NB: This blog post has been updated to reflect the current situation and is slightly different from what was posted on our main page earlier this week.
It’s official: Postimage.org is in danger and needs your help. Here’s why:
Who is at fault?
On October 27, 2016, CloudFlare abruptly cut us off from most of their services except DNS for abusing their system. This came as a bit of a surprise, since although we’ve been using one of their cheapest plans for a long time, we had reached an agreement earlier this month that we would be upgrading our account when the next billing cycle started. A couple of Skype calls later we learned the following:
- CloudFlare was very unhappy that the total traffic usage of our project had surpassed a staggering figure of 1.8 petabytes in the last 30 days.
- The amount of money we had to pay monthly to make them happy again grew after each Skype call as more people in CF got involved in examining our case: $200 became $1000, which in turn became $12k.
- The sales team was adamant that although CloudFlare did not officially have bandwidth limits, our violation of Section 10 of their terms of service could not be remedied by serving less image traffic and more HTML traffic (although, being an image hosting company, we have no idea how we would pull this one off anyway without blatantly gaming the system), and that at the level of petabytes of data, they would never allow that on a $200/month Business plan.
- We were officially screwed.
Let us make this absolutely clear: we do not hold a grudge against CloudFlare for refusing to foot our traffic bill any further. We do realize that we are costing them a ton of money, and it is solely our own fault that our current business model is not sustainable. We also recognize that the deal they offered is probably better than anything we could reasonably expect from any other CDN providers. The only thing we disagree with is that instead of publishing estimates of how much traffic customers are actually allowed to consume at each service plan, CloudFlare insists that their bandwidth is unlimited and declines to comment on the actual terms of service.
What should we do now?
A possible outcome is that Postimage.org will have to shut down, terminating nearly 140 million images embedded into some 450 thousand websites, first and foremost a number of great message boards (although a lot of online auctions, personal galleries and corporate websites will be affected as well).
However, thanks to a generous stream of donations it looks like we will make it through the next couple of weeks until we develop more permanents solutions both to cut costs and to implement a viable business model. We didn’t pay enough attention to making money off Postimage, as the advertising revenue was sufficient to support ourselves and our servers. We are now rethinking what we can actually offer to our community while minimizing the harm done.
While we are definitely bothered that the project on which our modest livelihood depends is shutting down, this latter circumstance bothers us much more. We would hopefully find other jobs, but a huge historical layer spanning more than a decade of some of the Internet’s most vibrant communities would be obliterated forever. Thus, at this point failure is not an option; we must fight tooth and nail to keep Postimage.org running and to correct the consequences of our earlier mistakes.
Where’s the money?
Historically, advertising revenue has been our main source of income [approximately a 50/50 split between AdSense and content recommendation systems]. While we’ve recently decided to experiment with header bidding platforms, we have yet to collect a single dollar from these experiments, so we don’t really know if this will work.
We are also considering the option of running a crowdfunding effort a la Reddit Gold or a donation system. Our main website is seeing 8 million unique users per month, and if just 0.125% of our user base sent us $1 every month, that would be enough to cover our bandwidth bills.
Finally, there is an option to try a different role in the digital marketing industry, perhaps even become a DMP data source as well as a publisher (our recent measurements indicate that we’re serving over 28 million unique daily users over our whole network of 450k websites). However, we have to first answer a couple of important questions such as if this data is actually worth anything, and if such a privacy-impairing trade-off would be acceptable for our users if that’s what it took to keep their images online.
We realize that there is a certain probability that neither these nor any other options we have will be able to cover our costs. Still, we intend to fight hard to save this project and all the amazing websites it has become an integral part of. Because some things are just worth fighting for.
P.S. If you have any suggestions or bright ideas, please contact us at firstname.lastname@example.org.