The vexed question of who was responsible for which claims in the infamous Iraq dossier revolve around minutes, diary entries and tape recordings. But you could also find out who edited it, and when, by looking at the document itself. The same goes for the fake Jamie Oliver cookbook circulating on the net. Many programs store this kind of information as metadata in the file.
Metadata doesn't collect information for a sinister purpose and it is often very useful. You can search for files you created on a particular date if you can't remember the file name or the folder you used. Digital photos include the make and model of the camera you used to snap them, and Windows XP copies that EXIF (exchangeable image file format) information with the photo. That means image editing software could correct for the colour gamut or lens characteristics of a particular camera. You can also compare photos taken with a new camera to old favourites, or check the lighting conditions and exposure on a particular shot.
But information about who opened a document or when it was last changed could also be useful if there is a dispute about the document. If you send a report in late, telling your boss you had problems with your email system, it is suspicious if the timestamp on the file shows you opened it several hours after you said you had first sent it. The sensible response would be to explain you opened it to check it one last time before you resent it. That won't work if a photograph claiming to show an event turns out to have been taken on a different day.
Metadata (or keepware, as some of the programs that save it have been dubbed) adds yet more details to office documents. If a document passes from one machine to another, it can include information about previous edits. Most Microsoft Word documents include a revision log listing the last 10 users who opened the document, and it is usually simple to work out the name of the writer from that. Changing the name of the file doesn't remove the information, and neither does using the Save As command; the revision log shows the name of the file at each stage, and the folder it was in - so you can see if files were copied on to a floppy.
Usually this log is hidden, but it is not hard for a programmer to write a utility to retrieve it. When the government published the second dossier on Iraq's security and intelligence in February as a Word document on the Downing St website, one curious reader looked at the revision log. He retrieved the names of a Foreign Office official, a Downing St official and two people in the prime minister's press office, along with anonymous users at the Communications Information Centre, where the dossier was drafted (and where the material from an old PhD thesis was incorporated).
Even without a utility, the file properties tell you the original author, how many times the file has been saved and how long it was open for in Word. They might include the title of an email used to forward the document and to whom it was addressed. As David Bennie, Microsoft UK product marketing manager for Office, has found, this can be embarrassing or amusing: "I've seen PowerPoint presentations where I've thought 'that's really good' and I open it up and the file properties have my name in. My presentation has been re-edited!"
When you collaborate on documents, turning on Word's feature to track changes - a feature in most word processors - means that even if they don't include comments, you can see exactly who typed particular sentences or changed individual words when you get the document back. You can use this if you are not sure about the changes you are making. Changing the view shows you the original version with the edits (known as markup) - or the final version with the changes incorporated, but it doesn't delete the comments. If you send the document out without remembering to accept those changes and delete the edits, anyone can turn on markup and see what was changed just before the document was approved.
In the current beta of Word 2003, whenever you open a document you didn't create, it automatically opens in the new Reading Layout, with change tracking and markup switched on to make it easier to collaborate on documents. That means if you forget to accept or delete edits, they will show up automatically.
Bennie thinks "people need to be aware that if they're leaving tracked changes and revisions in their document, they need to be taken out at some point". He also points out that you can tell Word to warn you if you have left tracked changes in a document before you print, save or end it, a feature he uses.
You can also automatically strip out personal details from the file properties when you save documents, or make hidden markup visible as a reminder when you open or save a document. These options are all in the Security section of the Tools| Options dialog and an IT administrator can set and lock them as part of user policies.
Other tell-tale details can lurk in documents. Several word processors, including Microsoft Word, have a Fast Save feature which includes the original text and instructions to hide deleted or changed text rather than saving a file that only contains what is in your document. Saving just the changes means the file saves more quickly but, again, a programmer would be able to retrieve some of the information.
Check the notes fields in presentations: these are often a rich source of confidential information put in for background and never removed. Styles, templates, hyperlinks to documents on internal servers and information in headers and footers could also reveal more than you wanted about a document's origins: they are often named for the company that wrote them. And if you have created multiple versions of a document inside a single Word file, remember to delete the ones you don't use.
Saving a file as an RTF (rich text format) document will strip out metadata and the other details, or you can do what the government did when it published a report on Iraq in June and distribute it as a PDF (portable document format) file instead. Bennie suggests using Office 2003's Information Rights Management tool to stop people accidentally forwarding documents with confidential information, and says Microsoft is using it internally to protect any document that includes the launch date for the new Office.
It isn't just text you need to think about. When you crop images in word processors or desktop publishing programs, the original image is often stored with the document - unless you choose to make changes permanent. American Tech TV presenter Cat Schwartz was caught out by a similar feature when she posted a photograph of herself on her blog. A curious reader who downloaded it found that while she'd cropped the photograph to show just her face, the file appeared to still contain the thumbnail image of an original, considerably more candid photograph.
How to minimise metadata in Microsoft Office documents
http://support.microsoft.com/default.aspx?scid=kb;en-us;223396
Coping with metadata issues
www.microsystems.com/Shares_Well.htm
