On Instagram and ownership: an Advertiser’s viewpoint

Monetizing social media relies on re-targeting your social/photo content to advertisers, and social media sites won’t… justify their high valuations unless they do it.

In my day job I am a lead developer in an online-advertising company. I’ve also worked as a full time book author (design books, mostly web design, 2000-2005), something that has forced me to know the difference between copyright/ownership and licensing.

I’ve been following the recent Instagram change of terms with some interest. I have found no article stating what Instagram’s end game on this is and how to avoid the downside without simply closing your account.

Instagram
Instagram is watching you…

Here’s the why, what and how, as well as how to avoid your photographs ever being used in an online advert.

As this post is proving popular, I will keep it updated as the position changes.  Relevant changes are listed in the Updates section at the end of the post.

Ownership vs Licensing

Many photographers and bloggers don’t understand the difference between ownership, assignment and licensing. Certain organizations are disingenuously stating things like ‘you will always own your own photographs’, when licensing (rather than copyright or ownership) is the key point. Let’s clear this up quickly.

The moment you create new content, you can do what you want with it because you own it and all rights to it. Nobody can take those rights from you but you can sell them on (or if you are unlucky, inadvertently give them away). When you sell rights, you can either license or assign rights.

  • Assignment means you sell all rights to someone else, and that someone else can act as if they had created the content: basically, they now own it. You typically assign rights when you cannot make money on your ideas without significant input from someone else, and its better for the someone else to do it all. As no social media company is attempting assignment (yet), we won’t discuss this further, except to say that if you are ever asked for assignment of rights without an up-front cash sum equal to the value of the content for perpetuity, my advice is say no.
  • Licensing is a legal contract whereby you keep ownership, but allow someone else to use the content. Although you still own the content, you have allowed someone else to commercialize it and make money from your stuff. They will typically pay you a regular license fee and/or an advance and/or royalties.

So, by licensing your content,  the other guy makes some money and shares some of it with you. There’s a pretty good summary of licensing from a lawyer here.

When licensing using a weak or vague contract or one that favors the licensee, the fact that you still own the content means almost nothing because you have sold the rights to make money from your stuff (and potentially lost the right to exit the license if you don’t like what is happening with your content), making ownership a moot point.

Think of your photographs as if they were a house you have built. You may own it, but if you signed an unconditional and perpetual lease for its use to someone else then you will never make money from it (nor be able to live in it!). Owning the house in this case would be of zero value because someone else has all the rights to it.

Even giving away such rights for one room of the house would make the house useless. You could no longer control who comes and goes within the house, and what use it is put to. Nobody in their right mind would buy the house from you because (like you)  they have no idea what that other room would be used for in the future, and by whom.

Unfortunately, this is exactly the legal situation that millions are letting their photographs fall into on the web!

Before we look at how social media companies want to sell your photos and posts, we need to know some background about the web in general: specifically about what happens when we search for something.

Search and The Semantic Web.

The web has changed drastically between 2000 and 2006, and one of the biggest changes is the move from content to semantic content. Let’s unwrap that a little.

If you are writing a blog post, your text and images are the content. Other people can read that content if they know where it is. Most of the time, they don’t know where it is and they have to search for it.

Users don’t manually search for your content: they use a search engine.  A search engine is a set of software scripts that can see the words of your content but doesn’t understand the meaning (or semantics) of your content. So, along with the content of your blog post, you usually add tags such as photography and landscape to let the search engine know that this is what your post is about. Current search engines rely on additional scripts that can auto-generate additional tags by applying rules to your content (‘page scrape’), but we’ll ignore that feature to keep things simple.

You may also know that you should use words in your content that match the tags. So if you are using the tag photography, you should use the words photo or photograph often when you are writing about your work, and you should not use image and shot.

Why? Well, the more snippets that the search engine finds in your content that matches the tags, the higher the search engine assigns to the relevance of your content to the tags when a user Googles for ‘photographs of landscape’. If you use image instead of photograph, the search engine will get confused on whether you are talking about Photoshop rather than Photography, and your post’s relevance to the tag photography will go down (and your blog will appear lower in the search results).

Relevance is very important in search: it is what ensures your blog post comes near the top of search results, and ultimately, is one of the big factors that generates traffic to your blog post. The biggest factor is the quality of your content (but that means little if you do not make your content searchable so that people can actually find it!). However, its always good to keep your content relevant to the type of person your site is targeting (in this case, photographers).

Web developers call the tags you attach to your blog posts metadata. Metadata is ‘information about information’, and is basically labels attached to your content that allows machines to know what your content is about without needing to be smart enough to understand your post. In fact, the semantic web is the cornerstone of what we call Web2.0. With metadata, web developers can write code that acts as if it understands content. This allows applications to intelligently acquire and process information. The trouble of course occurs when that information is yours and you didn’t expect it to be used in this way.

