The New York Times Company

The new LIFE photo archive hosted by Google is a rich trove of images dating back to the 1750’s.  Google has opened up the archive to the public and ultimately it will contain over 10,000,000 images, most of which have never been published.   This is an exciting and amazing digital collection and Google is to be praised for its continued adherence to  its “mission to organize all the world’s information and make it universally accessible and useful.”   The images may be used for non commerical purposes without any speial permission.

Aside from the enormous scope of the collection, Goggle is making available relatively large versions of these images.   From what I can see, the majority of images are available with one side equal to or larger than 1280 px.  This size easily  fills the screen of the majority of computer monitors.  While not high resolution, this size does allow users to see the image in enough detail to examine and use it for a number of purposes.   In contrast, many museum sites offer far lower resolution images of the works of art they make available in their digital collections.

Unfortunately it has some significant limitations and is a bit of a disappointment, as reported in the New York Times by Virginia Heffernan on February 27, 2009 .   Although it contains “millions of photographs from the LIFE photo archive, stretching from the 1750s to today,” and is obviously a rich trove of images, both searching and viewing the collection is limited.  Her article gives a decent overview of the collections wealth and limits and her criticism of the lack of a slide show method for browsing through the collection or any segment of it is well taken.

Here I would just like to focus a bit more on the enormous limitations of the search capability.  Heffernan notes that “the home page for the archive has little charm, and the categories assigned to what are presumably the most exciting photographic subjects (“people,” “places,” 1860-1970 by decade) don’t exactly tantalize.”  She fails to note that these are not even true categories of the material.  When you enter any of them you find that fewer than 200 images retreived.   If these were true categories they would contain all of the pictures appropriate to the category.  Rather these are merely teasers, to give the user a sense of the contents.  In that respect they are not far off.  One can see the time period covered by the collection, and the categories do reflect the staple content of LIFE’s photos: people, places, events, sports, and culture.

The underlying problem is that the search function merely searches the captions associated with the photos and nothing else and, as Heffernan notes, these have obviously not been enhanced in any way by Google.   Some of the captions are pretty descriptive, as old LIFE readers like myself will recognize.  An example is the best evidence of what I mean.

The caption of the photograph above appears just as it did in the Google LIFE collection and, presumably, in LIFE when it was first published.  The Google results also show, in at least some cases, the date and the photographer.  There are also “labels” (tags) associated with an image, but there is no explanation about how these were determined.  The labels for the photo of the Prince above are: Hyderabad Royal Family, Animals, Azam Jah, Elephants, India, People And Customs, Living Standards, Taxidermy, Caste System, Vintage Print, 1940s.  These are sufficiently different from the caption to have required some additional organization.

Others photographs, however, do little more than provide the name of the person photographed.  This works well if you are looking for a celebrity, Hollywood or political, but not if you are looking for other things.  If you want to see the dress styles worn in a particular year or decade, forget it.  If you want to see male portraits, or city scenes, or generic objects like a soda can, a television,  a record album, or an elephant you may be lucky and find a few that were important enough to the editor to be identified as such in the caption, but otherwise you will be out of luck.

Google does appear to have an algorithm so that a search for a particular term will also look for variations on the word.  A search for “taxidermy,” for instance, retrieves a number of images with captions using some variation on “taxidermy,” like “taxidermist.”  But it didn’t retrieve the photo above which both uses the word “taxidermic,” and has a label of “taxidermy,” because, as I later realized, I couldn’t see all relevant results.

A search for “diamond” did retrieve a number of photographs of those newsworthy stones, but none of Liz Taylor decked out in hers.  I only found this one, among many others, by searching for her.  There were quite a few, but I didn’t find any with the word diamond in the caption or as a label.  Surprisingly,  my “diamond” search it also retrieved 138 photographs of Lou Diamond Phillips.  Who knew?

The most problematic aspect of search, however, isn’t the limited text or labels available to search; it is the fact that all searches return no more that 200 results.  Google doesn’t make any effort to explain this.  Since all searches retrieve only 200 photos, the same search may return different results each time you run it. It took me a while to realize that was the reason I was having such difficulty finding particular images I knew were in the collection.  Many of my searches failed simply because all results were not returned.  This is such a huge deficiency that I find it mystifying.  Could it be that this was a condition of the LIFE/Google argeement?

