It's generally "how do I dial?", followed by them trying to press the holes in the dial, and, when told to rotate it, they rotate it to dial before picking up the handset.
Since you didn't modify the original phone, I wonder if the oldschool dialing way of digit-banging would work. For each digit N, you hit the hangup button N times, pause, then go to next digit. Used to do that on our family's Pupin that looks like this one [1]
I remember it being N at least in ex-Yu. But now that I think about it I forget if it was a reverse-click i.e. the pulse was created by releasing the pressure plate briefly. In that case we'd have a fencepost thing and there would be N+1 physical presses maybe
> Since I didn’t want to make any permanent changes to the phone, I didn’t want to remove these tabs, or to solder anything onto them. I just wanted to connect a cable to them in the easiest way possible.
Same here but I want to go for a xilink bluetooth adapter and maybe even sacrifice and old smartphone with a separate sim. Not sure if the AI stuff will work but it seems possible….
Interesting - summer 2024 I picked-up an old rotary, my plan is to make it ring when a "insert-corporate-instant-messaging/voip/meeting" application call comes in... But time, no time...
Neat! On a mostly serious note, I went looking to buy a handset that just plugs in via USB and is a normal speaker/microphone, and I was quite surprised there wasn't anything like that out there, at least not that I could find.
Am I the only one that wants something like this? Does anybody know where to get one?
Someone please make a Moto Razr form factor and snap bluetooth device so I can keep my big and costly device in my bag and use it only when I actually need to.
If you use such a thing on a regular kind of meeting that happens over zoom or similar, your arm will atrophy and fall off from having to hold that thing for the duration.
The only remedy I see is to give everyone such a contraption and make it mandatory.
Back in the day when phones of this style were the only ones available we learned to hold the handset between our shoulder and our ear, leaving both hands free for other things.
That was one of the things they considered when they designed phones. A lot of money was spent on ergonomics getting this "right". Of course we now know a lot more about ergonomics today, but they did the best they could and did a great job considering.
NB: Many of these older phones already came with metal weights insides to give a feeling of more substance to them. Not always in the handset, though it wouldn’t surprise me if some did have them.
Usually, the base of the phone itself was a sturdy metal plate. So while not the handset itself, the phone unit was usually a pretty decent self defense weapon candidate as was widely displayed in many a fight scene in movies from the time.
> Half an hour and fifty dollars later, I realized I had spent fifty dollars on this, and that this was not sustainable because, if anything, the code was getting more and more buggy the more Claude fixed it.
Off topic, this has been my experience with AI so often that it prevents me from exploring AI uses more.
I liked Cursor’s “auto” plan but that now seems gone. I’d happily switch to a provider that offers a similarly “unlimited” usage.
It's difficult to offer unlimited usage of something that's so expensive to run. OP could have used a $20/month Claude or Cursor plan for "unlimited" usage within their quota had they been willing to use a different model than the $75/Million Token Opus 4
It is possible to radically increase your chances of success. You have to speak the LLM’s language, just like you write Java or Rust. But it doesn’t come with a language spec, so you get to figure it out by trial and error. And a model change means revisiting what works.
Lots of tips on how to do this out there but one thing I do is have it try, throw away everything it it did, and try again with a completely restated question based on the good bits in what it was able to produce.
E.g., if you ask for a web app that does X and it produces a working web app that doesn’t do X, throw that away and just ask for the web app scaffolding. You’ve still come out ahead even if you take over fully.
> Lots of tips on how to do this out there but one thing I do is have it try, throw away everything it it did, and try again with a completely restated question
This is the thing that worries me about AI/LLMs and how people "profess they're actually really useful when you use them right": the cliff to figuring out if they're useful is vertical.
"You’ve still come out ahead even if you take over fully."
I just finished a weeklong saga of redoing a bunch of Claude's work because instead of figuring out how to properly codegen some files it just manually updated them and ignored a bunch of failing tests.
With another human I can ask, "Hey, wtf were you thinking when you did [x]?" and peer into their mind-state. With Claude, it's time to stir the linear algebra again. How can I tell when I'm near a local or global maxima when all the prevailing advice is "I dunno man, just `git reset --hard origin/master` and start again but like, with different words I guess."
We have studies that show people feel like they're more productive using AI while they're actually getting less done [1], and "throw away everything it did and try again" based on :sparkle: vibes :sparkle: is the state of the art on how to "actually" use this stuff, I just feel more and more skeptical.
For an update, Codex has been massively better for me. I should probably give it another go, especially now with Codex Max, as it has been much better at not getting stuck like this. Give it a go if you get the chance, although with the general caveat that people have massively different experiences with LLMs, for some reason.
