01/2003-01-31)
  • 2003/2005 (“Between the start of 2003 and the end of 2005”)
  • 2013-04-27/28 (“On April 27th or 28th 2013”)
  • 2013-04-27T14:01+01:00/ (“At or after 2.01pm BST on April 27th 2013”)
  • 2007-11-13T00:00/15T24:00 (“Between November 13th and 15th, 2007” — equivalent to “2007-11-13/15”)
  • gregorian|/ (“I don’t know, explicitly in the proleptic Gregorian calendar”)
  • gregorian|1800-01-01/1899-11-30 (“Any time in the 1800s except December 1899”)
  • x-wombat|hatstand/banister (“Between two values meaningful in the x-wombat calendar scale”)

And here are some invalid example values:

  • (the empty string) (invalid because the slash is mandatory if no dates are given)
  • gregorian| (invalid because the slash is mandatory if no dates are given)
  • 2001/2000 (invalid because the “not after” date is before the “not before” date)
  • 2000/2001/2002 (invalid because no more than two dates are permitted)
  • 2001/2001-04-05 (invalid because the “not after” date contains date components omitted from the “not before” date)
  • 2013-04-27/15:00 (invalid because the “not after” date contains date components omitted from the “not before” date)
  • hatstand/banister (invalid because no calendar scale prefix means proleptic Gregorian is implied, and neither date component is a valid Date as defined in the XMP specification)

Addendum

You know what? You can add individual fuzzy date ranges as keywords to your images right now. You don’t have to wait for anyone. Prefix each fuzzy date range with fuzzydate: and maybe, at some point, a vendor like Adobe or Flickr might notice. Flickr already spots some specially formatted “machine tags” and treats them differently than normal tags.

Examples:

  • “Sometime during the American Civil War” is one keyword: fuzzydate:1861-04-12/1865-05-09.
  • “Definitely not 1901” is two keywords: fuzzydate:/1900 and fuzzydate:1902/.

Away you go.

1 Comment

Filed under Random

Tagged as , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,

September 30, 2011 · 2:35 pm

Facebook timelines: mind the gap

Facebook’s new Timeline feature is seemingly intended as an autobiography in the automatic sense: your life distilled to a series of status updates interspersed with red-eye, all filtered and summarised by the invisible hand of logic, the unblinking, unknowing eye of the Zuck-3000.

I’m intrigued by what my own timeline will show. I don’t write the updates Facebook expects. There’s rarely a sense of where I am or what I’m doing. I never say “Sitting in Bill’s, writing a blog post about Facebook timelines.” They’re my tweets, echoed; my rants, hashtag games, rubbish jokes, and other nonsense. My own timeline is likely to consist of other people’s photos of me, some events, and a bunch of non sequiturs. Here’s me at Chris’s 40th; here’s a link to a blog about Facebook; here’s a photo of a typo. Nothing of deep historical or biographical interest, I imagine.

For eager, unjaded pubescents pouring every numbing detail of their lives into Zuckerberg’s database, the timeline will be great – until it isn’t. Until a friendship disintegrates – as it will. Until a relationship ends acrimoniously – as it will. Then the timeline becomes a sniggering, taunting reminder, Gripper Stebson poking Ro-land in his chubby face, forever.

Thus, inevitably, it will become another part of our life online that needs pruning, tending, curating, culling. It will beg for attention, Tamagotchi: The Next Generation. Users will constantly edit their lives. I am going out with Terry from form 3C, I have always been going out with Terry from form 3C.

Facebook wants your timeline to be your autobiography, but it won’t be. For most people it’ll be like a Hello! magazine puff-piece: all of the glamour and the shiny taps, and none of the hoovering.

You won’t see, perhaps, underage drinking at a dodgy party featuring a jazz cigarette, or cruelly excised former friends or partners. I think there’ll be more of the latter omissions than the former: people’s youthful indiscretions appear to be becoming less important. Clinton had to claim he “did not inhale” but Obama didn’t; Cameron successfully sidestepped questions about his own drug use and not even the Daily Mail proclaimed the End Times. I think this is a natural societal evolution, not caused by the Internet but certainly made more visible and – crucially – searchable by it.

However society adapts to decreasing privacy, it’s the gaps that are most interesting. Facebook doesn’t know what happens in the gaps, in the mini dark ages that pepper my history and everyone’s history. It aims to know all but does not, will not, cannot, even with the vast data-buckets that it and the internet in general can supply on-demand for each of us. Like the missing years in a CV or the crackle as an old film skips a few frames, what is absent is often far more interesting and revealing than what is present.

I therefore submit this humble prediction. Facebook’s timeline will ultimately be no tell-all semi-autobiography. It’ll be a sanitised, part-fictionalised history. Stalin’s airbrushing writ large; Big Brother’s ultimate rewrite.

