We are the robots …

( With apologies to Kraftwerk … )

Like a lot of people I post images to a photo sharing site. I use 500px but there many others, the most famous probably being Instagram. 500px has some features for marketing and selling photos and when I started I had some hopes that I might be able to supplement my camera gear budget by selling some pictures. However I was rapidly disabused of this idea, partly because my photos weren’t really good enough but mainly because they were lost in the tsunami of other material that was available.

After the initial disappointment that I was not going to get rich or famous via this route I realised that 500px was still useful because I could use it to force me to actually make time to create something on a regular basis. I got into a routine of posting one photo a week which meant that I had to be routinely going out with my camera, thinking of new ideas and improving my photo skills. This has worked well and, other than when I took a break to move and renovate a house, I’ve posted regularly and I now have over 120 good photos in my account.

( There’s an interesting story about quantity v quality that I came across here – Why Trying to Be Perfect Won’t Help You Achieve Your Goals (And What Will). I was initially a bit sceptical about this but, after my 500px experiences, this approach certainly works for me. )

However, human nature being what it is, I still get annoyed when I see photos that I think are no better than mine getting many, many more upvotes and likes. I know that the conventional wisdom is that you should spend lots of time on the site interacting with other users, commenting on photos etc. but I was doing that anyway. If I extrapolated from my activities there just weren’t enough hours in the day to get the number of followers that some people had. Recently I was motivated to dig a little deeper when I came across these images: ( I’ve taken screenshots rather than link to the original images because these accounts can vanish quite suddenly )

This is obviously an advert for office removals in Noida, India. However it’s had 23 views and 15 people have liked it. Looking in more detail at the people who have liked it then the three most popular have 9,798, 6,538 and 5,086 followers. Looking at the images that they have posted then they’re pretty good, none of these people live in India, and so there’s no obvious reason to like a stock advert image.

Or what about this one, advertising gym mats I think? Again it’s got 3 likes quite quickly from people with 11,647, 6,833 and 2,910 followers.

Obviously there’s something strange going on here and a quick search soon reveals some likely reasons. Essentially the premise is that if you follow somebody and “like” one of their images then they’re more likely to follow you in return. Your photos then end up in their feed and you’ll more than likely get a few likes for each one if your photos are any good. You, of course, will never look at your followers’ pictures because there are so many of them which rather ruins the whole “photo sharing” aspect of these sites. This approach appears to work on both Instagram and 500px.

I think that the maths make sense. For example, at the time of writing, the top landscape photo on 500px has 1,500 likes. For my photos then, of the people who view them, around 50% like them. If I have 10,000 followers then to get a photo to the top I only need 15% of them to like it which seems entirely possible.

The basic approach using an automated script and the 500px / Instagram API is as follows:

  1. Monitor the site for newly posted images, possibly filtering only those posted by newly registered users.
  2. “Like” that image and follow that user
  3. If, after a couple of days, that user hasn’t followed you back then unfollow them
  4. Repeat until you’ve got thousands of followers …

A clue to that this is going on can be seen by hovering the cursor over the ” XXX Following” text on a 500px user’s landing page. You’ll get a pop up showing how many other people that user has followed and unfollowed over the past week. For the two examples that I showed above then I saw some evidence of this: “This user followed 202 users today (and unfollowed 11). They followed 789 (and unfollowed 158) this week.” I think he needs to slow down a bit otherwise he’s going to get found out 🙂

Although this is prevalent on 500px there seems to be more happening on Instagram. This post by Tim Großmann from April 2017 is a very interesting analysis and experiment. He originally started the experiment to learn about Python and Selenium WebDriver ( web site testing tools ) and freely admits in his conclusion ” … As of writing this, I have ~2,800 followers. I plan to keep running my script until I either get banned from Instagram or rise to the highest heights.” It’s quite a lengthy post but well worth reading if you’re interested in the details. It’s interesting to note that he hasn’t been banned yet and he now has 4,226 followers. If you want to try for yourself then the code is freely available on GitHub – Instagram Bot.

Finally a couple of interesting comment threads on this subject on Reddit: ( Warning the language on Reddit can get a little robust 🙂  )

This final example is from Instagram ( via a Buzzfeed article ). Yes, that is a photo of a black square with 1500 likes and user salvatore_cafiero91really loves the photo” 🙂 The Buzzfeed article gives a lot more detail of the techniques used but it comes down to the same basic principles even though it seems to be via a commercial service ( Bots as a Service – BaaS ? )

All this raises the question of why you’d want to go to all this bother. Well I can think of three basic reasons:

Curiosity

As an engineer I can definitely relate to the “I wonder what happens if I do this …” question. The post by Tim Großmann is a very detailed and interesting account of such an exercise.

