Hacker News new | past | comments | ask | show | jobs | submit login
IBM Audible Random Timer (oldbytes.space)
97 points by luu 6 months ago | hide | past | favorite | 36 comments



It's a work sampling timer. When it goes off, you record what you were doing.[1][2] After a few days or weeks, a pattern of how much time is spent doing what appears. It's a management idea that never really caught on, because it's so annoying. Probably saleable as an app today, with a cloud-hosted back end.

[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC225293/pdf/mlab...

[2] https://www.hca.wa.gov/assets/billers-and-providers/particip...


> Probably saleable as an app today, with a cloud-hosted back end.

https://apps.apple.com/us/app/bereal-photos-friends-daily/id...

But a corporate version!


Oh I built one of those with Rust and FLTK: https://six-five-six-four.com/git/reactor/annoying_journal

I figured if it works for software profiling it oughta work for people. And I set the interval to some Golden Ratio thing like 37 minutes, with the hopes that it wouldn't log the same times every day, but also wouldn't be purely random.


Wait, so how well did it work?


I don't remember any big insights except that I'm pretty bad at estimates.

Everything that feels like a "half-day" task took an entire day or two. I would look back on every feature and think, "No way was I working on that for X days straight?"

Basically nothing takes less than half a day.


I’m going to claim that lesson alone made it a success

Small wins are still wins


I wondered if you could use a Geiger counter with background radiation to do the same thing, but it'd probably be thrown off by moving between different rooms.


I think you’d need a very insensitive Geiger counter. Mine is relatively cheap (GQ GMC320) and still clicks 30 times per minute as background radiation.


You could put it inside a box maybe.


Make sure you note down what you're doing each time the cat dies and hand the report to your supervisor at the end of each week


Yes you can it is a type of quantum random number generator. See https://en.wikipedia.org/wiki/Hardware_random_number_generat...

The biggest drawback is non uniform distribution and low bit rates. Another was radiation concerns.


Similar to Reporter:

http://reporter-app.com/


I have seen something like this as a safety mechanism. It fires and you are supposed to either check in or verify something that is critical to life. The randomness is to ensure that psychologically the alarm is not ignored or becomes habit to respond to.

Not positive if this is the same though.


I wonder how well this would keep me/someone from accidentally hyperfocusing-from experience, a 10 minute repeating timer helps somewhat, but I have always been curious about a random length timer instead.


Is hyperfocus a state you want to avoid? In my experience I spend most of my time (as a developer) hoping/trying to induce hyperfocus as that's when I get most productive work done...


I fairly regularly hyperfocus on something that isn't the thing I should really be hyperfocusing on and end up spending hours fleshing out a less-than-supremely valuable idea without meaning to. Hyperfocus in the wrong direction can be a real problem!


That's the blessing and the curse of ADHD/autism. It's easy to get hyperfocused, but it's not always on something productive.


Some people struggle more with directing their attention than maintaining it.


Same here. I go further and mute all notifications/my phone to avoid distractions.


I have a friend who has a pomodoro integrated DE. So whenever he does anything there is a colored line around the screen that slowly shrinks while it counts down. And when it reaches the end I think his screen shuts off or something, not sure what he set it to do. But it's following the famous pomodoro technique.


> So whenever he does anything there is a colored line around the screen that slowly shrinks while it counts down.

That might literally drive me crazy. Watching with dread as the line disappeared would disrupt my work completely. Eventually I’d just lose it like a character in a Poe story.


I use a Pomodoro timer browser extension. It’s a small toolbar icon that shows a minute countdown, so it’s not too distracting (until it’s supposed to be).


That makes much more sense to me.


Just try it!

I’ve used bash like this many times before:

  while sleep 600
  do
    say “hey!”
    sleep $(( $RANDOM % 300 ))
  done


If you're on a Mac, you can try enabling the system setting to "announce the time," if you like being startled by a robotic voice on the hour.


Does anyone happen to know an iOS app that does something like this? Preferably with configurable upper and lower bound for the random interval, and maybe times of day when it should be active, and/or a settable timer when it should stop being active.


Reporter. I saw a pretty cool visualization of someone's activities over a year, and decided to give it a shot. I gave up after the first day because of how annoying it is.

http://reporter-app.com/


Thanks. I don’t want an app that asks any questions about my activities or the like, just one that sounds an actual alarm (not a notification).


I've been using grandfather clocks ticking sound to help me focus on the laptop on a noisy environment.

Would find it funny to see IBM had an official tool to help people focus, but for now I'll assume it's for random sampling of something.


Similar to a dead man switch. You acknowledge the device on random intervals. Failure to acknowledge the device means the person is presumably incapacitated, so the heavy vehicle/machine they are supposed to be operating (train, boat, forklift whatever) automatically comes to a halt.

This seems conceptually similar but without being part of a fail safe mechanism.


What does the random distribution look like?


"This is I think where the "randomness" is generated - assuming the two counters run at different rates, the comparator would trigger the alarm when the two counters hit the same number at the same time."

I think the above probably isn't right...

The 4060 counters have a built-in oscillator. My guess is that it starts them at the same time and the alarm is when they don't have the same number...when there's drift as compared across the two oscillator sources. And that the clock dial probably changes the speed of one of the oscillators, so they match well enough to stay in sync for the rough desired period.

All that to say the random distribution is probably based on the variability between any two 4060 counters.


So these device all have a different distribution?


Even highly precise and expensive timers will drift.

Lesson learned: we built a 42-node several petabyte storage cluster. Worked fine as we were starting small, and got bigger, and bigger. When we “had it” and disconnected from our internal network and moved to an isolated network we discovered quickly that the timebase on all the systems drifted very quickly to be seconds and then minutes out of sync after several days. This was because they couldn’t access NTP.

Set up an ntp server and “resolved” it, except now we had an entire cluster of hardware that was (in lock step) drifting out of sync with the rest of the world. Fixed that. Moved on.

Another example: we bought a bunch of really cheap LED ropes to make our LGBT+ pride float’s backlighting for a pride parade several years ago. I wired it. The ropes were all independent and we planned to have them rotate colors. The little controller boxes had a mode that would smoothly transition through presets. It turned out that simply turning on the power would cause the lights to start in sync, and then quickly drift out of sync in a really mesmerizing way. They’d occasionally come back together but frankly it was better than anything we could’ve programmed.

We also learned we could control the rate of the “smear” by over or under volting the LEDs a bit.

Even if it’s digital, everything is eventually analog. And analog is weird.


Data sheet for the CD4060B, does have some info on variation of frequency between units:

https://www.ti.com/lit/ds/symlink/cd4060b.pdf


Ok, but what I really meant was: will the IBM random timers all have a different mean time between beeps (etc. for other statistical moments)?




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: