Google CAPTCHA codes (Completely Automated Public Turing Test to tell Computers and Humans Apart) were invented to reduce spamming and hacking on the internet. They are usually made up of a series of random, distorted letters and numbers that must be deciphered, and more recently, in the form of images that must be selected according to key words. However, Google has now built a new system that allows the user to simply click a check box, letting the algorithms behind the scenes determine whether or not you are a ‘bot’. This is done through background monitoring things such as mouse movement, IP address and behaviour. Not only does this raise questions about the line between security and privacy on the internet, it also looks at how algorithms behave as infrastructural forms.
In terms of infrastructural relations, we can examine whether Google CAPTCHA codes are an infrastructure, or when they are an infrastructure. To answer this, I will go through some of the qualities attributed to infrastructure by Star (1999) and provide a discussion at the end of the post.
- Embedded into other structures: Yes, within computers, phones internet, and internet browsers
- Transparent: It invisibly supports security tasks, and algorithms don’t have to be reinvented.
- Has a large reach: Yes, extends across the internet which has a global reach
- Learned as part of a membership: Yes, if membership means people who own computers and have access to internet. It is taken for granted unless they are computer illiterate or hearing/visually impaired. But it can read out letters for those who cannot see (also adds to large reach).
- Conventions: Specific way of inputting letters and numbers, creating new conventions – easier online security, computer scientists/hackers creating smarter robots (e.g. Computer vision technology)
- Embodies standards: The way they arrange and present letters to confuse bots. Providing ‘more human’ tasks like picking photos from a group, distinguishing distorted letters and numbers.
- Installed base: Yes, comes from algorithms, in computers, online. Algorithms make it random, computers/internet make it quick. But bots are becoming smarter, which means computers must also improve.
There are other points that Star mentions in her paper, but those above strongly suggest that Google CAPTCHA Codes act as digital infrastructure. The topic also questions what it means to behave as a human online, and versus robots. At what point (if there is a point) will computers be able to function as an ‘online human’? Similarly, Turkle’s study of MUDs back in 1995, showed ‘bots’ being mistaken for real people and ‘taking’ human identities through online interactions.
We can also look at these CAPTCHA codes as a completely different kind of digital infrastructure: Knowledge acquiring through collective brainpower, instead of security-minded. For example, Re-CAPTCHA codes use human responses to decipher words and numbers that computers could not recognise while digitising text from books, articles and Google Street view. Repeated human entries of these words are then added to the database.
Star, S. L. (1999). The Ethnography of Infrastructure. American Behavioral Scientist 43 (3), 377-391
Turkle, S. (1995) Introduction: Identity in the Age of the Internet. In Life on Screen: Identity in the Age of the Internet. New York: Simon and Schuster, 9–26