Bragging

This one’s also obvious I suppose. In my experience anything that can be measured will eventually become a form of competition or one-upmanship. Most people get over this by the time that they’re 15 years old but I guess that these days social media pressure is such that getting thousands of Instagram or Twitter followers is serious business which brings me on to …

Money

In my naivety I wasn’t entirely sure how to make money from Instagram but, as ever, 30 minutes with Google was very educational 🙂 Discounting the obvious “Use Instagram to promote your existing business” or “Sell your photos” advice there appear to be two main approaches.

Firstly there’s “Affiliate Marketing” which is exactly what it sounds like. You post about a product and you get a cut of the sales ( if any ). I suspect that bot generated followers may not be that helpful here because of the scatter gun approach of recruiting them. You may be interested in beauty products or video games or whatever but only a small percentage of your followers may be.

Secondly there are Sponsored Posts which can be summarised as  ” … a piece of sponsored content on Instagram is a photo or video that highlights a product or a brand. These posts are accompanied by captions that may include branded hashtags, @mentions, or links.”  From what I can make out you get paid a flat fee which is much more compatible with your bot generated army of followers 🙂

In either case it’s important to be an “influencer“:

An influencer is basically anyone who’s built themselves an online reputation by doing and sharing awesome things online. To their audiences, influencers are tastemakers, trendsetters and trusted experts whose opinions about certain subjects are respected.

This is all very well but, let’s face it, we all want to know “how much”? That’s tricky to answer but you’re probably going to need at least 5,000 followers to get started:

  • Buzzweb implies $700/month for 5000 followers, $5,000/month for 50,000 followers.
  • A Buzzfeed article says ” … you could expect to make 1% of your following per post (someone with 10,000 followers could charge $100 for a sponsored post, for example).
  • A survey suggests that a mid-level influencer can charge $200-$500 per post. ( The survey however doesn’t define what a mid-level influencer is )
  • There are various sites that purport to tell you how much you can expect if you give them your Instagram name – Tim Großmann above can apparently expect to earn between $30 and $60 per post. If you’ve got 150,000 followers then that rises to $300 to $500 per post.

Warning – take these figures with a very large pinch of salt. My personal BS meter went off the scale several times while researching the information. This is not meant to be careers advice 🙂

However it’s not all sunshine and roses. There’s some small pushback from hotel chains against wannabe influencers demanding freebies. The discussion in the Hacker News thread about this post has some more interesting comments and is worth reading. There is also some evidence that influencers don’t actually make that much money.

Related to this topic is an interesting YouTube video by Matthias Wandel on the effects that  sponsorship have had on his channel. He has more than 1.3 million subscribers and produces some great original content but, from what he says, sponsorship has had a detrimental effect.

Conclusion

What started out as my rather petulant “Why are those photos getting more likes than mine?” whine has turned into an interesting and educational investigation. Some of the slightly puzzling activity that I’ve seen on my 500px account is now a lot clearer and I’ve become a little more cynical. I now quickly check anybody who now follows me and if their profile looks like this …

Bot alert!

… then they go into temporary quarantine. If they’re still following me after a few days then I’m reasonably confident that they’re not bot generated and I check out their photos and usually follow them back.

Even though I’m sure that there’s reasonable money to be made, I’m not really tempted to set up my own bot recruiter ( despite the fact that it’s probably only an hour’s work and it would run fine as a background task on my Raspberry Pi music server.) The best interaction on blogs and photo sharing sites come from like minded people and if you start off on a dishonest basis then the whole relationship is not likely to go well. Also there’s always that nagging feeling with sponsored posts that the author is just in it for the money and that the product review is not entirely neutral. There’s plenty of good, unbiased content out there and I’ll stick with that thanks.

 

 

Posted in Longform, Photography | 4 Comments

Reflections

Water makes a great subject for photography, mainly because there are virtually an unlimited number of themes that one can explore. These photos are all about reflections, in some the water is still like a mirror, others the slight rippling adds an interesting blur effect.

This is Lac de Retournemer near Xonrupt in the Lorraine region of France. It’s obviously winter because of the snow on the bank and, despite the slight ripples, the bare branches of the overhanging trees can just be seen in the reflections.

It’s not often that you get such clear reflections in a river but the Meuse at this point is deep and slow moving, especially in the summer.

A completely still winter afternoon makes the Altausseer See in Austria seem like a mirror. The town of Altaussee and the mountains behind are reflected almost perfectly in the water. There’s a lot of haze because the previous few days had been very snowy and rainy. I suspect that a shot much earlier in the morning would also have been better so that the sun illuminated the mountains but it was cloudy and overcast at that time of day 🙂

There were just enough ripples on this canal in central Birmingham, UK to distort the image slightly but not so much that shape of the building is lost. I just managed to catch the two people in the top right hand corner to add some extra interest.

Posted in Photography | Leave a comment