Hacker News Reader

Hacker News

Comment by chubot | Hacker News Reader

Parent Full thread

chubot

Taleb talks about the Lindy effect and some of the surprises in this article can be explained it.

https://en.wikipedia.org/wiki/Lindy_effect

The rough heuristic is: the longer a technology has been around, the longer it will last into the future.

That is, it's NOT the case that every piece of technology lasts roughly the same amount of time, and then is replaced.

For example, a chair vs. an iPhone. Which one will be used further into the future? Almost certainly a chair.

----

Land lines have been around a lot longer than fax machines (both in the article), so they will likely outlive fax machines.

Will HTML or JavaScript last longer? Probably HTML, since it came first.

What about ASCII or HTML? Probably ASCII.

These have a "dependency stack" issue, but it applies regardless. And I think that is part of the phenomenon -- low level technologies that are widely adopted take longer to go away. Plumbing, electrical (AC vs DC), land lines, TCP/IP, BIOS, etc.

I can't find a link now, but there was a recent Kevin Kelly article about finding farming tools from a catalog in the 1800's still in use. I think he said that almost EVERY one was still in use, or able to be purchased, which is a nice illustration of the point. It takes a long time for old tech to die, and arguably it never does.

View on HN

A nit, the telephone and fax machine developed more or leas in parallel, and the first working fax predated the first telephone by 11 years!

They were sending drawings through telegram lines for newspapers during the American Civil War. Thus, the fax is older than the phone. But in terms of general population use, of course most people encountered a phone before they encountered a fax.

Interestingly, you can still send a telegram.. https://www.itelegram.com/

I guess these guys took over all the telegram infrastructure of Western Union in 2006.

Telegrams were always expensive but the price now is outrageous. I followed the link and chose Norway as the destination country and was told that it would cost GBP 22 + 67p/word! Even worse for Mexico: GBP53 + 67p/word.

It's cheaper to send flowers with a card!

That doesnt sound like fax. Or better, if you include this and fax together, you’d have to describe it as a generic technology that allows sending images remotely. Which may actually survive the telephone.

Source?

Google "fax machine history." This is incredibly trivial to verify.

Backing up the original point of the parent post's spirit, I actually see fax lasting longer. It's special cased in a lot of regulatory structures as 'secure' and has quite a bit more use than the HN crowd might think.

Fax is the cheapest way to get forecast if you sail offshore through a service called "weather fax". Alternatives works through satellite with hefty monthly subscriptions. That service is the reason I bought yet another raspberry pi with a software defined radio module.

This is, of course, silly but true, because fax has no endpoint authentication of any kind.

It kind of makes you feel ridiculous about the amount of work that we put into securing even systems that are in no way critical and then the world runs on unsecured mailboxes, homes which can be broken into by someone willing to kick hard enough, Social Security cards with no security features whatsoever, funds transfer that only requires your account and routing number to withdraw, and so on.

It's not that ridiculous. Securing stuff on computers is more important due to the scale at which attacks can occur. On the internet, you can gather information on millions of users in little time and without putting yourself in a place where you could easily be caught. And for a machine that isn't connected to the internet, it can still give you far more data than you could get by stealing paper that takes up the same amount of physical space.

I guess so, but there are many Internet accounts I would be way less upset about someone compromising than the other things I mentioned.

I'd say most of the parents' comment is applicable to the same scale. People can abuse your social security number or bank account and routing numbers across the Internet. Phone numbers being relatively easy to spoof and/or hijack makes fax trivial to mess with internationally as well.

I think you are understating the extent to which mailboxes and homes are secured by men with guns. Sure; there’s very little preventing someone from kicking their way into someone’s house, but if someone did so, they would very likely end up in a jail cell or even, in some places, dead.

Unfortunately, you can’t physically harm people over the internet, so different security measures must be taken.

Your odds are very good stealing from my mailbox and not that bad burglarizing my house. Maybe you have much more vigilant police where you live. What’s more, unauthorized access to computer systems is also a crime.

The odds of getting caught robbing a single mailbox are not high, but if you do it on a large scale, it starts to get pretty risky.

The odds of getting caught attacking a Russian computer from the US or attacking a US computer from Russia are essentially zero for any scale small enough to not have major foreign policy implications.

Homes are massively protected by people’s goodwill, not by the judicial system. In most of Europe and more than you suspect in US too, if you call the police, they come more than 30 minutes later (especially in France where they have no right to use their guns and don’t risk their own security), and if you “handle the matter yourself”, you are in just as much problems as the thieve, particularly in US where you still have to use a lawyer to prove your innocence. No, what really holds houses from being broken into, is mostly that people don’t do it (hence the use of a society where people earn enough to not be willing to risk physical fight).

