ST4 is older, and I think developed without thinking that we'd have a separate computer controlling the mount, the ST4 messages would be generated by the guide camera itself which would do the calculations and calibrations etc...
Guide Cameras these days don't have that ability, and the quickest way to ask the mount to do something is to ask it directly, not ask the camera to ask the mount to move.
ST4 is even more ancient than that. It has its roots in the old hand controllers used for manually guiding. In those days you would watch a guide star against the cross hair reticle and push the N/S/E/W buttons on the hand control to correct. ST4 autoguiding replaces the manual button pressing which is why it has 4 wires plus a ground - one wire for each of the N/S/E/W buttons on a hand controller. So a 100ms pulse is the same as holding down a button for 100ms. You could still guide manually via the ST4 port with a set of 4 push buttons if you wanted to. Each button grounds the relevant wire which causes the mount to move in the appropriate direction while the wire is grounded.
Pulse guiding uses two way communication between the mount and the autoguider. That way if something fails in the mount is is usually detected pretty quickly. With ST4 there is no feedback from the mount. Also, the ST4 cables are notoriously unreliable. If one wire fails, especially if intermittently, it is difficult to detect what is going wrong.
Hi
ST4 was the name given to a box manufactured and marketed by SBIG in or around 1990. It automated some guiding processes.
These days, the box has been replaced by electronics inside guide cameras which do the same thing.
If you've a choice, go with direct mount-computer connection. The main advantage for us being that as the software knows where the telescope is pointing, you need only calibrate once and re-use that same calibration; no need for aux-mounts or extra cables.
Cheers and HTH