Search and relevance were big issues for the web around the Web 2.0 years (2003-2008), but something else is more important now. That something is social media, and specifically how to monetise it. The way this is occurring is via metadata.

Social media: advertiser gold dust

There’s a big problem with all of this skullduggery going on behind your back though. It doesn’t actually work.

People generate their own content on social media sites, and their friends prefer to view this content rather than most other content because it has the ultimate relevance to them: the content is about them and their social group. You can’t get much more relevant than that!

Better still, users always add metadata to their social media. If you post a photo of you and your friends having a drink in a bar, you will inadvertently also include:

  1. The date
  2. The location (or at least, your cell phone will, because it will add geo-location data unless you specifically turn it off).
  3. How popular the picture is (via ‘likes’ from your peers).
  4. The explicit names of the bar and the people in the photo if you tag their likenesses.
  5. Link-backs to other users that would find the image relevant through your friended peers.

Another, less obvious social media output is review. You are more likely to believe that the Nikon camera you are thinking about buying is the one to go for if you see lots of reviews on Amazon and/or Facebook ‘Likes’ from satisfied peers. In fact, you are more likely to believe those reviews than you are to believe Nikon’s multi-million dollar advertising.

Finally, your ‘likes’ are a list of things you own, interest you, or that you want.

So advertisers have this massive amount of data generated for free, and that has high relevance in people’s perceptions on ‘what is cool and what is worth buying’. Users trust this data more than almost anything else when making buying decisions because it represents the impartial consensus. Best of all, the content is semantic by default because users always add accurate metadata. This means that software scripts can read the metadata to re-target social media content for advertising that is highly relevant to each person.

That’s a worrying concept: twelve year olds can be targeted by advertising that shows their peers appearing to endorse a product via tagged photographs taken on their cell phones and uploaded to Instagram, then licensed on to the advertiser (who may even deploy the advert outside a social media site) and although you as the photographer own the content, the licensing agreement doesn’t allow you to say no to this use.

This is the ‘house with one room leased’ example we talked about earlier, solidifying into what social media companies actually want to do in the future. That one room will be used for advertising and you will have no control over what your content is being used to sell, who to, and where… and this will occur for perpetuity.

There’s a big problem with all of this skullduggery going on behind your back though. It doesn’t actually work.

Branding vs Search

Search is pretty good. If you Google something you will usually find it. But what percentage of the time is the first result irrelevant? Frequently, but that’s okay, because you can carry on looking down the results, and the one you want is usually only two or three entries down.

Now, what if I was a Nikon advert, looking for reasons to buy a D600, targeted to you personally. The first post from your peer circle says ‘I use my D600 all the time, it’s very useful as a doorstop’. That person doesn’t want to like the D600 because he’s Canon. Software isn’t very good on sarcasm, so that’s the post that appears in the Nikon ad as the ‘relevant social media post from your peers’. It may not happen often, but I’m sure when it does, a screenshot of the ad showing this will go viral, and the ad campaign will get killed.

And that is the trouble with social media. Metadata is never 100% accurate. In search that’s fine, because search simply has to narrow down choice from billions of links to the five most likely, and we can then manually moderate the results and pick the best one.

For an advert auto-generated via software, the first result has to be 100% relevant because that is the one the ad will show. If that choice happens to be wrong, the brand can suffer.

The only way for the ad to get around this is access to human moderation, and that is the single reason preventing advertisers using user generated social media content directly in ads today. This situation may change as the technology gets better (and its something myself and my peers are working on for our day jobs).

I’ve seen and been part of online advertising that uses social media content and twitter feeds and can say it doesn’t currently work unless there is a human moderator. We often get companies coming to us using all the Web2.0 and HTML5 buzzwords, and we tell them ‘you need someone to sit there and check it all the time, otherwise the kids will upload pictures of goats all over your campaign, and then it will go viral, but not in the way you expected’. And they go ‘Ah, maybe we’ll try again in 5 years’.

But ‘5 years’ is only 5 years away.

Put another way, the wording of social media licenses are not important today because the technology isn’t there. Instagram don’t really care that you knocked them back in December 2012, because they don’t want this stuff in place for December 2012.

Update October 2013. Google has put in place changes in terms that allow them to use your review endorsements as Google ads from 11th November 2013. See note 6 at the end of this review for an update on this development. I strongly advise everyone to go into their Google account settings and uncheck the checkbox next to Based on my activity, Google may show my name and profile photo in shared endorsements that appear in ads.

The real issue is that 5 years from now when you totally forget about this stuff and the story isn’t newsworthy, the technology that can scrape your content for re-purposing in advertising will be there.  100% relevant,  auto-moderated content that is fit for re-targeting into advertising will be in place and working.