Most forms of security rely on the goodwill of most people in society; the entire problem consists of handling the minority of people who lack goodwill. These people exist in every society.

> Unfortunately, you can’t physically harm people over the internet

Yes how unfortunate?

I was at a DefCon talk where these researchers demonstrated pwning a network by sending a fax: https://blog.checkpoint.com/2018/08/12/faxploit-hp-printer-f...

You could put your fax machines in a locked container

I was thinking more of SS7-based attacks for remote interception.

Regulatory structures are not forever and will change.

33 years, if you go by the patent. Lots more fun facts in Tim Hunkin's excellent Secret Life of Machines:

https://www.exploratorium.edu/ronh/SLOM/0301-The_FAX_Machine...

Tim Hunkin is recently making new videos!

https://www.youtube.com/c/timhunkin1

This message is a telegram.

Interesting! Could you expand a little on what you did?

Wow, given the answers to that thread, it seems that the HN crowd has either no notion of conditional probabilities or that their “future is better” mindset is clouding their rational judgement.

The Lindy effect makes perfect sense has an heuristic to evaluate the remaining lifetime of technologies and species at a given point in time. It’s trivial to prove that it works for anything whose survival curve is convex.

Despite its apparent love for rationality (which I enjoy), it seems like a large part of HN is just as subjective as anyone and unable to accept things that don’t fit their mental model.

This field also seems to attract (or foster) a particular mindset that we might call "I am smart and hyper-rational", which then adds an additional set of blinkers and an enhanced ability to discard and ignore evidence, and an enhanced faith in one's own powers of deduction.

All people have that skill, of course, but this field seems to have more than its fair share of such.

> Wow, given the answers to that thread, it seems that the HN crowd has either no notion of conditional probabilities or that their “future is better” mindset is clouding their rational judgement.

I mean it's a tech entrepreneurial news aggregator on a tech incubator's website. Their whole raison d'etre is that "future is better buy my technology so I can get $$$" -- that's why this site exists.

Is it that it doesn't fit the mental model, or is it failing to read the room?

I believe that scientific rationality and engineering are part of YC’s DNA. This thread seems to prove I’m partly wrong indeed.

Humans.

humans...

It is an interesting observation to compare longevity times for current widely-used IT tech:

Smartphones (since first iPhone) - 14 years.

Laptops (since first Apple Powerbook) - 30 years.

PCs (since first IBM PC) - 40 years.

C programming language will be 50 years old next year.

SQL will turn 50 in 2024

> C programming language will be 50 years old next year.

This principle implies the C programming language will outlive Rust, Go, Python, etc. I'm not sure how I feel about that. There's a good chance its right, but I'm uncomfortable imagining my grandchildren learning C's weird quirks.

There's definitely codebases written in C today which will outlive me, like the Linux kernel. And most of those codebases will probably never migrate to a different language.

I think C will outlive those languages, in the sense that it will exist. But it doesn't necessarily follow that your grandchildren will have to learn it. There will be some other language like JS or foobar 2050 that's just way more relevant and popular :) But they certainly COULD if they wanted to.

The question of whether a technology simply exists vs. whether it retains its popularity is an interesting one. Chairs and forks will both exist and be popular. iPhones will exist, but they won't be popular on some time scale. So it is interesting to think whether C is closer to one or the other :)

C is the "lawn chair" of programming languages. Sure, it is neither particularly fancy, nor most comfortable - but beats having no chair :)

While C is not perfect, it is in a sense something that most other programming languages are not. It's simple and if you program in C you probably touch upon the entirety of the language on a regular basis. Because it is simple you can write a compiler for it with some reasonable effort if need be, even for completely new architectures.

In any case, I'd argue that most of the quirks are due to undefined behaviour as implemented in the compilers. And 50 years from now the language have probably evolved some more anyway.

I'm more worried my grandchildren will still be learning/using QWERTY than they have to learn C's weird quirks :). My opinion is that it's easier to get a group of tech people together and gradually replace C (like Android Rust development) than getting people to replace QWERTY or getting the US to move to Metrics system.

what's wrong with qwerty?

Arguments could be made that it's inefficient or un-ergonomic, though I'm doubtful how much difference it makes on a modern keyboard. There's an oft-repeated story about the qwerty layout being designed to keep people from typing fast enough to jam typewriters. I can't speak to the legitimacy of the story, but I do collect typewriters and I can say for sure that there's a lot more pressure and key travel (and time!) required to slam a type bar into the ribbon than is needed to press a modern keyboard key.

