I know, which is why I mentioned Spectrum as the mistake, not ZX-80 and ZX-81.
The bad keyboards did undoubtably impact sales of the ZX-80 and ZX-81 too, as it was facing competition from "cheap enough" machines with proper keyboards pretty quickly, but it's hard to say it was a mistake to build them that way when they were released.
The Spectrum is a different story, which is why I called it out as a mistake. Sinclair continued their tradition of crappy keyboards with that as well, and by then that was a deal-breaker for a large part of the market. Their machines were - together with the Oric 1 - often derisively referred to as "doorstops" even in the press, even though the Oric had a slightly better keyboard.
By the time the Spectrum+ came out, they (and Oric...) had fixed that mistake and it was clear what a difference it could be: The new model outsold the old 48K model (which was identical apart from the case) 2:1.
But the Spectrum+ had atrocious failure rates (the Spectrum had failure rates of 5%-6% - there are reports of failure rates of 30% for the Spectrum+). As a result I don't think I even ever saw one of them. Many dealers internationally never carried them.
When the Spectrum 128 came out, it was too late. I remember my friends and I were impressed with it - it was great "for a Spectrum", but by then "everyone" had already picked camps, and Spectrum was largely dead in the water outside of the UK and Ireland. Not even the subsequent +2 and +3 after Amstrad acquired the Spectrum could make up for that - by then the C64 was too cemented as the dominant 8-bit machine pretty much everywhere but the US and UK, only the UK of which the Spectrum had an established presence of note.
Commodore on their hand made their second attempt at a crappy keyboard with the Commodore MAX Machine (Ultimax/VC-10) in 1982, as a cost cutting measure.
That's what I was referring to by repeating mistakes - the MAX Machine was quickly cancelled after poor sales.
Commodore knew better, but tried again anyway, but because they always had a ton of different projects and different models, and different models were released in different markets, they saw the failure in Japan, and never invested much in getting it sold elsewhere, and cancelled it quickly to focus on the C-64. So unlike Sinclair, a failing model (or three) didn't affect them all that much - they had other models that sold well in most markets.
There's a long range of variations of the C16/C264/Plus4 range, for example, that were pretty much ignored by many Commodore subsidiaries, because they saw them underperform in other markets and didn't really want them.
Sinclair on the other hand didn't have better alternatives until it was too late to get a bigger international foothold.
The bad keyboards did undoubtably impact sales of the ZX-80 and ZX-81 too, as it was facing competition from "cheap enough" machines with proper keyboards pretty quickly, but it's hard to say it was a mistake to build them that way when they were released.
The Spectrum is a different story, which is why I called it out as a mistake. Sinclair continued their tradition of crappy keyboards with that as well, and by then that was a deal-breaker for a large part of the market. Their machines were - together with the Oric 1 - often derisively referred to as "doorstops" even in the press, even though the Oric had a slightly better keyboard.
By the time the Spectrum+ came out, they (and Oric...) had fixed that mistake and it was clear what a difference it could be: The new model outsold the old 48K model (which was identical apart from the case) 2:1.
But the Spectrum+ had atrocious failure rates (the Spectrum had failure rates of 5%-6% - there are reports of failure rates of 30% for the Spectrum+). As a result I don't think I even ever saw one of them. Many dealers internationally never carried them.
When the Spectrum 128 came out, it was too late. I remember my friends and I were impressed with it - it was great "for a Spectrum", but by then "everyone" had already picked camps, and Spectrum was largely dead in the water outside of the UK and Ireland. Not even the subsequent +2 and +3 after Amstrad acquired the Spectrum could make up for that - by then the C64 was too cemented as the dominant 8-bit machine pretty much everywhere but the US and UK, only the UK of which the Spectrum had an established presence of note.
Commodore on their hand made their second attempt at a crappy keyboard with the Commodore MAX Machine (Ultimax/VC-10) in 1982, as a cost cutting measure.
That's what I was referring to by repeating mistakes - the MAX Machine was quickly cancelled after poor sales.
Commodore knew better, but tried again anyway, but because they always had a ton of different projects and different models, and different models were released in different markets, they saw the failure in Japan, and never invested much in getting it sold elsewhere, and cancelled it quickly to focus on the C-64. So unlike Sinclair, a failing model (or three) didn't affect them all that much - they had other models that sold well in most markets.
There's a long range of variations of the C16/C264/Plus4 range, for example, that were pretty much ignored by many Commodore subsidiaries, because they saw them underperform in other markets and didn't really want them.
Sinclair on the other hand didn't have better alternatives until it was too late to get a bigger international foothold.