5 ms·

I think the Lindy effect can work in reverse. Like if you had a collection 1000 things that are all just invented, then the likelihood would be only a small per

I think the Lindy effect can work in reverse. Like if you had a collection 1000 things that are all just invented, then the likelihood would be only a small percentage of them will be around and in use in 50 years. Compared with a collection of 1000 things that have all be in use for 100 years, the percentage of them that will be around in 50 years will be much greater.

toss1 4y ago
Yup, it seems that this 'effect' is simply extrapolating from randomness and averages. Take any random thing and random time (with zero knowledge of actual lifespan), and on average, you are in the middle of it's lifespan. Therefore, if Thing-A has existed for 28 years, it is likely to last for another 28, if Thing-B has existed for 6 years, it's likely to exist for another 6, and so on. It may be somewhat informative for comparisons but not in real life. You are hiking away from a disaster with all your possessions and life's savings in your backpack, and are now at a muddy riverbank needing to cross. You ask me how deep the river is and I tell you the average depth is 6 inches. That sounds great, but I have most definitely NOT told you that you'll be able to get across without finding a deep spot and having to drop your backpack to survive. Using this effect to make judgements about product lifetime is similarly uninformative. It is a hint leading to only a possible inference, not data leading to a valid prediction.
- syntheweave 4y ago
  With respect to software objects, the most likely reasons for their life to end are: 1. Not sufficiently useful relative to involved costs for most applications(data formats, configuration maintenance, etc.) 2. Disrupted by something that is "10x better" for the purpose(e.g. using a spreadsheet instead of a text editor for 2D, cell-oriented data) 3. Outside forces invading the ecosystem and obsoleting dependencies(new OS, hardware, etc.) So what the Lindy effect describes in long-lived software is just the software that is relatively cheap to keep around, is hard to greatly surpass and resists invasion - which describes a lot of "worse-is-better" software, where the UX kinda sucks, and it's actually a bit too unstructured for any particular application, but not enough of these things that anyone cares to address it in the relevant professional scenarios where it comes up: instead the user just girds themselves to fight it into submission because they can spend six hours fighting it and one week debugging it or two weeks making a Right Thing that is much less compatible.
PaulDavisThe1st 4y ago
It's not reversible, inasmuch as your first statement is correct, but cannot identify the things that will be around.
- Eddy_Viscosity2 4y ago
  It doesn't claim to be able to pick which specific things will survive, it only gives an estimate of the probability that something will survive over a certain time frame.
skykooler 4y ago
You can't apply it endlessly in reverse though. Say you have a piece of software that's been around for 25 years - assuming that it's halfway though its life is reasonable. If something was just released 25 minutes ago, assuming that it will be abandoned in another 25 minutes is preposterous, since products tend to have a minimum amount of time for which they stay relevant. You could use this argument to create an "ultraviolet catastrophe"[1] for the Lindy effect and argue that virtually all software should be abandoned microseconds after being released. [1]: https://en.wikipedia.org/wiki/Ultraviolet_catastrophe https://en.wikipedia.org/wiki/Ultraviolet_catastrophe
- esrauch 4y ago
  I'd actually assume most GitHub repos get exactly one commit and are effectively abandoned in actual zero time.
- bee_rider 4y ago
  If software is only 25 minutes old, you probably aren't using it.
- Eddy_Viscosity2 4y ago
  It's a statement about probability. You can't assume that that specific piece of software will be abandoned in 25 minutes. But, if you had to guess which was 'more likely' to be around at some future time, then the Lindy effect would suggest its the software that's been around for 25 years over the one that was just released.

toss1 4y ago

Yup, it seems that this 'effect' is simply extrapolating from randomness and averages. Take any random thing and random time (with zero knowledge of actual lifespan), and on average, you are in the middle of it's lifespan. Therefore, if Thing-A has existed for 28 years, it is likely to last for another 28, if Thing-B has existed for 6 years, it's likely to exist for another 6, and so on. It may be somewhat informative for comparisons but not in real life. You are hiking away from a disaster with all your possessions and life's savings in your backpack, and are now at a muddy riverbank needing to cross. You ask me how deep the river is and I tell you the average depth is 6 inches. That sounds great, but I have most definitely NOT told you that you'll be able to get across without finding a deep spot and having to drop your backpack to survive. Using this effect to make judgements about product lifetime is similarly uninformative. It is a hint leading to only a possible inference, not data leading to a valid prediction.

syntheweave 4y ago
With respect to software objects, the most likely reasons for their life to end are: 1. Not sufficiently useful relative to involved costs for most applications(data formats, configuration maintenance, etc.) 2. Disrupted by something that is "10x better" for the purpose(e.g. using a spreadsheet instead of a text editor for 2D, cell-oriented data) 3. Outside forces invading the ecosystem and obsoleting dependencies(new OS, hardware, etc.) So what the Lindy effect describes in long-lived software is just the software that is relatively cheap to keep around, is hard to greatly surpass and resists invasion - which describes a lot of "worse-is-better" software, where the UX kinda sucks, and it's actually a bit too unstructured for any particular application, but not enough of these things that anyone cares to address it in the relevant professional scenarios where it comes up: instead the user just girds themselves to fight it into submission because they can spend six hours fighting it and one week debugging it or two weeks making a Right Thing that is much less compatible.

PaulDavisThe1st 4y ago

It's not reversible, inasmuch as your first statement is correct, but cannot identify the things that will be around.

Eddy_Viscosity2 4y ago
It doesn't claim to be able to pick which specific things will survive, it only gives an estimate of the probability that something will survive over a certain time frame.

skykooler 4y ago

You can't apply it endlessly in reverse though. Say you have a piece of software that's been around for 25 years - assuming that it's halfway though its life is reasonable. If something was just released 25 minutes ago, assuming that it will be abandoned in another 25 minutes is preposterous, since products tend to have a minimum amount of time for which they stay relevant. You could use this argument to create an "ultraviolet catastrophe"[1] for the Lindy effect and argue that virtually all software should be abandoned microseconds after being released. [1]: https://en.wikipedia.org/wiki/Ultraviolet_catastrophe https://en.wikipedia.org/wiki/Ultraviolet_catastrophe

esrauch 4y ago
I'd actually assume most GitHub repos get exactly one commit and are effectively abandoned in actual zero time.
bee_rider 4y ago
If software is only 25 minutes old, you probably aren't using it.
Eddy_Viscosity2 4y ago
It's a statement about probability. You can't assume that that specific piece of software will be abandoned in 25 minutes. But, if you had to guess which was 'more likely' to be around at some future time, then the Lindy effect would suggest its the software that's been around for 25 years over the one that was just released.