I learned Dvorak a decade ago and loved it unless someone else had to use my computer or I had to use a shared computer. I'm typing this on a qwerty keyboard and I'm okay with that now.

> There's an oft-repeated story about the qwerty layout being designed to keep people from typing fast enough to jam typewriters.

Typewriter layouts are designed to avoid jamming, but not by reducing typing speed. They place letters that are pressed in direct succession apart, so that the levers don’t collide. Trained typists can write pretty damn fast on a typewriter.

I thought I remembered something about that, thanks. Dvorak does seem to cluster a lot of commonly used letters, which would definitely jam neighboring type bars pretty quickly unless they used a Blickensderfer-[1] or Selectric-style unified type element.

I've always wanted a Blick, their "scientific" DHIATENSOR layout intrigues me. They also made a qwerty version and I seem to remember a story that their salespeople would make you sign a waiver that you were choosing a less efficient keyboard if you bought one. That's certainly 90% marketing for their layout and maybe 10% fact, but I find it amusing (if true, but I can't seem to find a reference to it any more).

I'll bet typewriter-trained typists can really move. It's definitely a separate skill, at least for the manual typewriters in my collection which have about an inch of key travel and a heck of a lot more actuation force than the gateron red switches I use at work. :-)

Ultimately I find that the fastest layout is the one you know.

[1] https://en.wikipedia.org/wiki/Blickensderfer_typewriter

> They place letters that are pressed in direct succession apart,

... and that really reduces typing speed. When i switched to dworak I was amazed how my fingers just roll on a keyboard, forming sentences faster than I even form a conscious understanding in which succession should I move my fingers. It just happens as if by itself!! Just think of a word to type and it is already on the screen.

> ... and that really reduces typing speed.

That may be the effect for you, but it's not a goal. It's maybe a tradeoff, but machine writing is a trainable craft. Usual results for trained people are at about 200 - 400 characters per minute measured over 10 minutes. Championship results are at up to 900 characters.

Yeah, and all championship results are on Dvorak keyboards. [1]

[1]: http://www.recordholdersrepublic.co.uk/world-record-holders/...

There are plenty of unofficial records and leader boards for typing speeds where qwerty typists are just as capable (more so even due to sheer availability) as Dvorak typists at hitting 200+wpm, and while that alone shouldn't be used to judge the quality of a particular layout it shouldn't be ignored that there's no difference in attainable max speeds done in short bursts. That says nothing about sustained typing of course, but I strongly suspect if you're already a competent and fast typist in qwerty you'll be comparably as fast in Dvorak and vice versa.

Dvorak is actually pretty bad as a smartphone keyboard, it's advantages with two-handed typing on real keyboard become disadvantages if you type with your thumb.

Dvorak has specialised right- and left-handed layouts. Maybe it also needs specialised one finger layouts, too.

I had wondered about phone keyboards. Part of the reason I switched back was that I couldn't change the layout on my phone and got tired of switching back and forth. (I know there are probably some options now, but there weren't on iPhone 4, Kindle touch, or feature phones with physical keyboards.)

I can imagine you'd need to use both thumbs extensively or you'd be jumping back and forth. I definitely found that when I was laying on my side on the couch, propped up with one arm, qwerty was easier to use with one hand. Not that that's an ergonomic or sustainable way to type, but I was a lazy college student and I didn't care.

> I'm uncomfortable imagining my grandchildren learning C's weird quirks

What about them learning English's even weirder quirks? Does that make you uncomfortable as well?

I have come to the point that I sort of accept C's quirks, and, with the appropriate mindset, I can even love them. I certainly see myself as an old man teaching these arcane quirks to a young and innocent audience.

Also COBOL is still around and most people don't need to learn it anymore.

Most people don't need to learn C either; or any programming language, for that matter.

Sure but I think programmers is an implicit assumption here when talking about which programming languages to learn. We don't need to start splitting hairs here.

And COBOL is extremely niche, while C is still quite common and useful to know.

Isn't that the whole worse is better thing in action? People complained about C and Unix when they were new, but they picked up so much traction, and people invested enough effort into working around the flaws, that it just achieved critical mass. Same thing with JavaScript; it's never going anywhere.

Certainly higher-level languages have edged out C in many domains though.

> Laptops (since first Apple Powerbook) - 30 years.

That kind of Apple revisionism is not correct. The NEC UltraLite predated Apple clamshells by 3 years (1988), and luggables with a battery configuration have been available as early as 1981 (Osborne 1 with an aftermarket 1 hour battery add-on).

