During a visit to University of Waterloo in 1984 summer, my host, aware of my interest in software for symbolic computation since 1971 and use of such software since 1973 (first IBM PL1-Formac on IBM360/50 computer in Australian National University), mentioned that an open demonstration of Maple would occur during one hour in a particular location. When I arrived, I saw 48 terminals operating off a DEC VAX11/780 computer, with 47 occupants of the associated chairs. I sat at the remaining terminal and tried some tests with Maple 4.2 (?). I had previously undertaken some major symbolic calculations with Reduce 2 on a DEC KL10 computer, that I was constrained to use between midnight and 7 a.m. because, when I was running Reduce, all other users of that not insignificant computer had the impression that it had ceased to operate. In contrast, there were I, and 47 other users of Maple on a less powerful computer, all (supposedly) happily trying various calculations, each likely thinking that he had the computer all to himself.
The credit for this achievement must be partitioned between the VAX VMS operating system, which provided a much improved 'architecture' with virtual memory for multiple users than the KL10 computer, and the designers of Maple, who developed a small kernel that operated efficiently in the fraction of the total memory allocated to each user.
When Maple in recent versions with the 'standard' (or Java) interface takes so long to open, and sometimes also to close, a session, I recall warmly that rapid response. Now that that standard interface is so large and bloated, with no compensating property that could not be arranged otherwise, there is a real need to return to the roots of Maple, and so to provide a worthy successor to the 'classic' interface.