• Everything about Commander is web based, but Insight is a desktop app. Who wants to drop into a desktop app just for a quick visualization of a build process?
• A significant portion of Insight is directed towards analyzing serializations and conflicts in builds, but Commander has no analogous concepts (if there’s a serialization in your Commander job, it’s because you put it there!)
• Commander has the concept of steps-within-steps (for example, subprocedure calls and broadcast steps); Accelerator (and thus Insight) has no similar concept. You could argue that this is something like submakes, but you’d quickly find that the attributes of a step that calls a subprocedure just don’t map well to the attributes of a submake.
• Insight is necessarily tightly coupled to Accelerator, for performance reasons. Remember, Insight is designed to handle analysis on builds with literally millions of distinct jobs. Abstracting the implementation to allow it to handle Commander concepts would hamper our ability to handle those large Accelerator builds efficiently.
• Last but not least, the domain vocabularies don’t even match up. Accelerator has “makes” and “jobs”; Commander has “jobs” and “steps” and “procedures”. Although it is certainly possible to factor this detail out of the interface, that seems like a great deal of work to effectively make a square peg fit into a round hole.

Thanks for the question!