This is the end game social media companies are counting on. It is why Instagram wants to clarify licensing well in advance. It is why Facebook was so keen on hiding the tick-box that says ‘don’t allow my social media content to be searchable’ until people made a fuss.

The Facebook share price movement has been downward. It will go lower still if they don't find out how to sell user data to advertisers
The Facebook share price movement has been downward since its float: 35% down vs the Dow Jones as of this writing. It will go lower still if they don’t find out how to make a larger profit from user data…

Monetizing social media relies on re-targeting your social/photo content to advertisers, and social media sites won’t make enough money to justify their high valuations unless they do it.

It is worth noting the other side of the argument: that advertising is actually good for the internet. Without advertising, most sites would not be free to browse: advertising is the largest (and often only) income for all non e-commerce and non-subscription sites.

There is clearly a balancing act here between  a free internet for all and an internet that becomes a free-for-all.

Conclusion: Keeping your content out of Advertising.

So now we have a good understanding of what the technology behind this is and what advertiser’s (and therefore also Facebook/Instagram’s) end game actually is and when they might be there, how do we avoid getting our photographs being re purposed in online advertising?

The way advertisers would ever get to your content would be via its metadata, so you have to make sure you add enough metadata for search, but not enough for branding. If advertisers can’t find your photo, an advertiser’s license is useless because they will never pick your content.

How to do this?

  • Don’t add location data. If you are taking pictures of Hawaiian sunsets, don’t allow the camera to geo-tag the photos for Hawaii, because then an ad for holidays in Hawaii will easily find them. Geo-tagging is the single best way for an advertiser to associate your photo with a real place, sales area, or even demographic.
  • Add the correct search data that a human would search for, but the incorrect (or no) metadata that an advert’s script would use to confirm 100% search relevance. Again, for your Hawaiian sunsets, tag then with Hawaii and sunset, but also tag them with the geo-location for Siberia. That will reduce relevance for an ad looking for Hawaiian sunsets (conflicting data: less than 100% accurate) and nudge auto-moderation into a ‘something is wrong with this: don’t use it’ result, but would be okay for search (no human would Google for Hawaiian sunsets using geo-location co-ordinates, and they would be able to moderate on image only).
  • Add general information, and never specific brands. If you are blogging about your new tablet, tag it with Android, but not Galaxy. That way an ad for Galaxy will never use it because they can’t risk becoming a laughing stock by showing a Samsung.
  • Realize that search accuracy for branding is some way off and you have time to fix the problem through killtagging. A machine that can understand human content is 5 years off, but a machine that can spot intentionally deceitful human content is not coming anytime soon. Willfully miss-tag a small percentage of your content so that your social media can never be searched at the 100% accuracy that advertisers will need to protect their brand.  I call this killtagging. In the long run, this is better than simply closing your account because it reduces the usefulness of all social media to advertisers in general.
  • Don’t ever link an account you have with an e-commerce site (such as Amazon) with a social media site (such as Facebook or Google+). They want to do this so they can use your endorsement directly in an advert. Leaving reviews for people who are already interested in a product is different from allowing the same review to be used in advertising to people who are not already interested in the same specific product. The first is a social operation, and the latter is a commercial marketing operation, and I would expect to get paid for the latter.
  • Don’t ‘like’ a commercial product on social media. A Facebook ‘Like’ may currently go onto your wall, but I foresee that  such systems will eventually be turned into adverts that may appear outside a social media site (specifically Google and Google Ads, an intention made concrete by Google change of terms October 2013, to be implemented from 11 November 2013).

You may not want to do any of this if you have shares in Facebook.

As with most internet changes, the best way to deal with this is to either assume a scam and change your habits, or see an opportunity and embrace the change in a way that maximizes benefit to you

  • For avoiders: don’t upload any photograph that you cannot later delete completely including from search results. This ensures deletion is identical to delicensing, something you would need to do if a paying client wanted to exclusively license your online work.
  • For embracers: make the ad work for you: add a link to your website on each photo (and not a logo: that’s not as searchable), and only upload preview content (i.e. lower resolution, cropped, etc), so that you can argue that only the preview (and not the associated final work) falls under any license.

Updates

Update 2013: added text to the section Branding vs Search to discuss Google’s new Shared Endorsements system, and added note 6 in Notes.

Notes

  1. The only usefulness of ownership following licensing occurs if there are exit clauses in the contract (i.e. if the other guy stops paying you, or uses your content for out-of-contract uses), you can exit the licence, and ownership is still with you.
  2. It’s worth emphasizing that licensing is often limited to regions: if I write a book, then I might sell it myself in Europe, Oceania and North America (because I understand the language and markets there), but licence it in China and Russia (because I don’t understand anything about their language and markets, and want to work with a licensee that knows those markets better than me). Social media companies prefer to err towards global terms, so you will be giving up everything.
  3. Be very wary of any license that neither pays you nor has any exit clauses. In this case, even though you still have ownership, you can never exercise it, and your ownership actually has zero value even if you are allowed to sell your content to someone else, because nobody else will touch it. Far from it for me to suggest this is the route all social media companies would prefer, and their game plan is to slowly push everyone in this direction…
  4. It’s worth noting that the issue of licensing vs ownership works the other way for the Photographer when considering software: Adobe would prefer that the law sees you as a Photoshop licensee rather than a Photoshop owner. See this article.
  5. Using  words such as photograph and photo in your blog posts is actually more important than adding the photograph tag to your blog post, because most modern search engines weight what appears in your content higher than the tags. This is because many websites used to add tags to content simply the get higher search relevance (a process called cheating or Search Engine Optimization). Adding both the word photograph and using the photography tag and linking your blog to lots of other websites that use the same terms goes some way to getting high up in search if someone Googles ‘photography’.
  6. Update October 2013. Google has put in place changes in terms that allow them to use your review endorsements as Google ads from 11th November 2013. This seems fairly innocuous, except that the Google terms leave it open as to where those reviews are being harvested from. They specify the source as Google and ‘…third-party applications connected to your Google Account’, which opens up the service for anyone who adds a new checkbox to your terms with them. For example, they could easily use your Amazon reviews in the future, something that you may not have expected (given that you wrote the review in good faith to help people with their buying decisions, and not in an advert to entice people with no intention of buying the product), or Google may in the future buy a review site so it becomes part of Google without you knowing.  Google, rather speciously state ‘You retain ownership of any intellectual property rights that you hold in that content. In short, what belongs to you stays yours’ without specifying the full terms of their licensing rights. If you have read the earlier sections, you will already know that this is a means for Google to trample all over your ability to profit from your work (its the ‘house with one room’ example again!).

Painting with ice

Natural textures tend to be similar, so using ice textures to change a sky is not as odd as it might seem

Although Photoshop can now generate organic noise and texture, back in the day we had to use our own. Even now, I find photographing my own textures, and using them instead of synthetic textures is much better. One of my favorite sources is ice.

A few days ago, I was wandering about the garden with not much to do. I came across the wheelbarrow. Not much to look at…

A boring wheelbarrow
A boring wheelbarrow

It contained  leaves covered by iced water. I pulled out the ice to drain away the water and was about to walk away, when the sun came out. Never mind the mundane subject: I could only see the wonderful textures the light had suddenly created. I ran for my camera.

Textures 1
Textures 1

The photos here have not been re-colored in any way: I just waited until the side of the wheelbarrow was in full sun.

The photos have an ‘alien sunset’ feel to them. This is the sort of thing that comes in very useful in Photoshop, when you want to layer textures over an image to give it a more painterly or grungy effect. Using colored textures such as these, rather than just solid color filter values) is also useful when you want to add warmth or coolness to your photos.

Layering the ice texture onto a sky
Layering the ice texture onto a sky

Here’s what happens when I layer my alien sky over a simple and initially boring tree line.

I took some more photos…

Texture 2
Texture 2

Using a macro, I now zoomed into the ice surface.

Texture 3
Texture 3

1 to 1 pixel closeup of ice
1 to 1 pixel closeup of ice

All the images are taken with my 24MP Sony Alpha A77 at ISO 64. Looking at an actual size section above, you can see that there is zero noise, and lots of lovely organic texture (and as you can see, ice is rarely white: there’s every color of the rainbow there if you look close enough). Perfect source files for Photoshop layer effects!

Okay, onto an example of using our textures…

Fractal flowers
Fractal flowers

The photo on the left is as-shot. Looks nice, but for me it was missing something: the background is a little boring and dead. Layering in some ice texture in these areas lifts the final photo up considerably by adding movement, turning the final image into something much more organic.

Conclusion

Consider using your own textures rather than relying on Photoshop generated textures, except when you want pure noise. Natural textures tend to be similar, so using ice textures to change a sky (say) is not as odd as it might seem: both are natural fractals and therefore easier to blend into each other than (say) starting with synthetic Gaussian noise and trying to blend that into the sky.

Using your own photographed textures in Photoshop rather than using something like Instagram or Pixlr is preferable because the former (a) gives you a higher resolution final image, (b) you can create something more unique and personal to yourself because you are the sole content creator for the image and effect, and (c) with Instagram, any idiot who knows how to press a button can’t do just that and copy you!

Notes

  1. Photographic equipment used: Sony Alpha A77, Sony 50mm/f1.8 prime, Sigma 24-70/f2.8, Tamron 90mm/f2.8 macro.