Happy Accidents, Semantic Photography, and Tim O’Reilly
May 6, 2010
The first US domestic talk I gave about my book and the semantic web is now online. It was at Marco’s SemWeb meetup in December.
Today I’m going to answer questions from readers.
Chris writes: My questions revolve around whether the data locker will ever be accurate enough or care free enough to be useful for true pull. Just a few errors could twist the data into an unusable mess. It’s a chicken and egg thing: to get to the point where the data is useful it has to first be useful.
It’s clear Chris has read the book and is a keen observer. In real life, these things are difficult to pull off. However, we’re already on the right track with Master Data Management, Product Lifecycle Management, XBRL tools, and Data Synchronization. It’s not easy, but the tools are getting sharper, and the rewards are clear. The SEC has mandated XBRL not to give everyone an extra expense (which is normal for the SEC) but in this case to give people a break. Once everything speaks XBRL, the tools will get cheaper and the overhauls will be much fewer. There’s a tremendous financial incentive to going with common formats – companies can focus on solving problems rather than on converting data from one format to another. Look at Adobe. I keep saying they have so many products that read so many formats that they have to keep making patches and connectors, and now the patches are becoming the products. You can’t own the world. You should let your customers own and control their data, and just add value without reinventing the wheel (more on this in a future blog). Over time, the errors go away (and, in case anyone hasn’t noticed, we have plenty of errors with the way things are today). Open linked data is already proving to be useful and will continue to get more and more investment. The days of proprietary data formats are waning.
Tim writes: I recently finished your book PULL. Very insightful ideas and information… as a photographer who shoots commercial and stock photography I was wondering if you had any particular ideas for the photography industry (concerning the power of pull) that you could share via your blog? Or other resources that could be used as a guide? It seems most advisors in the photography industry are in denial and confusion as to how to deal with the internet; your insight would be, to say the least, enlightening.
Oh, I could write a book about this. Hey, I kind of did – it’s about setting up the infrastructure so things work automatically. A short list of my photo hopes, demands, and predictions:
- The main DSLR companies (Canon, Nikon, Olympus) should agree on a new DSLR lens mount standard, one that includes both mirror and no-mirror options for autofocus and has room to expand into the future. We don’t want to wait 20 years for this, and we don’t want to rush in and make something no one wants, but we can follow the example of the Micro 4/3rds standard and put something together everyone can build on. It’s hard. There will be a lot of complaints. But it’s doable, and it’s what consumers want. No one uses F-mount lenses any more. In 20 years, I hope the current mounts are not the state of the art, and it would be a shame if they weren’t compatible.
- Let’s lose the digital film. We don’t need CF or SD cards! We need bandwidth. Let’s shoot and ship and go right to the server.
- The server should recognize people, places, and things. The camera and server should collaborate to do our tagging for us, so we don’t have to do it manually. Everyone in a particular photo should get a notice that it’s up and available automatically, after the photographer has had a chance to make edits and changes.
- Stock is interesting. What if you’re an art director and you want a certain shot? I’ve spent days and days wandering the web looking for specific shots. Art directors should be able to put a quick request online and have photographers send their stuff in or bid on going out and getting the perfect shot. This would blur the line between stock and commercial shooting, and it needs blurring.
- More open standards for tagging photos will let us store photos anywhere and mash them up as we want. For example, I should be able to put a timeline together of every photo I’ve ever been in, even if I’m in photos on hundreds of servers around the web. Today, all the stock places have their own in-house tagging vocabularies, and that’s too bad. The more we share vocabularies, the more powerful our searches will be.
- Finally, we’ll eventually have all our apps online. Desktop apps are not the future. I just looked into buying a skin-smoothing plug-in for Lightroom, and it’s $200! Give me a break. Charge me 30 cents every time I use it instead – you’ll probably make a lot more money. The online photo editing software era isn’t here yet (though the sale of Picnik should encourage more venture capitalists to fund such tools), but I hope it’s just around the corner.
Let’s face it. The more connected we are and the better the gear, the fewer people will make a living solely through photography. I already lived through that in type design. There is no law that says all photographers get to earn a good living doing photography forever.
Another Tim writes: How do companies differentiate their products/services via semantics given that the differentiation will almost certainly not be structured? The trade off between innovation which implies a new way of doing things which implies a new standard which will/can not(?) be included in semantic search?!
This is probably the best question I’ve gotten so far. A lot of people say, “If all the data becomes structured and you can say exactly what you’re looking for, what about the happy accidents?” This, to me, is not a good question. For starters, you can always turn off your semantic search and go wander around the keyword web like back in the old days. Second, it takes 20 unhappy accidents for every happy accident – wouldn’t you want to have a personal ontology that can help you find things in your periphery that you might actually be interested in? Now Tim asks something different. If I am able to specify exactly what I’m looking for, but I don’t yet know about something new, how will it ever find me? Today, we do that using notes from friends, reading in magazines, reading news blogs like Gizmodo or Techcrunch, etc. So, for starters, we’ll hear about new things from other people. In addition, our personal ontology and our query parameters will help make our queries fuzzy, and they will know what we tend to go for and what not, so they will actually help us find things we might be interested in even better than anything else we’ve seen so far.
Darlyn writes: I love your book. I wished you had more videos online. It helps me grasps the concepts better. I am writing you to ask, what are your thoughts on Tim O’Reilly’s statement regarding the semantic/RFID web being “an evolutionary dead end”? His definition of the semantic web is restricted only to RFID aspects but in your book you delved deeper. Does he, like many others, just have not yet grasped what the semantic web is?
First, I do have more videos coming. Come back in a week or so and check in for an exciting announcement. Second, I have tremendous respect for Tim O’Reilly. I think Tim is a force of good in the world, in the same way that Bob Marley was. I did see this video, which I think was made a year ago, and I think it’s certainly worth watching. But I agree with you that he’s wrong about RFID. I just bought a pair of $16 shorts for my little boy at Gap and it came with an RFID tag. Many of the items in their stores have RFID tags now, and more are on the way. Tim says things will be recognized without tags, and for faces, people, places, and artwork he’s right. But when I was at the Galeria Kaufhof in Germany and I took an armload of shirts to the counter and just plopped them down in a pile, the display immediately showed the list of shirts, their prices, and my total. No camera is going to do that. As Tim himself says in the video, when a Wal-Mart customer pulls a product off the shelf, Wal-Mart says that’s a vote. This is pull, and he’s squarely onto it. So I think he must have been thinking about RFIDs in a bigger context, and he was just pointing out that there are other ways to disambiguate, so let’s use whatever’s appropriate. That’s my take on it anyway. Thanks for sharing that video link – I enjoyed watching and listening to Tim O’Reilly, thanks for sending me that link.
Speaking of links, this is a huge web site. See the links on the left hand side and do some exploring before you leave.
And email me if you have questions about pull, the semantic web, and business: david at thepowerofpull.com
