Metadata, you’re the Devil

by Bellhalla (CC: BY-NC-SA)

TP ( i.e. that person, a friend of mine) sent me some pictures, one of which inadvertently contained rather spicy metadata. The pictures were completely harmless in nature and very beautiful. But that one pic was one embarrassing accident.

And here is how it went. My Operating System displays icons for image files by extracting the preview that’s embedded in the pictures metadata. When TP cropped the image, the preview remained unaltered. As I saved the files to my harddisk one icon showed more than just a portrait. I won’t elaborate.

TP was obviously unaware of this detail and I just knew, he or she wouldn’t want anyone to see it. But part of the damage had already happened. I extracted the preview into a self contained file and mailed it to TP, explaining what had happened. Afterwards I eased erased every instance of the compromising data (You’d be surprised, how much you have to delete for this). TP was, of course, shocked. Luckily that picture had not been published anywhere. Just in case I provided TP with a cleaned up file for replacement.

This little story should be instructive to all of you previously unaware of the Metadata problem. It is not academic, it has had it’s impact in other cases already. The Metadata problem divides into two issues, one technical, the other human.

The technical part is hard to tackle. JPEG images work extremely well because applications handle them in a very liberal manner. Usually an application leaves all metadata untouched that it is not associated to itself . That metadata allows you e.g. to show off your carefully set exposure, aperture etc. settings with your pix on flickr. It makes putting your  images on a map a piece of cake (especially with the new iPhone). It brings of course it’s drawbacks as shown above.

The human side of the issue is the part to be tackled. Before you share pictures, wipe all their metadata. There is software to assist you.


Further reading:


Disclaimer

I did my best to protect TP’s identity. While the description of how it happened is 100% true, all identifying details about TP are altered or invented, if not omitted.