Sheron Resnick of 20/20 Software has sent a brief note with very cogent arguments as to why my ideas for an
Image Creator Locator are impractical and probably won’t work. She said:
“Dear Jim,
“I feel it’s important to respond to your idea of an image locator. While it sounds like an easy thing to do, there are many, many considerations to actually make it happen. First, to have a repository of all the images worldwide and have new images added on a regular basis, someone would have to have MASSIVE, MASSIVE storage capabilities. Secondly, that company would have to have paid support staff available 24/7, as well as an infrastructure that could support and manage that much content. Thirdly, that company would have to license or build software for visual recognition, then build a website that has not only search and display capability, but also upload capability for the agencies to add new content. And, also important, with that many images, the company would have to have buildings full of servers so the searches aren’t slow. Also, with visual recognition of that many images, there are bound to be many images that are very similar.
“So, who would pay for building and maintaining all this infrastructure? Would the contributing libraries be willing to pay a per image monthly or yearly fee for infrastructure, for storage, for support staff, and to license visual recognition software to maintain, and store their images in this type of database? And, what size image would the libraries have to store to be sure that their images are easily distinguishable? It seems to me that the only company with that much computing power, infrastructure, and finances, would be a company like Google, and would libraries want their content all stored with what would become a super-agency? On the other hand, have you spoken with David at the Copyright Hub? I believe they are trying to accomplish something similar, but perhaps on a much smaller and perhaps more manageable scale.”
I would like to examine some of her concerns and criticisms and offer some thoughts on each of the issues mentioned.
Storage Capabilities
First, massive image storage is getting cheaper and cheaper. I’m not proposing to store image files large enough to fulfill the needs of every image user. I am told that for the sole purpose of doing a visual match all that is needed is a very small, approximately 700K (700,000 bytes) image file. A terabyte is 1,000,000,000,000 bytes so roughly 1,425,000 files could be placed on a 1 TB drive.
Dell is selling the Dell Power Edge 100 Terabyte drive for $11,549. Roughly 285,000,000 files at this 700K each could be stored on 200 TB at an approximate cost of $23,098, (and the cost is going down.) There are also cheaper drives, but I’m not sure how efficient they are.
There will also need to be an identifying number attached to each image that can be linked to a database of contact information for the library or copyright holder. No keywords or caption information will be necessary because the sole function of this database of image files is to determine if one of them is a visual match for the file being searched.
Obviously, extra storage will be required to run the search software, but it should be minor compared to the space required to store the images themselves.
Currently, a number of companies including Microsoft are offering 1 terabyte of free storage to all their customers. Thus, it would seem that storage is not an expensive a factor – and getting cheaper all the time.
I have assumed that the total number of unique images stock agencies and individual photographers around the world have on offer for licensing is currently about 2 billion.
If storage cost for this number of 700K files is prohibitive it may not be necessary to store every one and still accomplish a great deal of what ICL is designed to do.
I believe most of the images that are used without permission are not being taken from the databases of the large agencies. Most users know that if they are looking for a free image they don’t want to search the files of a large stock agencies whose business is to license rights to images. Stealing from such sources is not a good idea. Rather they go to other sources like Facebook, Instagram, Pinterest where photographers have placed their images, or web sites where in the first instance, at least, the image was probably legitimately licensed. Then they grab the image because there is no clear way to tell if the image needs to be licensed, or not.
Assuming that is the case then the only images we have to worry about are the images that have been legitimately licensed at least once. No one steals the images that no one else wants to buy. My guess is that if we look at the large agency databases no more than 10% the images have actually been licensed. Maybe the percentage is a little higher, but not much. Thus, initially the only images we may want to put in the ICL are those that have been licensed. One way or another, it is these images that have probably made their way to one or more Internet sites.
Such a strategy wouldn’t prevent 100% of unauthorized uses, but it would give many users the choice of a legitimate legal option. If 85% of the professionally produced images currently found on the Internet are unlicensed, then the existence of an ICL should mean that a lot more images will be licensed, or at the very least many users will stop stealing professional images and keep searching until the find an image where licensing isn’t required.
Paid Staff To Support 24/7
Once the system is built, I don’t think the ICL will require a lot of additional paid staff to operate. It is the existing agencies and photographers who will actually be servicing the clients. All the ICL will be doing is directing customers to the agencies and photographers. The ICL process will be very automated.
Of course, if the number of customers increases dramatically, then the agencies and photographers may require more staff, unless systems can be developed to automate a lot of the sales activity. Suppose, you’re currently making 10 sales a day and suddenly you receive 100 request a day from people who have been stealing your images and are now asking to pay you money to use your images.
Yes, more sales staff may be required, but I presume you’d rather make the additional sales and get the money rather than give your work away. Of course, it depends on what those customers are willing to pay. If the price the customer is willing to pay is not enough to cover the cost of servicing them then the photographer you can always set a minimum price and politely tell the customer that the image is not available for anything less than that price.
At that point the customer may decide not to go through with the sale. But the customer is unlikely to go ahead and make an unauthorized use of the image. After having contacted the photographer that would be taking a great legal risk. If a photographer loses enough sales, he/she may decide to turn the licensing and fulfillment over to a stock agency that represents many photographers and images.
Presumably, all photographers trying to license rights to their images want to make more money. The only way to do that is to get more customers.