I think maybe it has to do something with the prompting style, my hypothesis is that some people's prompting styles fit certain LLMs better. I don't know how else to explain the fact that my very experienced friends prefer Sonnet to Codex, for example, whereas I had the opposite experience.
https://petrovs.info/post/2023-01-12-shaiba/ This is my USB rotary dial. It's always fun with the young people when I bring it to IT conferences.
I bet the reaction was similar to what my teenage kids had when I showed them some VHS tapes I found in storage.
What was their reaction like, for those of us who don't have teenagers at home to repeat the experiment on?
It's generally "how do I dial?", followed by them trying to press the holes in the dial, and, when told to rotate it, they rotate it to dial before picking up the handset.
It might have been similar to your reaction when you saw (depending on your age):
a floppy disk (3.5")
a floppy disk (5.25")
a floppy disk (8")
The feeling of the hang up is so satisfying. An important piece of UX.
Plus the sound through the handset sounds like nostalgia!
I enjoy how this article shows the issues with both image generation models and language models
Since you didn't modify the original phone, I wonder if the oldschool dialing way of digit-banging would work. For each digit N, you hit the hangup button N times, pause, then go to next digit. Used to do that on our family's Pupin that looks like this one [1]
[1] https://in.pinterest.com/pin/rare-vintage-pupin-bakelite-rot...
>For each digit N, you hit the hangup button N times, pause, then go to next digit
N+1 times for each digit N. The digit 1 was two clicks of the hookswitch, and the digit 0 was eleven.
Was it? I've dialed like that a few times and I always clicked N times.
Here's the code for the phone, BTW, as I forgot to include it in the article:
https://github.com/skorokithakis/dialogue/blob/master/src/ma...
I remember it being N at least in ex-Yu. But now that I think about it I forget if it was a reverse-click i.e. the pulse was created by releasing the pressure plate briefly. In that case we'd have a fencepost thing and there would be N+1 physical presses maybe
No, the hangup rocker goes to another GPIO than the dialing rocker in the rework.
I enjoyed the llm generated rotary phone image which had no “3” position but two “8”’s
The author did too:
> Rest assured, though, the image is almost exactly what the phone looks like, except with a bit more 8 and a bit less 3.
Relevant Seinfeld bit: https://www.youtube.com/watch?v=uvTYSnIxb-k&t=54s
> Since I didn’t want to make any permanent changes to the phone, I didn’t want to remove these tabs, or to solder anything onto them. I just wanted to connect a cable to them in the easiest way possible.
Same here but I want to go for a xilink bluetooth adapter and maybe even sacrifice and old smartphone with a separate sim. Not sure if the AI stuff will work but it seems possible….
It's a good approach! I used an xlink for my super hacky (but very easy) bluetooth rotary phone project: https://hobinjk.github.io//hardware/bluetooth/hackery/2024/0...
I've got you: https://www.stavros.io/posts/irotary-saga/
Interesting how 10 years later we saw an actual cellphone with this color...
Did we? Which one?
I had a Blu Win HD in basically that color: https://www.meetgadget.com/gadget/71500/BLU+Win+HD/photos/im...
Although, that's contemporaneous with your post.
Interesting, I didn't know of this!
there is accually a marketing guy (founded competeitor to redbull in germany) who mad this comercially.
https://herbworld.com/en/products/retro-telephone-red
I like this guy's optimism of letting AI create a USB sound card out of a microcontroller from scratch.
Probably not a bad litmus test for current and future generations of LLMs though, I'd be curious to try it out on the latest crop.
Only so long as nobody knows you are using it. Soon as anything becomes a metric people with game it.
It got really close! Audio played, it was just crackly/poppy. The microphone didn't work, though.
Interesting - summer 2024 I picked-up an old rotary, my plan is to make it ring when a "insert-corporate-instant-messaging/voip/meeting" application call comes in... But time, no time...
There’s almost always time, as in wall clock time. It’s usually energy and attention that is in limited supply.
I like to say that everyone has the same 24 hours a day, it's a matter of priorities.
Very cool but I think my wireless bluetooth one is even cooler ;)
https://blog.waleson.com/2024/10/bakelite-to-future-1950s-ro...
It actually supports using the rotary dial to call phone numbers on your smartphone.
I think my full-blown mobile rotary phone is EVEN cooler:
https://www.stavros.io/posts/irotary-saga/
It actually makes calls itself and has a SIM.
ok you win
Someone should make a Bluetooth box that outputs to a standard phone connector so any standard phone would work. Has anyone done that?
Someone does - the Cell2Jack is only $36 USD:
https://www.cell2jack.com/
I use one with a 1970's vintage rotary desk phone and it works well.
I think I've seen it, I was thinking of making a USB one. I might!
love the hangup part. so, i can finally "rage quit" the meeting that discusses 99th revision of a doc where a comma should be added here or there.