For a collection of this size there should be a way to obtain complete results, and to filter them so as to distinguish between the caption, labels, and the photographer.

The bottom line:  the limited results are very problematic and the search functionality is thus not  even as good as the advanced search for any images in Google.

I can’t help believe that the caption and other information that is associated with these images during the process of digital capture does not permit some additional ways to parse the text and improve search.  Is there not IPTC coding data for any of these photographs?   The first IPTC standard was approved in 1979, but there were guidelines for metadata for news images for years before that.  The weekly LIFE ceased being published in 1972, but it was published intermittently until 1978 and as a monthly from then until 2000.  There is some overlap.

How can Google make retrieval better?  Hiring experts to classify the 10,000,000 photos is not an option.  Google has create a user tagging tool, Google Image Labeler, “a feature of Google Search that allows you to label images and help improve the quality of Google’s image search results,” and this is a step in the right direction.   Like STEVE.museum, Peek-A-Boom and Artigo, this is an effort to harness the collective intelligence of users.   If Google can’t figure out how to do this effectively on a large scale, then who can?  If they succeed, I suggest that they gift this system to the museums of the world.  By the way, I’d love to know more a lot more about this endeavor.  I think this will be the subject of a future post.

In regard to areas other than search, what else would I like to see ?  For one thing, high resolution versions of these images available to scholars and academics for reprint at no cost, as the Metropolitan and other museums have done.

One final note: ironically, as I wrote this post, a segment of this same collection was made available on the site www.life.com.  The display is similar but, at least for the Prince of Berar photo, the information was more complete.  I learned that the photo was taken on Jan 01, 1946.  The credit is Margaret Bourke-White/Time & Life Pictures/Getty Images.  Moreover the image is linked to a Getty Images item record with a more detailed set of keywords (oh when will there be agreement on this terminology).    Interestingly,in contrast to Google’s reassuring message that the image can be used for non-commercial purposes without any special permission,  the LIFE site adds a potent threat for copyright violators.  The bottom line may be the same but the message is different.

Title:   Azam Jah
Circa 1946: Indian Prince of Berar of the Kshatryis caste, a mighty hunter & heir to the throne of Hyderabad, posing next to taxidermic specimen of the elephant he shot, in the game room of his palace. (Photo by Margaret Bourke-White/Time & Life Pictures/Getty Images)
Image #:           50873062
License type:           Rights-managed
Photographer:           Margaret Bourke-White/Stringer
Collection:           Time & Life Pictures
Source:           Time & Life Pictures
Credit:           Time & Life Pictures/Getty Images
Date created:           01 Jan 1946
Bar code:           00732768
Object name:           00732768.jpg
Copyright:           Time & Life Pictures
Keywords:           Hunting, Custom, Vintage, Game, Room, Lifestyles, People, Order, Lifestyles, Palace, India, Animal, Elephant, Natural Disaster, Religion, Prince, Specimen Holder, Hyderabad, Males, Portrait, Animal Themes, Taxidermy, Throne, Print, Advance, Hyderabad, Arts Culture and Entertainment, Human Interest, Prince. Find similar images
Release information:           Not released. More information
Restrictions:           Contact your local office for all commercial or promotional uses.
No resale application use without the prior permission of Time, Inc. Contact your local office to see if we can clear this image for you.
Availability:           Availability for this image cannot be guaranteed until time of purchase.

Getty Images reserves the right to pursue unauthorized users of this image or clip. If you violate our intellectual property you may be liable for: actual damages, loss of income, and profits you derive from the use of this image or clip, and, where appropriate, the costs of collection and/or statutory damages up to $150,000 (USD).

Tagged as: Elizabeth Taylor, Google, Image resolution, International Press Telecommunications Council, Lou Diamond Phillips, New York Times Company, Search Engines, The New York Times Company, Virginia Heffernan

{ 0 comments }