Hacker News Reader

Comment by BarkingOven | Hacker News Reader

I'll tell you exactly why it's nine minutes. Because in the olden days where everything was made with discrete components, the math of 9 minutes is WAY cheaper.

Let's say it's 6:00, and you hit the snooze button. Save the current value of the minute minus 1, into a comparitor, Trigger the alarm when the comparitor goes off automatically.

If you save the current value into the comparitor, the alarm triggers immediately. Now you need to add delay logic. Just use current minus 1, since it's only a few half-adders.

You can't use the high order minute digit because then your snooze length will depend on your current low order minute digit. Any logic to compensate starts adding a multiple of the number of components required.

View on HN

"In the old days when everything was made with discrete components" and you talk about "comparitor" (comparator) and "minus one"

No, this is too advanced.

For that you would be looking at 4 bits storage + comparator + the "minus one" circuit (which is very complicated) for a BCD counter.

You don't need the "minus one" circuit. Just always store the previous minute value whenever the minute is incremented, and then latch that value when the alarm goes off.

Yes, of course. You are correct. Saves even more logic real estate. :-)

It was established in the age of mechanical clocks. Because of the way the gears were generally configured, the two sane options were 9 minutes or a bit more than 10. They chose 9. Digital clocks followed suit.

Thank you, I would bet on this. Examining an early digital clock with 9-minute snooze could probably confirm that the 9 minutes is implemented using some smaller number of components than any 10-minute system could be.

I also vaguely recall that such on such clocks, the alarm itself (if not silenced) would tend to run for 9 minutes before self-silencing. The same logic might apply: it's the longest finite interval that doesn't require some sort of multi-digit carry-logic.

I wouldn't be surprised if early circuit layouts even made it so that the wiring to toggle some [digit ± 1] triggering state was shortest/cheapest, in a circular arrangement, compared to any other potential offset. (I don't even think there was anything binary in the earliest digital-display clocks, rather some [10m, 6h, 24h] or even [10m, 6h, 12h, 2meridiem] cyclical counters.)

In the olden days, our clocks were analogue/mechanical. Even the ones showing digits.

Your argument makes most sense around here :)

This rings true.

But are clocks using a base-10 counting system? I'm thinking no.

The oldest digital clock my family had used a mechanical display, so it had ten values for the right-most digit, six for the next, and twelve for the left-most. So this explanation would definitely make sense for it.

But I don't recall if it actually had a snooze button...

Usually BCD (http://en.wikipedia.org/wiki/Binary-coded_decimal)

Reply on news.ycombinator.com