#502 1983 · Sequential Circuits, Roland, Yamaha, Korg and Kawai (the MIDI working group) · Electronic musical instruments
Rival synth makers agreed to a standard by deliberately building the cheapest version, not the best one
the problem
Every synthesizer maker's instruments spoke a different, incompatible electronic language
background
By the early 1980s, synthesizer manufacturers each built their own proprietary way for one instrument to control another — Oberheim's system linked only Oberheim gear, and every other maker had its own closed scheme. Musicians who wanted to layer a Roland synth under a Sequential Circuits keyboard, or sync a drum machine from a third company, simply couldn't; each brand's instruments were an island, and the arrival of affordable home computers was about to make that isolation worse, not better.
The straightforward engineering path for the companies drafting a shared protocol was to build the most capable one their 1983 electronics could support: continuous polling for expressive detail, wider data words, room to grow. That was also the path several early proposals took, calling for faster transmission and richer resolution — the obvious way for competing engineers to prove their own instruments' features mattered.
what everyone would do
Build the most technically capable protocol the era's chips could support — richer resolution, continuous status updates, wider data words — since that is what would best serve professional musicians' actual needs, and it's the version every synthesizer engineer's own instincts pointed toward in the early proposals.
what they saw
Smith and the working group saw that a communications standard's value comes entirely from how many machines speak it, not from how much any single machine can say — and a sophisticated protocol that only well-funded manufacturers could afford to implement would guarantee a permanently split, smaller network, no matter how impressive it was on paper.
the move
Dave Smith of Sequential Circuits and Roland's Ikutaro Kakehashi brought rival makers — Roland, Yamaha, Korg, Kawai — into direct collaboration, and the group repeatedly chose the simpler, cheaper option at every design fork: a slow 31.25 kBaud serial line, events sent only when something actually happened rather than continuously polled, and data words capped at 7 bits. The Musical Instrument Digital Interface (MIDI) specification, published in August 1983, was deliberately engineered to be implementable on the cheapest microcontroller chip available, not the most powerful one.
why it works
Capping the spec's cost of implementation to whatever the cheapest available microcontroller could handle meant no manufacturer, however small, had a cost excuse not to adopt it, so essentially the entire industry implemented MIDI in its first year rather than splitting into competing camps the way VHS and Betamax did. Because every company had a real stake in the same shared network from day one, no one had grounds to defect to a rival scheme, and the resulting universal adoption became self-reinforcing: the more instruments that spoke MIDI, the more valuable speaking it became, which is exactly the value a technically richer but narrower standard could never have generated.
the payoff
Every synthesizer maker, including the smallest and least capitalized, could afford to add a MIDI port, so the whole industry adopted it within the same year the spec published rather than splitting into rival camps. Dave Smith later put it plainly: MIDI "works because it's simple and free... which led to it being ubiquitous." Forty years on, the 1983 spec is still the operating standard for electronic instruments worldwide, natively supported in Windows, macOS and Android, and new MIDI gear still talks to 1983-era hardware without modification.
where it breaks
The same minimalism that won universal adoption becomes a permanent ceiling: once an entire industry has settled on a shared minimal spec with no single owner, upgrading it requires re-winning consensus across every manufacturer again, which is why a genuinely necessary technical advance (higher-resolution control, tighter timing) took nearly 40 years to ship as MIDI 2.0, and only by staying backward-compatible with the original. It only works, too, when adoption cost is the actual bottleneck to standardization — if the blocking problem is something else (incompatible business models, patent disputes, regulatory approval), building the cheapest version doesn't remove it.
what came after
The MIDI Manufacturers Association (formed 1985) has maintained the spec ever since, and MIDI 2.0 (2020) was built to stay backward-compatible with the original rather than replace it — an explicit design constraint the group accepted precisely because breaking compatibility would have meant re-winning industry-wide consensus from scratch. Smith and Kakehashi jointly received a Technical Grammy in 2013 for the standard.
references
- [1]MIDI History Chapter 6 — MIDI Begins, 1981-1983The MIDI Association, 2023midi.org
- [2]MIDI daddy Dave Smith: '30 years of version 1.0 shows we got it right'The Register, 2013theregister.com