The Grid Compass was probably the first true laptop in 1982. But it ran a proprietary operating system by default and costs about $8,000 in 1982 money. The Data General-One came out in 1984 and might be a better candidate. But neither of those were really mainstream.

Don’t neglect the Model 100...

“The TRS-80 Model 100 is a portable computer introduced in 1983. It is one of the first notebook-style computers, featuring a keyboard and liquid crystal display, in a battery-powered package roughly the size and shape of a notepad or large book.”

https://en.m.wikipedia.org/wiki/TRS-80_Model_100

It wasn't a laptop in the modern sense though.

For any reason other than it doesn't fold in half?

The iPhone was definitely not the first smart phone.

Nor was the Powerbook the first laptop nor the IBM PC the first PC. Those are all, however arguably responsible for popularizing the technologies in something resembling their current form.

I disagree. I am of the opinion that those form factors were inevitable, and would have come along anyway with or without Apple's offerings.

How is that inconsistent with the parent statement? Even if the form factors were inevitable--e.g. someone would have decided that smartphones didn't need a physical keyboard--someone had to be first to popularize. (Not sure I agree on the laptop but the iPhone pretty clearly popularized the modern smartphone form factor.)

To say they are "responsible" for something can be read as "allowed to take credit for". Being the first to popularize something is a business achievement, which is quite meaningless compared to the technical achievement of being the first to build something.

Something that's a mainstream business success is far more interesting in general than progenitors that never really took off for whatever reasons. They're may still be important as technical achievements but the history books notice those who took things mainstream.

James Watt didn't actually invent the steam engine. He just came up with an invention that made it a lot more efficient.

I can see this in the case of the PC and the laptop, but proto-smartphones were around for years before the iPhone and they were not developing in an iPhone-like direction. Every iPhone-like product that came after was the result of copying. I don't think it was inevitable at all.

One thing I do think is true is that Apple developed sufficient brand permission to be able to do things that were definitely outside of a lot of buyers' comfort zone and arguably needed some iteration to really nail.

I'm still inclined to think someone would have jumped to a smartphone without a keyboard. But it's also true that the iPhone had plenty of critics early on and arguably didn't fully hit its stride until the 3GS.

No, but it was the first wide spread smartphone and the first one most people used. This is the only practical starting point.

Blackberry, Palm, and Windows Mobile phones were more widespread than you're giving them credit for, as were laptops considerably preceding Apple's PowerBook series.

The mindblowing part about these kind of numbers for me always is the sheer amount of smartphones out there.

I mean, imagine a parallel world where those smartphones weren't designed to shove ads down your throat and where they could be used to be as productive as with a laptop, and where people could help to automate their lives on their own with it.

That would be so amazing.

Smartphones are a huge productivity tool, that's why they took off in the first place. Especially Blackberry, which offered the magic technology of accessing your email and calendar from anywhere. The ads are not an obstacle to this, especially not on iPhone.

They are far more than that. For a large number of people, the smartphone is their first and only computing device. Enabling internet access is like rocket fuel for advancing socioeconomic conditions for those in developing nations.

Entire generations have been lifted from poverty due to it.

And "first and only computing device" pales to insignificance next to "first and only bank account" which they also are, in a way. That's huge.

I don't know about you, but a consumption-only "first and only computing device" sounds incredibly dystopian. I get that they have a positive impact, but it feels like we could be doing better.

> I don't know about you, but a consumption-only "first and only computing device" sounds incredibly dystopian

Smartphones aren’t consumption-only.

Doing anything but consumption and lightweight content discovery on smartphones is basically a farce. They are not usable as general-purpose creation machines.

> Doing anything but consumption and lightweight content discovery on smartphones is basically a farce.

I disagree, even when talkingn about a smartphone as a human interface device and not, as was the actual context of the thread, a computing device.

Having a, say, DeX-enabled Android (or even potentially powered by fairly traditional Linux, if one doesn’t just use a major maker device with stock software) smartphone as a computing device doesn’t preclude standard desktop I/O devices fo interact with it in manner very similar to a standard desktop PC.

> They are not usable as general-purpose creation machines.

Again, as an exclusive computing device, they are just about as much as is any computing device.

Even as an exclusive HID, there’s a giant excluded middle between “consumption only” and “not usable as general purpose creation machines”.

