Wednesday, 6 April 2011

Is google changing the way we think?

This post was triggered by two separate stimuli: first, seeing a demonstration of one of google's development projects on machine translation, and secondly, reading Is Google making us stupid?, an article by Nicholas Carr published in the Atlantic a couple of years ago.

The notion that google seems close to being able to provide simultaneous translation between dozens of different languages is impressive enough, but the thouht that this is achieved entirely through processing power and straightforward statistical analysis of the way words occur in different language texts available to google search engines, is really stunning, at least to me. It was demonstrated on Hans Rosling's programme for the BBC on the importance of statistical data in the modern world. He texted deliberately obscure messages in Swedish to the google researcher in California, who produced more or less simultaneous translations (as far as we could see), that were largely accurate. Rosling tried to confuse the programme by using phrases such as 'bisexual bishop', but the programme was not fooled.

Carr's article is one of dozens of publications over the past few years pointing out what seems to me obvious: we are being changed by digital technology. In fact this type of article is implying two main questions: is digital technology changing us in any way, and if it is, is this a good or a bad thing? These articles argue, with more or less evidence and analysis, that we are being changed, but the vast majority I have seen, while strongly implying that this is a bad thing, don't provide evidence for this assertion.

It seems to me that the use of any tool over a long period will have the effect of changing the user, both physically and mentally. When I played tennis from morning till night as a teenager, my right forearm became much larger than my left. When I more or less gave up playing, in my thirties, they gradually returned to being the same size. Furthermore I think it is clear from all sorts of scientific research, as well as from common sense, that our brains get 'patterned' to a greater or lesser extent, by our activities - craft workers develop tacit skills and knowledge through using a small selection of tools over and over again on the same materials. If large numbers of people use particular tools this will quite naturally also begin to have an effect on the culture and organisation of the society they inhabit. The introduction of the printing press had an incalculably complex effect on the social and political history of Europe in the C15th, starting with its effect on individuals and local groups. Similarly, though perhaps much more quickly, we must expect the same process to operate with digital technology.

We need to remind ourselves that these processes are not evolutionary ones, in the technical sense of Darwin's theory, as far as we are able to see; this is because the time scales involved are too small for genetic mutations which might confer advantage in a digital world, to be observed as having done so. That would require thousands of years for the human species as I understand the theory of evolution.

So, technological changes and the use of new tools do bring about physical and cognitive changes in individuals and groups. If widespread enough, these can bring about observable changes in society and culture. But change is not necessarily a bad thing: few would now suggest that the printing press and books had negative effects on people and societies (though some people certainly warned about them at the time - they were afraid that people's capacity to memorise would be lost if everything was written down - this sounds uncannily similar to the kind of worry Carr has about the digital world). The question of whether such changes are positive or not is not a scientific one for the most part, it is a moral and political one. The question Carr needs to ask explicitly, and the one such articles should really be aiming to answer, is whether a digital world is a better world, and why.

Instantaneous machine translation would be a morally useful tool in my view because it would allow for easy verbal communication between any people in the world without the elimination or reduction of the vast range and diversity of different languages, each of which, in my opinion, embodies a particular view of reality and of moral issues: this variety, just like biological diversity, is actually a survival mechanism - we need it to remain human.


  1. check out new scientist 02/04/11 'Evolution in the fast lane' article by Michael le Page, and also the emergent field of epigenetics. Adaptations are occurring much faster than previously thought. An example might be today's F1 driver, eg: Hamilton. Brought up from an early age as a pc gamer, the modern F1 steering wheel resembles a gaming console. The question is...has this caused observable effects on neural pathways (certainly) and will this cause effects in subsequent generations (very much a perhaps) percy aggett

  2. Thanks for this Perse. There are two main points here as I see it:

    1. Evolutionary change depends on genetic mutation, not on any other kind of physiological change. Genetic mutations occur randomly as a result of the inexact process of copying that takes place during reproduction. Some of these (probably very few) confer advantage to the offspring in terms of survival: as a result, over many generations, more of the offspring with the mutation will survive, and less of those without it. Eventually those with the mutation become 'the norm' for the species. For bacteria, which reproduce very fast, many generations might take place within a few hours, and humans might therefore be able to observe evolution in action: I believe this has happened with diseases. To observe the process in humans we would need to be able to 'see' a timescale of thousnds of years, if not more - and we can only 'see' the process with hindsight, because we can't predict when mutations ocur (they are random) or whether they will confer any evolutionary advantage.

    2. Lewis Hamilton may have developed his neural pathways, but how will he pass this on to his offspring? His genes are just the same as they were when he was born.

    There is also the question of whether his special neural pathways confer evolutionary advantage anyway. In some way we don't know about, they may actually make survival less likely.

    Anyway, as he persists in taking part in highly dangerous activities on a regular basis, common sense suggests that he is less likely to pass on his genes, neural pathways or not, to future descendants than many slow-moving, bus-travelling, TV-watching, neurally-challenged couch potatoes!

    You are a musician, and have been playing frequently for many years. You too will have highly developed neural pathways as the result of this repetitive activity, in the same way that if I play tennis every day for three years, my right arm will get noticeably bigger than my left (this happened when I was a teenager!). But here is no mechanism for passing this development on to my offspring through my genes. That's my main argument against your point. You are nothing but a 21st century Lamarckian! That sounds a bit like song - shall we write one?

  3. Google translator may not be as good as it promises after all. I've been listening to a group of Swedish academics on the radio talking about how in order to get published (and there as here, their jobs depend on getting papers published) they have to write in English. One of them said it takes him twice as long to write in English as in Swedish, whereas another said she quite liked it, and was proficient at it, but was well aware that she was not getting the nuances of meaning she would be able to in Swedish - that sometimes, as she said, she sounds like a child - but this obviously isn't a major problem for her or her employer: it gets the job done.

    So why don't they use google translation software? I assume because it doesn't work well enough. Actually I wrote this in Swahili, can you tell?

    Seriously, I don't know any Swedish, but here is what I have just written translated by google into Spanish:

    Traductor de Google pueden no ser tan bueno como se promete, después de todo. He estado escuchando a un grupo de académicos suecos en la radio hablando de cómo el fin de ser publicado (y no como aquí, sus puestos de trabajo dependen de conseguir los documentos publicados) que tienen que escribir en Inglés. Uno de ellos dijo que le toma el doble de tiempo para escribir en Inglés como en Suecia, mientras que otro dijo que le gustó bastante, y era experto en eso, pero era consciente de que ella no estaba recibiendo los matices de significado que ella sería capaz de en sueco - que a veces, como ella dijo, ella suena como un niño - pero esto obviamente no es un problema importante para su empleador o ella: se hace el trabajo.

    Entonces, ¿por qué no utilizar el software de traducción de Google? Supongo que porque no trabaja lo suficientemente bien. En realidad, yo escribí esto en swahili, puede usted decir?

    En serio, yo no conozco a ningún sueco, pero esto es lo que acabo de escribir traducido por google al español:

    Looks pretty good to me!