Tag You're It

I think by now, we can all stop our frantic worrying and just agree that when it comes to their mission to “organize the world’s information and make it universally accessible and useful”, Google has come further than any mechanism since we invented buildings to put books in and called them libraries. We all now live in a solidly post-Google world thank-you-very-much, with a seemingly endless stream of filtered and categorized information at our fingertips. The newest article of intellectual faith of the twenty-first century is that the addition of user data improves any system, and that Google are the masters of it.

As long as we’re talking about text that is.

Google, like almost every one of its major competitors, has struggled with the ability to make image search in any way as effective or meaningful as their world changing text search product. For example, if I search on Google for the word “God”, the first image I receive is a badly drawn child’s cartoon of a man shaking his fist at the sky, hardly representative of mankind’s lifelong common artistic journey to depict his Creator. The problem is simple, there is a meaning gap between how Google’s search engine detects meaning and how we as visual and conceptual beings attribute meaning to images. Even today, Google’s vaunted search algorithms have no way of knowing what a image is of unless someone takes the time to tell it.

The truth is that text, by virtue of it’s makeup, contains it’s own definition and images don’t. By looking at a picture of a tree, we don’t automatically communicate the word “tree”. This means that in order to successfully operate at Google’s habitually high standard of search an image file requires a human created text file to describe it. This text is called “metadata”. This basically means that when a search engine is searching through image files, it’s actually searching through the descriptive text files that tells the engine what the image is contains.

The problem is that of the billions and billions of images uploaded to the internet every year, the majority lack any “metadata” of any kind other than their filename, most of which are at best, only vaguely descriptive. This means that images still require actual humans with actual eyes to actually supply the missing meaning for the computers to categorize. If Google paid a mllion people to spend all day assigning adequately descriptive metadata to the images on the web, they would still have to go through billions and billions of new images each and every day to even get started. This means that with regards to image search, Google’s mission has hit a seemingly unscalable wall of meaning because the best image detection software on the market still has problems telling the difference between a small photo of a horse and a large picture of a dog.

So, is there any way of getting over this problem or should we just accept a lower standard of search and move on with our lives? The really interesting thing about this it that there is already a search engine that has successfully managed to mobilize millions of people to work tirelessly for free supplying metadata for all it’s images, one that now seems to be winning the huge game of tag that is the internet.

Facebook.

The current giant of social media relies on one simple social mechanism for search success: people will voluntarily add incredibly detailed metadata to photos that they and their friends appear in, making them easily and meaningfully searchable. It is this loyal user behavior more than any other that has successfully separated Facebook from Google and means that Facebook now poises a genuine threat to Google’s dominance of the internet. It is this new reality where they have a willing armies of meta taggers at their beck and call that Facebook are now looking to take advantage of by for the first time ever allowing brands to be tagged in photos. This means that if a user wants to show off their new Nike’s or Christian Dior, they can tag their photo. This enables Facebook users to associate themselves with a brand they love while at the same time creating the basis for an incredibly lucrative Pay Per Click advertising system. The desire for a solid PPC monetization system is no surprise considering Facebook's CEO Sheryl Sandberg used to be Google's Vice President of Global Online Sales and Operations.

Facebook has always been at it’s financial center, a marketing platform, but this may be a departure as risky as Google deciding to sell adspace on the pristine whiteness of their homepage. Now that the cat is out of the bag, Facebook will have to be incredibly diligent in preventing brands from giving incentives to users to turn their photos into yet another messy, over saturated adscape that will dominate the user experience.

In the meantime, if Google or one of their competitors want to bring image search into the 21^st century they will have to either figure out how to mobilise the internet into caring about providing comprehensive metadata for the images they upload, or use their vaunted engineering core to solve the problem with software. In the meantime, tag your photos people. Social media has taught us that they get lonely otherwise.

-----------------------------------------

Rudhraigh McGrath