Time to close some doors

In a recent standards tcon I was struck by the statement:

Don't close the door to innovative solutions.

I think it is long past time to start closing that door.  Standards are not research.  The successful standards are created when a group of people realize that:

• They agree that there is a common problem (aside from minor details)
• There are multiple well understood solutions to that problem
• There needs to be agreement on one common solution for progress to be made.

Standards should be created after the innovation has occurred and after there is experience with that innovation.  Experience shows that there is a long period of experimentation and experience gathering between the innovation and the successful standard.  If there is still substantial innovation needed, then standardization is premature.

Fortran

This was a highly successful standard.  There were several key dates:

1958 - The successful introduction of Fortran II into operational use.  It was one of multiple, incompatible but similar languages called Fortran.

1966 - The publication of the Fortran 66 standard, more commonly known as "Fortran IV".

That's eight years between the earliest operational use of the one of the versions and the finished standard.

Ethernet

Another successful standard.  It's key dates are:

1968 - Publication of the Alohanet paper, describing the technique and successful experiments.

1973 - Initial operational use of the major contributor to the eventual ethernet standard.

1982 - Completion of the 802.3 standard for 10 Mbit/sec ethernet.

In this case it was nine years from initial operational use, and fourteen years from the initial idea.

TCP/IP

Another successful standard.  It's key dates are:

1973 - Publication of the Catenet paper, describing the internetworking technique and early experiments.

1975 - Initial operational use of the earliest internetworking protocol.

1982 - Publication and transition to TCP/IP, as a finished standard.

This went faster.  It was only nine years from initial idea, and seven years from initial operational use to the point where a standard was ready.

All of these standards continued to evolve and grow.  But the initial innovation was many years before the standardization.  The interim was spent with research, experiments, evaluations, and improvements.  The standards efforts were debates between advocates with experience and experimental results to justify their claims.  The implications of compromises and decisions could be understood.

None of these standards was perfect or complete.  Fortran continued to evolve, Ethernet has grown into a variety of faster versions, and TCP/IP continues to change.  The innovations have been gradual improvements.