Improved player and video delivery infrastructure
We have decoupled our player from the view.ly app and created an improved player build pipeline. The new player loads faster, consumes less bandwidth and allows us to ship embedding feature sometime in the near future. We have also fixed several bugs and UX issues.
The improved video delivery infrastructure and routing have contributed to faster buffering and seek times.
Dynamic gas fees
Spam attacks on the Ethereum network in the past month have exposed the insufficiency of our previous gas estimation mechanism, leaving users with poor experience during 10-fold spikes.
We have tuned our gas estimation algorithm such that it responds to congestion spikes better. The primary objective is for users to have fast confirmation times, while paying the least amount of fees, and our new gas estimation mechanism seems to be performing well in that regard.
Privacy respecting analytics
view.ly has no third party trackers installed. This is great, because users can watch any content they like without being spied on. It does however also present a challenge, as we are flying in the dark and have no objective tools at our disposal to make the service better.
This is why we are exploring a middle-ground, where we collect anonymous usage data (anonymous meaning that no personally identifiable information such as IP, cookies or browser fingerprints are collected or stored, and the metrics collected are not linked to user accounts). We would like to know how many people watch what, but remain willfully blind to who watches what.
Currently we only record number and duration of views, as well as traffic origin (hint: it’s mostly organic discovery from Google). This will allow us to improve the service by allocating more nodes to more popular videos, as well as offer very basic statistics about videos to their respective creators.
At this point I would like to re-iterate that our mission at view.ly is to find an alternative to the ads based internet, and thus have no need or incentive to collect personal information about our users.
If we ever have a technical need to learn more about our users to provide additional features (ie: a recommendation engine), it should be highly transparent about what data exactly is being collected and only enabled by users’s opt-in.
Onwards to Alpha 0.5
In 0.4 we have generally made existing things better. Viewly Alpha is more reliable and has less rough edges. We have also received a fair amount of user feedback, which is very important to us. The next point release will bring forth a major new feature to address the most common pain point so far.