"What has been will be again, what has been done will be done again; there is nothing new under the sun." – Ecclesiastes
"Time is a flat circle." – Rust Cohle
Remember 2010? It was a good year. I got married that year. The iPad launched. My friends got me one as a wedding gift. I brought it on the honeymoon. My wife wasn't thrilled.
Focussion also launched in 2010. You probably don't know Focussion, because from what I can tell it never had more than a few thousands users. I didn't know about it either until some late-night Googling a couple days back. I found a lot of products in the penumbra of "better feedback on photography" but nothing spot-on. Yet, here was Focussion with its big, boasting headline splashed across the screen:
"Improve your photography skills, get constructive feedback for real. Join the community and get critiques that's nothing like 'Great photo' or 'Wow!'"
Wow, indeed. Questionable syntax aside, that's it. These are my people! Why hadn't I heard about this?
Something wasn't quite right—and I sensed it immediately. The stark Bootstrap 1.0 aesthetic. The empty section where the "Trending" photos should have been. I clicked through to the top post in "Latest Forum Activity". It was from March 13, 2014. I clicked the second post, dated July 23. No year listed. The cryptic title of the post was "Kolkata the core expertise of Packing and shifting". I thought maybe Kolkata was a brand of camera I hadn't heard of. It's not. It's the name of a packing and moving company in India, spamming the forum for Google juice.
My suspicions were correct. It was a digital ghost town. As I poked around some more, it seemed the last residents had vacated sometime in 2014, roughly four years after its founding. The C-H-E-A-P VIAGRA™ tumbleweeds blew through the comment sections that once seemed to be a flourishing digital society.
I wondered: Who were these people? What happened here? What could I learn from them? I was going to find out.
The first thing I did was reach out to the founders via the contact form on the site. I figured it was a long shot. They never responded.
A "spider" is a program that "crawls" the "web" and records what it sees. It sucks data from web pages and follows links to other pages. It's what Google uses to feed its massive internet brain.
Thanks to the internet's open infrastructure, anyone can run these spiders. From the perspective of the website they're crawling, they look mostly like a human user would. The trick is training them to look at the right things.
And I had a mounting list of questions I wanted answered. Like, how many users did this site have? How often did they post? What was the kind of feedback people would give? Who were the power users?
Then I stumbled onto Portia. Not Ellen DeGeneres' wife. The popular web scraping program named after the Portia spider, a genus remarkable for its intelligent hunting behavior.
With Portia, you pick any web page and annotate it, giving labels to the parts of the page you care about. These are the parts the spiders will look at with their laser eyes and send back to the hive-mind, so you can retrieve the aggregated results later. It's remarkably easy. In fact, if you're interested, let me know and I'll do a post on all the technical details.
There were two types of pages I was most interested in on Focussion. The first was the photo pages, where a user-submitted photo lived, along with all the feedback from other users. The second was the user profile pages, which listed some basic info about the user, along with a summary of their activity on the site.
I unleashed my spiders on the Focussion site before I left for work one morning. By that evening, they had exhausted their search. They had crawled every page they could find.
The data they collect comes back as a big JSON file (aka: a fancy text file). I wrote a few custom scripts to clean up the data, and then dumped the results into this Google document for analysis.
My little spiders found 7,106 user profiles and 24,175 pieces of feedback posted.
Very few submissions received no feedback. Almost all submissions received less than 10 feedback comments, with the most frequently occurring number being 2.
This graph paints a relatively rosy picture of what life in Focussion looked like. Relatively few people went hungry (for feedback).
What about the velocity of life in Focussion? Was life there more Wall Street trading floor or retirement home? I looked at days until first feedback (Fig. 2), which shows that the overwhelming majority of posts received feedback on the same day.
I wondered how good Focussion was at getting people to share, versus lurking awkwardly towards the back (or signing up and never doing anything at all). Many community sites show a lot of in-balance here, and Focussion seems to follow the trend, with a sharp fall-off in the number of photos submitted starting from just one submission. Although, more people submitted one photo than zero.
I noticed a curious little uptick there to the far right. It was a major outlier. But more on that later.
Graphs are cool, but I wondered if I could discover anything meaningful by looking at the actual words people were posting as feedback, so I analyzed 837,403 words that the spiders found in the feedback comments.
Positive words dominate the content of the feedback. Nice (1.5% of all words), love (0.9%), good (0.7%) and great (0.7%) were all in the top 5 most frequently used words (after throwing out filler words). Distracting (0.09%) was a top negative word, and was used 17 times less frequently than the top positive word. Boring (0.005%) came in way down the list at number 1480, and ugly (0.002%) at 2847.
When I looked at photography words, color(s) (0.66%) beat out composition (0.52%) as the top most used, then light (0.3%), focus (0.3%), crop (0.2%) and contrast (0.2%), with exposure (0.1%) way down the list. As for actual colors, white (0.15%) and black (0.16%) were pretty equal, followed by blue (0.1%) then green (0.06%) then red (0.05%).
As for popular subjects, eye(s) (0.3%) was a popular topic, along with sky (0.2%) and water (0.2%) . Background (0.3%) was five times more popular than foreground (0.06%).
I'm happy with the results of my little archaeological expedition. But I wanted to go deeper. Remember that little uptick at the end of Fig. 3? That uptick was a someone. A Focussion power user. I dug through the data. His name was Sean Allen. More soon.
Raw data is available in Google Sheet here.