An upstream incident at Heroku caused our Screen product to be inaccessible for five hours. The length of this outage was due to our needing to route around the vendor to solve the problem; we are now in a much better position to avoid a scenario like this in the future.