The general name for this is mechanical sympathy. Humans have advanced sensory processing abilities, and if we design systems that use them, the performance of the humans operating the system improves.
It improves for the simple reason that richer feedback leads to a better understanding of the state of the system, which in turn lets the operator predict the effects of their actions more accurately.
I thinkn we could be a lot better at using human sound, visuals, and spatial navigation abilities in understanding software systems too, but I've never gotten around to implementing anything despite plenty of ideas.
My latest experiment is using tones or speech synthesis articfacts to get insight into LLM operation -- ongoing resoning, tool calls, etc. Still haven't really dialed it in (hah!) to be useful yet.
As well as all the other good reasons: modems had speakers because they COULD have speakers, because modems were connected to a network which was designed to safely transmit human-audible analog sound waveforms, which is exactly what cheap speakers are designed to emit.
If we directly connect a speaker to a different data network format, such as Ethernet, we probably will hear nothing useful. Maybe some clicks and pops, or the smell of a blown speaker.
I used the modem speaker for all kinds of things in the BBS days. To name a few:
- Monitoring to minimize DTMF duration (S11 register, IIRC) and maximize redialing speed[1]
- Doing the same thing with dialtone and busy tone detection times
- Recognizing BBS callers by the unique sound of their modems, and using this to be annoying
- Surprising other BBS sysops who were themselves trying to call out, by issuing an ATA and dropping them into my BBS's login screen when I heard dialing instead of busy, ringing, or carrier tones
- Wardialing (I gave up on that forever when the voice from the machine said "911, what is your emergency?" and the usual automated hang-up-and-try-different-random-number routine did not disconnect the call, and instead produced an increasingly-agitated 911 operator)
[1]: A firmware update to the SupraFAXModem v.32bis allowed this timing to be so tight that the phone switch would miss digits, so tuning was in order
I too tuned my dial tones as short as the exchange would allow. BBSes used to give us busy signals all the time, so modems would redial on a busy signal. It would dial and hang up silly fast, then redial. (The delays by default were not super short between redials, but again, just a value to tune)
A major commercial radio station used to have give aways of music- they’d say they will be “opening the lines soon, first 20 callers win!” My modem was very good at reliably getting me in the first 20 :)
Of course that wouldn’t work without me sitting next to it and a speaker in it!
When that high pitch squeal emanated from the modem speaker you’d run across the room and start typing at the BBS because if you didn’t, it would timeout and hang up on you.
Also, if there were issues connecting, you could get a sense from the sounds. Sometimes it told you there were hardware issues afoot and you’d be better off taking a BBS out of the dial rotation for the evening.
(You could redial a group of BBS’ to increase your odds of connecting, as opposed to redialing a singular BBS continually…)
Connection protocol codependency: the human follows along to ensure the connection is good because goodness help us you can’t trust the modem to figure it out and get it right every single time alone.
I wasn't really allowed to use the internet as a teen because it cost money. My parents would never pay for a flat subscription. Luckily I figured out you can disable the speaker in the driver settings, at least you could with ours. Then I preceded to spend half of my teenage life awake at night on the internet after everyone else had gone to bed. This screwed up my sleeping patterns for years.
Sound let you monitor progress and success/failure. It also provided some security against a program phoning home on its own (as long as the speaker was on).
And not just binary success or failure - you also got to know the handshake sound and could predict if it was successfully connecting at high speeds or not - and so quickly hang up and try again if it was going to be slow.
Once I dialed the number of my grandma with my modem and my uncle answered the phone. So it also guards for when you inserted the wrong number and some person is actually on the line, and not a computer
It improves for the simple reason that richer feedback leads to a better understanding of the state of the system, which in turn lets the operator predict the effects of their actions more accurately.
I thinkn we could be a lot better at using human sound, visuals, and spatial navigation abilities in understanding software systems too, but I've never gotten around to implementing anything despite plenty of ideas.
My latest experiment is using tones or speech synthesis articfacts to get insight into LLM operation -- ongoing resoning, tool calls, etc. Still haven't really dialed it in (hah!) to be useful yet.
If we directly connect a speaker to a different data network format, such as Ethernet, we probably will hear nothing useful. Maybe some clicks and pops, or the smell of a blown speaker.
A major commercial radio station used to have give aways of music- they’d say they will be “opening the lines soon, first 20 callers win!” My modem was very good at reliably getting me in the first 20 :)
Of course that wouldn’t work without me sitting next to it and a speaker in it!
As a parent now, I wish my kids devices had such a feature.
Also, if there were issues connecting, you could get a sense from the sounds. Sometimes it told you there were hardware issues afoot and you’d be better off taking a BBS out of the dial rotation for the evening.
(You could redial a group of BBS’ to increase your odds of connecting, as opposed to redialing a singular BBS continually…)
I miss that era. Now if a mlcc can be cut to save a fraction of cent, it sure will be.
If you're going to be posting old things, add dates to them please