Can computers write poetry?

March 27, 2013 at 9:43 pm (art, computers, interesting, internet) (, , , )

Working and being fascinated by computers and the Internet, I regularly come across interesting applications where someone has taken technology and done something quite unexpected with it.

But every now and again I happen upon something even more unexpected – a side effect of technology in action that is both quite unanticipated and extraordinary.  The website Google Poetics is one such site.

If you use Google, I’m sure you’ve seen how it anticipates what you wish to search for.  It is at times an interesting barometer for what the world is thinking about a topic.  I remember a very topical time when typing ‘how to quit’ had ‘how to quit facebook’ right at the top.  Probably after some Facebook update that was annoying people at the time.  Today it has how to quit … smoking; smoking weed; your job; drinking.  I don’t think this is personalised to me!

Now Google works very hard to anticipate everyone’s needs all as part of their mission to ‘return one search result – the one you want’.

But I doubt that even the engineers at Google anticipated that sometimes it will return a set of results that strikes a chord with people and actually means something quite profound.  That is what Google Poetics is collecting – examples of poetry made from the accidental (or at least, algorithmic) collisions from millions of people searching using Google.

Here are some of my favourites so far.

Would you like me to

  • Would you like me to be the cat
  • Would you like me to seduce you
  • Would you like me to
  • Would you like me to rephrase the question

We are not p

  • We are not permanent we’re temporary
  • We are not pilgrims
  • We are not pirates we are fishermen
  • We are not promised tomorrow

As I turn

  • As I turn the pages
  • As I turn away
  • As I turn up the collar on my favourite winter coat
  • As I turn my back on you

Sometimes I p

  • Sometimes I pretend to be normal
  • Sometimes I pretend
  • Sometimes I pretend I’m a carrot
  • Sometimes I put my hands in the air

The examples above bought to you courtesy of @GooglePoetics.  I could lose quite some time reading some of these.  And all as a consequence of the Google algorithms (far to) honestly regurgitating the behaviours of millions with some quite profound results.

And another interesting property of these poems, is that they are changing and not always the same for all readers. Try it yourself – type the titles into Google and see what you get back for you.  It will depend on your location, your search history, what everyone else has searched for recently and hundreds of other ‘small signals’ that combine within the walls of Google HQ to give you what it thinks you want.  Imagine attempting to design a system from scratch that could do this.  Talk about an emergent property!

Try it – its not quite as easy as I looks, and it can sometimes be dominated by song lyrics.  But every now and again you might find a gem.  If you do, make sure Google Poetics get to know about it.

Here is one of mine.

She is missing

  • She is missing
  • She is missing you
  • She’s a freak never missin a beat
  • She is missing me

Can computers write poetry?  I think this is proof that they can, albeit as an unintended consequence of something quite different.



Permalink 1 Comment

Faces in the Sand

August 5, 2012 at 10:25 am (art, computers, interesting, internet, science) (, , , , , , , , , , , )

I had an hour spare in London yesterday so took some time to visit the Science Museum.  Now this is something I like to do from time to time, if nothing else just to pop in if I’m passing that way to see if the Listening Post is still there.

This time I had two aims in mind.  First to catch a little of the Alan Turing exhibition, although most of it I already knew, from having visited Bletchley Park in the past.  Then, from the birth of modern computing, I wanted to visit the Google Chrome Web Lab.  This is a recently opened exhibition, running until June next year.

So, I find the museum, wander in and immediately see the Turing exhibition, so walk through that.  Mission accomplished, I walk on through the main gallery, past the steam engines, through the space section, past the ‘building the modern world’ gallery, home of the cut-in-half mini, the cray supercomputer and the ‘decades’ displays.

I had a brief pause when I noticed the prototype Clock of the Long Now.  This is something I have literally just read about in Richard Watson’s book ‘Future Minds’ (which I totally recommend).  In the book he presents a search for slowing down, worried about how the ‘always on, always interrupting’ nature of the modern, connected world leaves no time for quiet thinking, reflection and the kind of deep thought that really leads to new ideas.  He mentions the clock as a symbol of “thinking slowly”.

The prototype of the 10000 year clock ticks once every 30 seconds and has been installed in the science museum.  The final clock is being built inside a mountain in the US (like the fictional Colossus)! The web page lists the prototype as being created December 1st 01999 – not many projects will list a 5 digit year! (reminds me of As Slow as Possible, but I could digress about such topics for ages, so I won’t!).

Once past the main galleries and into the newer Wellcome Wing, an immediate right turn will take you down some steps to the Chrome Web Lab.  And after all this old, long term thinking you are brought right into the current, connected present.

