‘TinEye’: Searching for the DifferAnce
See also the complete TinEye’s analysis here
‘Language depends on difference, as Saussure showed … the structure of distinctive propositions which make up its basic economy. Where Derrida breaks new ground… is in the extent to which ‘differ’ shades into ‘defer’ … the idea that meaning is always deferred, perhaps to this point of an endless supplementarity, by the play of signification.’ (Christopher Norris, 1982)1
Difference is not purely ‘otherness’. At least not if we take the notion of the French theorist Jacques Derrida who by using the anomalous ‘a’ in his way of writing ‘difference’ – differance – set up a disturbance in the general understanding of the word (concept). By doing this (applying a distinctive mark in the word difference to form differance) Derrida opened up the insight in which a word could be continuously in motion to generate new meanings without erasing the trace of its other meanings. Meaning as well as representation is in this sense never finished or completed, ‘but keeps moving on to encompass other, additional or supplementary meanings’.2 In other words, signification is always deferred (postponed) and Internet image search engines like TinEye enable us to be aware of web image differantiation.
Web 2.0 application TinEye, created by Idée Inc., was officially launched on May 6, 2008. It was announced as the first image engine on the Web to use image identification technology. As TinEye’s website describes; by ‘using sophisticated pattern recognition algorithms, TinEye creates a unique and compact digital signature or ‘fingerprint’ for each one [TinEye’s crawled images from the web] and adds it to the index’. In contrast with regular search engines such as Google Images where the user has to enter text in order to find web pages that contain that specific text, TinEye has the user submit an image to find web pages that contain that image. It works, in other words, as an iconographic search engine from which the user, by uploading the image of his choice or by submitting an URL (of page or image), can find where and how a given image appears all over the web – even if that image has been modified. This last point is important because by analysing image attributes and comparing uploaded the image’s fingerprint with every single image in the TinEye search index, TinEye offers a detailed list of any site using that image which may even have been cropped, resized or heavily modified (however, the first images in the result are the closest match to the submitted image). In this way, TinEye makes visible Derrida’s deferred process of meaning in web image representations that is emphasized by TinEye’s option to compare the user’s uploaded image with any other result from the search task (see for instance TinEye Example Matches or Embeddable Widgets). Therefore, image differences and deferred image meanings could be explored with TinEye.
The problem with TinEye is that at the present day this search engine has only indexed a fraction of images on the Internet (about 487 million images). Therefore it can hardly compete with the dominant image search giants such as Google, Yahoo or AltaVista. An additional problem is caused by its troublesome accessibility. Although since August 15, 2008 invitations are no longer required in order to access TinEye services, one still needs to sign up and accept all terms of service that limit TinEye’s easy and direct way of use (compared to the aforementioned search engines that work by just typing the name of a desired image).
TinEye’s group is still developing user friendlier methods to access their services (such as TinEye’s plugin which is intended to search for images in a faster way using a FireFox or Internet Explorer browser), but the obstacles described above make TinEye at best a search engine for experimental usages. Nevertheless, the implications that TinEye incurs in the sense of it being a search practice based on image instead of text open up reflections around the way the current visual culture reinforces both its presence and the importance of visualizing as much as possible of daily life processes. In this case by searching for visual information through visual information itself.
In addition, search engines such as TinEye raised the question what this kind of image searching practice could mean for the society at large if the indexing database methods are extended and further developed to become much more efficient (for example by a much more sophisticated image identification technology). For instance: what could this practice mean for companies if they were able to recognise more accurately the way in which their original media products (logos for example) are modified and exploited by their consumers/users? Furthermore, with regard to the popular culture side, what could this practice mean if one becomes much more aware of what peer producers do in the re-appropriation of cultural icons labelling their own ideologies? In this way, the awareness of the differances across cultural images concerning, for example, the logo of the brand Puma might give the company important insights into the ways popular culture transforms this image into a certain ideological statement (for example by changing the ‘P’ into ‘F’) and therefore have it adapt its commercial strategies. Similarly, knowing that the ‘Fuma’ statement is already made by finding it through TinEye, the user/consumer may adjust his statement or apply new forms of deferred meanings to consolidate alternative significations around this cultural visual product in particular – either against the grain or not.
1. Norris, Chistopher. Deconstruction: Theory and practice. London. 1982: p. 32.
2. Hall, Stuart. ‘Cultural Identity and Dispora’, in Colonial Discourse and Postcolonial Theory: A reader, ed. P. Williams and L. Chrisman. New York: Columbia Univ. Press. 1994: p. 397.
*By means of ‘The Idée blog’ I was informed that TinEye has now indexed around 900 million images which is obviously almost the double of the 487 million images as described in this analysis. However, this number of indexed images was taken from the introductory to TinEye videoclip (see intro) which then seems to have not been updated yet. Anyway, thanks for the comment.