If a lot of requests for usage start coming from other countries in a language the photographer doesn’t speak then photographers may find it advisable to work with agencies, or licensing representative who speak those languages. Fixed pricing schedules and the ability to accept credit card, or Pay Pal payments may solve some of the problems.
Visual Recognition Software
Visual recognition technology is becoming very common. A number of companies are offering it, often as a free service. I doubt that the cost of implementing such technology will be a serious hindrance to anyone who can see some advantages to offering this type of service. More about possible advantages later.
Search And Display Capability
Yes, anyone who would decide to implement such a system would have some initial cost in building a search and display capability, but I would think it would mostly be a one-time cost that would not require extensive continuing maintenance, although there would certainly be some.
Upload Capability To Add New Content
Most medium size and larger stock agencies have systems that allow their suppliers to regularly upload additional new images. I don’t think building such a system would represent a major portion of the overall operating costs. I can’t see why something similar can’t work for the ICL.
In addition, I think their needs to be a system that allows each registered user of the ICL to remove some or all of his/her images at any time if the user finds certain images no longer need protection.
While there might be multiple agencies representing the same image, there needs to be an automated way for the registered creator of the image to remove the agency’s name from the list of representing agencies. This facility should be built into the initial system, but it seems unlikely that the technology would require much in the way of regular maintenance.
Building Full Of Servers
Given the size of 100 TB drives today, I don’t think we will need a building to store 10 or 20 drives capable of storing 100 TB of data each. (Check out the
pictures of the sizes of these drives.)
Super Agency
The suggested design for the ICL would not make it a “super agency.” The ICL would not have large deliverable files. The ICL would simply refer customers to existing agencies. The ICL would only offer visual search, not keyword search.
Some photographers dissatisfied with the price agencies have been charging might try to cut the agencies out of sales generated by the ICL by uploading their images to the ICL directly. Then they could set their own prices. Customers who use the ICL to determine if an image needs to be licensed could then go directly to the photographer.
On the other hand, even if some photographers leave certain agencies, it seems likely that going forward the agencies will make many more sales of images belonging to the photographers they continue to represent than they had previously. Many customers who had been stealing images will now start paying for the images they use.
Copyright Hub
The UK’s
Copyright Hub has some good features, and in many ways is a “proof of concept” for the ICL. However, in my opinion there are several flaws that make it an unlikely substitute for the Image Creator Locator.
First, it may have been designed to do too many things for too many copyright holders. In the end it seems unable to adequately address any of the goals. The ICL is focused on simply identifying and providing customers with contact information for still image creators only.
It has been almost 5 years since the Copyright Hub was launched by the UK government. It still doesn’t seem to be getting off the ground. The main problem seems to be in finding a way to independently fund its on going operations. The UK government funded the initial pilot project, but is unwilling to fund it on a long term basis.
The site was designed primarily to service creators from a single country, the UK, and to be used primarily by UK consumers. Given the nature of the photographic industry where creators from every country sell into every country, there needs to be one universal site that makes an effort to collect content from creators around the world and make it the go-to site for those, anywhere in the world, who are trying to determine if permission is required to use a particular image and how to go about obtaining that permission.
In my opinion, the idea of there being multiple sites in multiple countries is unworkable because there would be too many places for customers to go in order to do a comprehensive search. In addition, creators would have to place their work in multiple sites to be sure that they can potentially reach all consumers and that would be impractical.
What It Will Take To Making The ICL a Success?
In order for the ICL to be a success I think a large technology company will need to recognize the advantages that providing such a service might offer.
For example, possibly Adobe could see an advantage in knowing who the customers are that search the broader Internet for the images they need rather than going to a major stock agency. Are they just looking for free images? Do they just happen to stumble on something they like? Are they willing to pay a reasonable price to use certain images? Those people could be potential Adobe customers, not just for photos, but for all the many other products Adobe has to offer.
There is a good chance that Adobe doesn’t even know that many of these potential new customers exist.
In addition, many infringers are making unauthorized uses of images Adobe has in its collection. If these people can be encouraged to pay for the images they use rather than steal them that could generate additional revenue for Adobe.
Adobe might also realize that many of the photographers who have not contributed images to the Adobe Stock collection, might decide to place their images with a company that is trying to help photographers, protect their copyrights and maximize revenue.
Then there is Google. Google knows that many of their customers use Google Images search, but I don’t think they have a clear understanding of how many of those customers actually use the images they find on website, or maybe even in print.
If they were operating the ICL they would have a better handle on that information. Maybe it has marketing value.
They could also adjust their existing image search into two separate collections - images that Require Permission to Use and images that are Free to Use. Then customers who are looking for a free image could just search the free collection and not waste time searching the collection where permission is required.
Of course, there is the risk some of the images in the Free to Use database might be there because the photographer had failed to upload his/her image into the Require Permission database, or because the software failed to catch that the image is in the Require Permission database and failed to remove it from the Free to Use database. Maybe the database should be called “Probably Free to Use” instead of “Free to Use.”
The question: “Is it better to have a system that is 90% accurate, but not perfect, or is it better to make no attempt at all to better inform users and generate additional revenue?”
The ICL would also demonstrate to the European Union that Google is willing to offer a compromise that would provide an additional level of protection for some copyright holders. All in all, It would be very positive PR.
Microsoft might decide to operate an ICL as a way to draw more customers to its Bing search engine.
If the ICL existed, all technology companies might benefit because it would likely reduce the number of DMCA (Digital Millennium Copyright Act) takedown notices they would have to deal with.