The idea behind the lab is to make the Internet seem real.  A very topical subject for me right now as I’m currently reading Andrew Blum’s ‘Tubes’ about his discover to find the real, physical Internet (but more on that another time).  Of course whilst educating the populace, it has the side effect of raising awareness of Google’s own web browser technology, Chrome, showcasing its future looking HTML 5 application development ability.

Google have created 6 experiments that can be run locally in the museum and online at  Visitors to the museum can see the online participants and those online can see elements on the physical displays too.  Based on the kinds of Internet projects I’ve seen Google push in the past, I was very interested in seeing what they will do themselves.

When you first walk in (or register online) you get a “lab tag” which you put in the experiments to “log in”.  You can also hold it up to your webcam when you get home and link back to your time in the lab.

There are five experiments in total, but the ones that really appealed to me were the Universal Orchestra and the Sketchbots.

The Universal Orchestra is a set of six digitally controlled musical instruments.  A marimba, xylophone, drums, tuned drums, shakers, wood blocks, etc.  It is controlled by programming using a system of dots that appears to owe quite a lot to a Tenori-On.

Three instruments can programmed from within the physical lab and three are programmed only online.  A computer provides an ethereal accompaniment track based on the notes chosen at any one time.  The continuously changing nature of the music provides a fascinating aural background to the rest of the lab.

But the experiment that prompted the title of this post is the collection of sketchbots.  Again, six in total, three controlled only online and three from the lab.  You insert your tag and stand in front of a webcam.  It takes your picture and then goes through a series of image processing steps to isolate the key lines of your face.  These lines are then drawn in some smooth sand by a robot arm.

There is a wonderful irony at play here.  On the one hand, the robot draws your picture, the table rotates a quarter turn and in the space of three rotations your picture is no more.  Washed away just like a picture in the sand at the beach.  Forgotten to all.

But in parallel, the digital version that allowed the robot to draw the picture in the sand has been remembered.  It sits on Google’s servers, linked to your own tag to be recalled at a moments notice.  How long will it remain?  Who knows.  “Storage is cheap”.  The cloud is forever.  (Note that the T&C say data will be deleted when the exhibition closes).

We are in an age of digitally never forgetting.  Whilst once it took lots of effort to remember – initially sharing a song around the fire, passing on tales and stories, then writing and language allowing written records.  Finally the printing press allow mass distribution.  Still, recording was an effort.

Today it is almost totally the reverse.  If it’s digital, it is remembered by default by something, somewhere.  We have the technology to record every moment of our lives, but when would we find time to watch it?  Many digital photos are “write only” taken and automatically preserved, but never looked at again.

It is now more effort to forget digital information.  Google knows what you’ve done – what you’ve searched for, what emails you’ve sent – in some cases where you’ve been.  To ask it to forget is next to impossible.  It will maintain your digital footprints in their digital sand for as long as its useful to them.

Hence the wonderful irony enshrined in the Google sketchbots – my picture is long gone in the sand, but lives on (as does the record of my visit) online.  And to me it points to a future where what was once transient is becoming permanent.  Interesting Times indeed.

It is very well worth a visit to the web lab – in the physical space if you happen to be passing, or online.  A very nice way to spend an hour.


Permalink Leave a Comment

NGrams are beautiful?

December 14, 2011 at 7:07 pm (computers, internet, odds) (, , , , )

Was looking at the website “Information is Beautiful” ( and found an entry about the Google labs NGrams product that shows occurances of pairs of words in the digitised libaries of Google books, some of which go back hundreds of years.  There are some interesting ones – I liked the ‘age old debates’ such as chicken vs egg, region vs science, etc.

So I had a bit of a play myself.  War and peace was interesting – some very clear peaks in the early 20th Century as you’d expect.  Eventually I started on some more odd ball ones, including Google vs Microsoft.  Well that yielded a surprise …

Don’t know if you can see that … but just what is that peak around 1900 for Microsoft?

Well, looks like it must just be a mishap in the Google scanning of books … top of the list of books returned from searching for Microsoft from 1880 to 1950 is the following:

The corporate software guide 1938 – Snippet view

Microsoft FORTRAN also has a high degree of Xenix source-level compatibility, a comprehensive set of utilities, and direct interlanguage calling with Microsoft C, Pascal, and Macro Assembler routines.

So there.  Writing about Xenix, Microsoft C and Pascal in 1938 … so that’s where Alan Turin got all those ideas from!

But on further investigation, it appears there are even earlier examples (obviously very hushed up, but presumably can be released now)

Principles of Food, Beverage, and Labor Cost Controls Using Microsoft Excel for Windows

No cover image Paul R. Dittmer – 1920 – 256 pages – No preview

So now we all know.



