Scriptlets have two advantages over ordinary ActiveX controls. The first one is their innate structural simplicity. The second point is a little more subtle. The HTML language is a Web standard worldwide. DHTML is supposed to replace it by the end of 1998, when the W3C consortium will hopefully approve that standard.
From then on, DHTML (maybe with slight differences from the current version supported by IE4) will become a new worldwide standard and all the browsers will agree and align to it. Having the DHTML object model available will quickly increase the demand for reusability. And scriptlets will be there, already up and running.
While this isn't a current scenario, it is a reasonable prediction. In this way, scriptlets may well become really cross-platform and cross-browser components, maybe at the expense of ActiveX.
Microsoft is working hard on transferring COM/DCOM over to non-Windows platforms, such as Unix. This will take a lot of effort, and will be very complex. On the other hand, programmers need interoperable, integrated, and possibly cross-platform components. Above all, they need them today, not in the remote future. Furthermore, these components should be portable and exploitable across the Internet, through the widest possible range of browsers. Specific technologies like ActiveX or JavaBeans all have pros and cons. Moreover, be practical. Are you really sure that there will ever come a time when Microsoft, Sun, Netscape, and other vendors, will agree on a unique and general standard for components? We don't think so.
Instead, HTML (followed by DHTML) is the standard. And the scriptlet proposal offers reusability and componentware inside the DHTML standard. Do you have a better idea?