You can quit anyway, just say "I've given you write permission, I trust you"
pkill -9 -f '(chrome|firefox)' is my rage quit, it's like hanging up but hitting the keys on the keyboard as hard as you'd throw the phone down
Neat! On a mostly serious note, I went looking to buy a handset that just plugs in via USB and is a normal speaker/microphone, and I was quite surprised there wasn't anything like that out there, at least not that I could find.
Am I the only one that wants something like this? Does anybody know where to get one?
Someone please make a Moto Razr form factor and snap bluetooth device so I can keep my big and costly device in my bag and use it only when I actually need to.
If you use such a thing on a regular kind of meeting that happens over zoom or similar, your arm will atrophy and fall off from having to hold that thing for the duration.
The only remedy I see is to give everyone such a contraption and make it mandatory.
Back in the day when phones of this style were the only ones available we learned to hold the handset between our shoulder and our ear, leaving both hands free for other things.
That was one of the things they considered when they designed phones. A lot of money was spent on ergonomics getting this "right". Of course we now know a lot more about ergonomics today, but they did the best they could and did a great job considering.
That’s why you’d get the big foamy cushion you could stick on the back so you could rest it on your shoulder
Cue neck pain
https://www.etsy.com/listing/631698424/vintage-rest-a-phone-...
There were countless products available when I was a kid that were meant to make holding a handset between your ear and shoulder comfortable.
These young people don't know what the nineties were like.
They must think the little things people stick on the back of their phones meant to make holding the device easier for longer periods such a novelty.
They make those shoulder-rest addons for phone handsets. Not too bad to use for extended periods with one of those, with both hands free.
This would be an excellent way to make sure the standup stays terse, we could even put lead weights in the handsets to further increase this virtue.
NB: Many of these older phones already came with metal weights insides to give a feeling of more substance to them. Not always in the handset, though it wouldn’t surprise me if some did have them.
Usually, the base of the phone itself was a sturdy metal plate. So while not the handset itself, the phone unit was usually a pretty decent self defense weapon candidate as was widely displayed in many a fight scene in movies from the time.
> Half an hour and fifty dollars later, I realized I had spent fifty dollars on this, and that this was not sustainable because, if anything, the code was getting more and more buggy the more Claude fixed it.
Off topic, this has been my experience with AI so often that it prevents me from exploring AI uses more.
I liked Cursor’s “auto” plan but that now seems gone. I’d happily switch to a provider that offers a similarly “unlimited” usage.
It's difficult to offer unlimited usage of something that's so expensive to run. OP could have used a $20/month Claude or Cursor plan for "unlimited" usage within their quota had they been willing to use a different model than the $75/Million Token Opus 4
I would have, had I realized how much it would cost.
Well if cost is an object, there's always the quick & easy way :)
>Vintage Retro 3.5Mm Telephone Handset
https://www.amazon.com/Telephone-Receiver-Radiation-Micropho...
It is possible to radically increase your chances of success. You have to speak the LLM’s language, just like you write Java or Rust. But it doesn’t come with a language spec, so you get to figure it out by trial and error. And a model change means revisiting what works.
Lots of tips on how to do this out there but one thing I do is have it try, throw away everything it it did, and try again with a completely restated question based on the good bits in what it was able to produce.
E.g., if you ask for a web app that does X and it produces a working web app that doesn’t do X, throw that away and just ask for the web app scaffolding. You’ve still come out ahead even if you take over fully.
> Lots of tips on how to do this out there but one thing I do is have it try, throw away everything it it did, and try again with a completely restated question
This is the thing that worries me about AI/LLMs and how people "profess they're actually really useful when you use them right": the cliff to figuring out if they're useful is vertical.
"You’ve still come out ahead even if you take over fully."
I just finished a weeklong saga of redoing a bunch of Claude's work because instead of figuring out how to properly codegen some files it just manually updated them and ignored a bunch of failing tests.
With another human I can ask, "Hey, wtf were you thinking when you did [x]?" and peer into their mind-state. With Claude, it's time to stir the linear algebra again. How can I tell when I'm near a local or global maxima when all the prevailing advice is "I dunno man, just `git reset --hard origin/master` and start again but like, with different words I guess."
We have studies that show people feel like they're more productive using AI while they're actually getting less done [1], and "throw away everything it did and try again" based on :sparkle: vibes :sparkle: is the state of the art on how to "actually" use this stuff, I just feel more and more skeptical.
[1]: https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...
For an update, Codex has been massively better for me. I should probably give it another go, especially now with Codex Max, as it has been much better at not getting stuck like this. Give it a go if you get the chance, although with the general caveat that people have massively different experiences with LLMs, for some reason.
I think maybe it has to do something with the prompting style, my hypothesis is that some people's prompting styles fit certain LLMs better. I don't know how else to explain the fact that my very experienced friends prefer Sonnet to Codex, for example, whereas I had the opposite experience.
Toll.... so worth the effort.