Permalink Leave a Comment

Subtle Marketing …

September 22, 2011 at 6:00 pm (internet) (, )

Just in case it slipped anyones attention,  GooglePlus is now open to all.

I do like the way Google are using very subtle marketing techniques just in case anyone hadn’t noticed. In fact it’s so subtle, you might not have noticed yourself, so I thought I should point it out to you …


Permalink 2 Comments

The Creative Internet

February 9, 2011 at 10:50 pm (art, computers, interesting, internet) (, , , , , , , )

Been meaning to post about this for a while now.  Google put together a presentation of loads of different Internet related projects that people have done.  Next time you see some piece of media highlighting how bad the Internet is, pick one of these at random and redress the balance a little.


Some personal favourites of mine include:

Fantastic stuff.  Tech at its best.


Permalink 1 Comment

What HAVE Google done to their image search?

August 17, 2010 at 5:55 pm (internet) (, )

I’ve regularly used the Google image search – simple, quick, find the pictures you want.  But recently, they changed the interface, and boy is it annoying.

Now, you get expanding images when you move the mouse around the screen, constantly distracting your eye as you scan over the pictures for what you want.  When you click on ‘more results’ it seems to jump to random page numbers.  Once you reach the bottom of the page there isn’t an obvious ‘next page’ link.  And it is so slow to load!

Thankfully there is a release – at the bottom there is a ‘switch to basic version’ link (read, switch to the working version).   However, there doesn’t seem to be a way to make this the default choice.

The cynic in my wonders if this might be a new policy of Google: start releasing really irritating user interfaces that people will have just to login to iGoogle or their Google account to change to a usable version … I can’t think of any other reason why anyone would consider this an improvement.  It is so anti-Google – their user interface is usually so slick and clean.  This is just so clunky.  An odd choice.


Permalink Leave a Comment

The best Google Logo yet?

May 21, 2010 at 8:24 pm (internet) (, )

The Google Doodles have been getting more and more elaborate recently, but I really think today’s logo is going to take some beating …

And whats more, if you click on ‘insert coin’ (where the ‘I feel lucky’ button is usually) you get to play it.  A fully playable pac-man logo.  Very neat indeed.


Permalink 2 Comments

Virtual Revolution

March 1, 2010 at 10:16 pm (internet) (, , , , , , )

I’ve just finished watching the BBC’s Virtual Revolution programmes.  These were presented by Aleks Krotoski and shown on 4 consecutive Saturday evenings a month or so ago.

It follows the evolution of the Internet from its humble beginnings as the ARPANET and a collection of academic networks, through the non-commercial era, to the dot-com boom and bust.  Finally, it talks about how people like Amazon and Google worked out how to make money from the Internet by exploiting out information in exchange for us using their services for free.  We get the benefits of the services and haven’t really had any negative impact, yet, from giving up our personal information to them.

The final programme looks at the whole issue of online social networks such as Facebook and looks at some of the, as yet unknown, future effects on society.

One interesting conclusion, well discussion but looking like a conclusion, is that the web seems to be encouraging more associative brain functions than linear … people prefering short, associated chunks of information rather than large, linear books.  This is one reason that many of ‘generation web’ don’t read books!

A very interesting series.


Permalink Leave a Comment

Just Google!

January 28, 2010 at 9:54 pm (internet, music) (, , , , , , , , )

Can’t remember quite how this started, it was probably a bit like this xkcd, but after an odd series of random links and searches, I ended up googling for ‘google song’ and turned up the song ‘Just Google‘, with words about searching using google, sung to the tune of ‘Beat It, by Micheal Jackson.  Its a bit daft, but some of the lyrics are actually quite clever … “how to milk a camel, how to cut your own hair.  37 years with an electric chair and pictures of a man being killed by a bear – Just Google” …

Its certainly a lot more inspired than this other effort.  Still made me laugh but for all the opposite reasons – this one is just odd.

Of course, songs about Google are not new – take this Cookie Monster song from Seasame Street.  It was recorded some 16 years before Google as we know it came about at all!  Just goes to show, theres nothing really new at all …

Oh, and in case you are wondering, there appear to be quite a number of Facebook songs too.


Permalink Leave a Comment

Painting via Google Street View

September 11, 2009 at 7:58 pm (art, interesting, internet) (, , , , , , , )

There have been several articles in the mainstream media (Telegraph and Daily Mail to mention a couple) just recently, about American artist Bill Guffey, who paints traditional scenes, but instead of painting from a visit to a landscape, or a photo, he uses the pictures provided by Google Street ViewSee his gallery hereIn another gallery, he’s painted a picture from every state in the US (apart from Hawaii, that apparently doens’t yet have Street View).

Its a great application of modern technology to a more traditional pastime.


Permalink 4 Comments

Next page »