A very well written article. I had come to mutch the same conclusions as I've been dealing with image galleries. I came to that conclusion after using google, and gmail really sunk the idea home.
But I don't think Shirky has the entire right of it when he talks about the value of categorization. When I think about a Coppermine without any categories and albums, no matter how you want to slice and dice it, it no longer becomes a gallery.
Eliminating categorization when you have a job to do and you have an inkling of what you are seeking is one thing, ie: I need to find images that are based in red and show sports being played. Thats something searchable. What about things that aren't so tangable or searchable.
I often go to users galleries, not because I know they have something I want to see, rather I just want to see what they are showing. I have no agenda other than to find out if I like what they are showing. If I don't like what they are showing I move on, If I like it, I may poke around a bit more.
First you have to define ... what is this gallery, why did I choose to create this gallery, how will my users USE this gallery.
If you go across the population of users of coppermine you will find many different uses. I know of sites that show trains and planes for categorization, technical reasons why they might want a photo of a SPECIFIC train or plane. You have galleries used to show products, you have galleries to show production quality photos that the artist has taken himself/herself.
So a gallery may be used as a repository, reference, advertising, entertainment, the foundation of Coppermin can support almost any use.
The point I'm trying to make is that the value of the gallery as a whole which includes how the author has categorized the items has inherent value. A site categorized well will do better than a site on the same subject with the same goal and same content categorized poorly. (Even though both galleries have the same capability to search)
Think of it another set of terms. When does google fail you? It fails you when you don't know what to look for. What do you do when you have a desire to look at something but don't know how to find it? You go to a different source, such as a magazine (isn't that a category essentially? you wouldn't go to a Lowrider magazine to find out about playing the piano), or maybe even yahoo's categories. You search categories a little more until you come up with some specific terms, THEN you go to google and now you find success.
For example, If I wanted to buy a computer case, I could go to google and type "computer case", but I doubt I would be very successful.
What I personally would do is go to yahoo categories and look at "computer case manufacturers", I would then sample the main manufacturers websites, when I found something I liked I would then go to google and type in "ASUS T10" and I think I would be very succesful in finding a vendor for that case. Two different tools for two different purposes, but one deliverable: "I wanted to buy a computer case that I liked".
There's a reason it's easy to call a gallery owner an "author", because you give 10 people the same 1000 images, you will likely get 10 different galleries. The most successful gallery will be the one that categorizes the content specific to thier targeted audience that makes the users experience most pleasing.
Viewing a gallery is just like real art. Would you go to the smithsonian and ask the steward,"Um yes, I would only like to see the art that has red in it, and has people playing sports". Only if you had a job to do. Otherwise you pick the area that most pleases you and you would start browsing. A curator has the same job a gallery author has. Perhaps we should call them gallery curators instead.
The bottom line is that I believe that a gallery is more than the sum of it's parts. Your ability to search, categorization for its targeted users and quality content all work together to make a succesful gallery. If you are missing any of the ingredients you will have a lackluster gallery.
In future versions you will see us refine both the categorization methods as well as the search methods, but I can't see Coppermine eliminating either no matter what happens.