At first, it was an acronym:
That was ok for 3 founding institutions. It even worked when Western Michigan University joined - with only a slight amount of strain.
After Michigan Technological University and Oakland University and Lake Superior State University and Eastern Michigan University and ... it didn't work anymore. And the name changed from being an acronym to just a name (and not a bad name either.)
When Merit got a domain name, there were only 3 top level domains - .com, .gov, and .edu. Merit didn't fit clearly into any of them - but since it was a non-profit supporting higher level education, owned by several higher level educational institutions - it got .edu. Later, .net was available - and Merit has merit.net too. But merit.com is someone else.
This is by no means a complete history. Merit has it's own history pages, which are well written, detailed, and precise. This is largely from my own memories of what I saw as an employee there during that time, involved in some of it, marveling at the rest.
When Merit was formed, ARPA (the Advanced Research Projects Agency) was investigating data communications and funding a number of different projects. Thinks like the early Internet (Arpanet). Merit received its initial funding from the National Science Foundation (NSF) and the State of Michigan. Like the Arpanet, Merit developed communications technology - protocols, machines, and software.
Merit didn't last long on its initial grants--just long enough to demonstrate its three node network. But its own technology was promising enough for the three founding institutions to fund it so that researchers at the three institutions could use resources at each of the three institutions.
Merit also benefited from it's close relationship with the University of Michigan, both the College of Engineering and the Computing Center. Collaborating together, Merit and the Computing Center developed similar technologies on similar hardware, using identical tools. (The University of Michigan's MTS academic computer system hosted the tools.) The University of Michigan Computing Center developed the Data Concentrators (and later Remote Data Concentrators), while Merit developed the CPs (Communication Processors, later PCPs - Primary Communication Processors) and SCPs (Secondary Communication Processors.) All of this was on DEC PDP-11 and LSI-11 hardware.
Merit and the Internet
After the ARPAnet got to be more than the plaything of a few wild-haired scientists funded by ARPA (later DARPA), the National Science Foundation got involved and started funding a network to facilitate the scientific research it funded using the technology proven by the ARPAnet. This was NSFnet phase I (1, one). It worked pretty well, but in a sometimes ad-hoc sort of way. There were a lot of smart, dedicated people involved - but they had other things to do too, money was tight (isn't it always), and the NSFnet (and ARPAnet) would get bogged down from time to time.
Incidentally, a large part of NSFnet phase I was run on LSI-11 equipment - the same hardware Merit used for all of it's SCPs and a some of the PCPs.
It had gotten a bit messy, as new things like this will, as people learn from experience. So, the National Science Foundation asked for proposals for a bigger, better, faster, new and improved NSFnet. This would be NSFnet II.
Merit pulled together a remarkable coalition of organizations - the State of Michigan, MCI, IBM, and itself. Merit had operational experience running a network (it's own technology) over an enormous span of space. Michigan is HUGE - when you're driving from Detroit to Marquette. IBM had some fancy technology (routers), and MCI had data connections all over the nation.
Since Merit was proposing that it would manage things - but not use it's own technology - it was viewed as impartial - a good thing given the number of different technologies involved at the time. And Merit's experience, as a large distributed regional network was important too. Many of the other regional network that responded to the NSF's proposal were operated entirely on a hub-and-spoke basis, whereas Merit was truely, physically distributed. And Merit knew how to get along with people on site to get things fixed.
Merit ran the global Internet with MCI and IBM until 1995. Global Internet.
Having one organization run The Internet - doesn't work very well once the internet starts growing up. And a lot of other people wanted to get involved in this (and make money.) Since the National Science Foundation stopped funding the NSFnet, no one organization has had as much control over and insight into the Internet as Merit, IBM, and MCI did.
From "Which Host?" to the early Web
One of the services Merit provided it's member institutions was dial-up service for faculty, students, and staff associated with those institutions. The earliest versions was pre-TCP/IP, pre-PPP. It was a "dumb terminal" solution (i.e.: Teletype-like.) You would have your terminal (or microcomputer) call a phone number, and Merit's equipment would answer with something like:
%Merit:Hermes (H02:CCB3) Which Host?
For a few years, "Which Host?" would change to "Witch Ghost?" on Halloween, but this ceased when it started breaking simple minded scripts on microcomputers written to watch for particular phrases from the dial-in service. Imagine Google not being able to play with their landing page because someone's PC won't work!
This was based on Merit's own PCP/SCP technology. It's hard to maintain your own technology when you have a small staff - but when it was the only way to provide the service, it was necessary. Once businesses were formed to provide this kind of technology, it was inevitable that Merit would have to abandon it's own, although it took a very long time for commercial technology to exceed what Merit was able to do itself using what was - at one point - leftover and obsolete DEC equipment.
Merit implemented SLIP (Serial Line Internet Protocol) and PPP in it's dialup service (and TCP/IP in it's SCP & PCP systems, of course.) Merit was providing dialup TCP/IP begining in the late 1980s on a wide scale - a first.