And employers won’t look at your timeline to decide whether to interview you or hire you: they’ll look at your friends’ timelines.

Leave a comment

Filed under Random

Tagged as , , , , , ,

September 22, 2011 · 5:29 pm

Facebook and the two minutes’ hate

In George Orwell’s 1984, Big Brother enforces a regular two minutes’ hate to ensure the citizen-slaves are fully au fait with the enemy of the day. It’s a term that originated in WWI’s artillery bombardments and is now mostly evident fully fourteen letters later, on that other battleground, the WWW.

Today’s enemy – as on so many other days – is Facebook. If it’s Wednesday, it must be another home page revamp. This latest change does away with those pesky ‘Top Stories’ and ‘Recent News’ options, which always used to default to ‘Top Stories’ even though I never, ever wanted anything other than Recent News. Now they’ve inflicted some cockamamy algorithmic hodgepodge that combines those two options, and the entire internet has exploded in a fireball of wtf and zomg.

This is not Big Brother’s doing: nobody at Microsoft or Google or Apple pressed a blue button marked with a cuddly sans serif f and sat back cackling with a smartphone full of white, fluffy lolcats. This is entirely a bottom-up reaction. Were I a lazy reporter desperately scrambling for a headline, I might even make some cackhanded pun on Arab Spring. But I can’t think of one right now.

People despise change because we are creatures almost entirely of habit. We might like to think we have will and self-determination and can stay up all night if we want to, you can’t stop me, mum. But we don’t. We laugh at dogs and their pavlovian reactions, and then it’s 4 o’clock and time for tea, oh and I mustn’t miss today’s Pointless I do like that nice Alexander Armstrong, don’t you? and the punched paper tape loops through our brains one more time.

Slow, gradual changes are easily accepted, embraced, like a frog being slow-boiled. I bought a cheapo wifi-capable printer last week, now perched on an ex-server in my spare room together with the military-grade safe in which I keep its priceless ink. The first few times I walked past the room my inner lizard shouted There’s something in there – it was a new, unexpected pattern on the retina and I turned my head involuntarily. Now, meh, I’ve been retrained. The new pattern is absorbed, the newness has gone, the routine is back.

More radical change takes longer to process. In a new house you’re constantly jumping and starting at its various ticks, cracks and wheezes, getting lost in cupboards and locking yourself in fridges. And all the time you’re swearing like a navvy on jankers. Your brain is clunking and clanking away rewriting the paper tape, and your eyes spin like the MacOS X hypnowheel (other operating systems are available) until the updated universe can be paged back in.

Thus it is with Facebook’s latest update. Millions of people are using the site’s various existing features, which they railed against last time they were changed, to protest these latest modifications. In a couple of days the chances are the Facebook juggernaut will thunder on to its next redesign unperturbed by self-immolating gifs and its users will wonder why they were all worked up about it in the first place.

That’s not to say I like the changes. I think they’re daft. And here’s why.

Let’s look at the UI. There are arguably four focal points: the icons at top-left next to the word ‘facebook’, which gain red numbers when something interesting happens; the list of stuff that seems to change in a way I don’t quite understand down the left (favourites, lists, apps) which gain blue-grey numbers when something interesting happens; the main body, which updates in a way utterly unfathomable to mortals, gaining and losing ‘recent stories’, ‘top news’ and other sections when something interesting happens; and the ticker on the right, which updates in a hazily understood way when something interesting happens.

Now, I don’t know about you, but I think that’s a lot of something interesting and a lot of I don’t know what makes something appear here. Oh, and some of it may appear in multiple places.

I’m not interested if you have figured out how each of these sections works and when it updates. Well done, have a banana. You are special. The point is, most people don’t. To most people, the way all this works is mysterious and magical and they just hope to jebus that they can find the thing they want to when they need to. This is not because they are dumb, it is not because they are lazy. It is because they just don’t care enough to work it all out: their goal is not to understand all this. To a first approximation their goal, to use the sainted jwz’s expression, is to get laid (link is geeky, SFW!).

Facebook’s current incarnation does not make this any easier than its last. It makes it harder, because the vital goal-fulfilling information is scattered amongst the something interesting and I don’t know what makes something appear here sections.

In Joel Spolsky’s words, the fundamental rule of user interfaces is: “A user interface is well-designed when the program behaves exactly how the user thought it would”.

Facebook does not behave how users think it does, because most people have no idea how it behaves. There are too many places where slightly different, possibly overlapping pieces of information are presented in slightly different ways, and those pieces of information are chosen using slightly different, possibly overlapping, closely guarded, unfathomable algorithms.

Which brings us to the end of today’s two minutes’ hate. I think I overran. Ah well, time for a cup of tea and Pointless.

Leave a comment

Filed under Random

Tagged as , , , , , , ,