But really it's all about framing. For example on desktop computers it's not possible for people to create new instructions for their CPU to handle. At some layer there will be an API that user needs to use to interact with the device. As times goes on I think it's natural for that layer that users are expected to interact with their device with to become higher level. I believe the top level comment is framing this issue such that current phones don't have an API that matched how it worked for UNIX computers and that is a bad thing. The commenter is too focused on how things worked in the past and doesn't want to allow for things to change.