What makes a smartphone a smartphone is the ability to use it with just the touchscreen. A smartphone with desktop peripherals is just that, a desktop. What I am arguing is that a bare smartphone (aka what most of those people in the developing world can afford) cannot meaningfully serve, due to being a form factor with a very imprecise input method, as a creation platform for any sort of precise content.

> What makes a smartphone a smartphone is the ability to use it with just the touchscreen

You seem to be conflating the ability to be used with a limited interface with a restriction to be used with only a particular interface. I desktop PC has the ability to be used without a graphics tablet. It would be a much more limited creation device if it were restricted to being used without one. And the same is true of everything that isn’t a touchscreen, microphone, or other built in human interface and smartphones. But a smartphone as an exclusive computing device does not imply the limitations of its built-in interfaces, since having external interfaces is also a standard feature of the class.

While I agree with your definitions of what a computation device is, I also don't see Android being developed on Android in the near future. So many UX conceptual problems won't allow this, so I think that as long as computers are required to build smartphones, they cannot be described as "general purpose" machines.

Its still a poor version of the internet when compared to a PC (laptop/Desktop).

That world cant exist. A flaw of anti capitalist alternatives is that they're non-natural.

Any alternative where trade didn't follow optimal path is due to regulation / force. And in most, the force required to steer humans away from their nature also kills innovation.

Ergo: you cant have a miracle chip in your pocket with no one using it to sell you potato chips.

1. Nobody mentioned capitalism or alternatives in the parent comment.

2. Capitalism is not "natural" either and only started developing in the post-renaissance world.

3. Natural selection demonstrates conclusively that relying on purely natural processes to drive "innovation" often lead to highly non-optimal and harmful trends.

4. Most of the major scientific and technological innovations of the 20th century were either the direct result of or funded by "force" which I take to mean government.

5. Advertisement was not the only monetisation strategy that the internet and telecommunication industry could have taken. The internet could have very easily gone down the subscription route and the only reason why anybody thinks otherwise is because of the decade long marketing that has normalised getting everything for "free".

Laptops existed far before the Powerbook.

It really depends on how you define laptop. There was this sort of thing in the 80s, (https://en.wikipedia.org/wiki/Toshiba_T1100), of course. And there were a couple of laptop-shaped laptops from 1988 on, but they were either spectacularly compromised (the NEC Ultralite had a max of 2MB of storage, for instance) or spectacularly expensive or both. In terms of laptops that were shaped like laptops as we know them, and that were actually usable outside of very specialized applications, the Powerbook 100 and Thinkpad 700T could be reasonably claimed to be about the first.

There’s also the TRS-80 Model 100, launched in 1983

Every time someone brings up the Lindy effect I can't help but roll my eyes. It should be replaced with "Survivorship bias". Almost every technology that humans used that lasted for a long time and no longer used has disappeared and is no longer in used, tautologically. The Lindy effect just seems to be a list of examples of cherry-picked technologies.

Based on your eye rolls and subsequent "explanation", it's clear that you don't understand the Lindy Effect. It's not about listing examples of things that have been around for a while. It's about predicting the likelihood that something will continue to be around given how long it has already been around. This effect is well studied and just a cursory glance at the Wikipedia page will give you some solid sources for more rigorous understanding.

There are no "solid sources" there. Its a bunch of books and articles.

Well-studied? By whom? In what journals?

The Lindy Effect may be true, but based on that Wikipedia article's sources, you can't make a good scientific claim for that being the case. Even if you could, you still run into all the current problems such a nebulous branch of science must contend with, such as the peer review problem and the reproducibility problem.

I think you could mount some interesting objection to the Lindy effect, but this isn't it. I'm not really sure what you're trying to say.

It's not claiming to be a scientific law; it's a heuristic for making decisions. The rest of Taleb's books are also about making decisions, not "being right" (whatever that means).

A concrete example is if I'm writing a blog, and I want people to read my posts in 5 or 10 years. Do I go with the cloud platform that just launched or an older hosting provider? This is a decision people make every day. Of course there are many people who don't care if their blog is readable in 5 years; this isn't a judgement.

The Lindy effect is not about what's "better"; it's about what lasts longer. It's also not making statements about the present, which is what survivorship bias typically means.

Another helpful angle is to consider things that aren't Lindylike. People for example -- we expect older people to die sooner than younger ones, not later. And radioactive nuclei -- we expect their ages to be irrelevant to their expected future longevity.

Yeah there are definitely some subtleties, and they would be interesting to tease out.

Older people obviously will die sooner, but having survived does give you some information. For example, your life expectancy at age 1 is 75 years, but at age 40 it's closer to 79 years.

e.g. https://www.annuityadvantage.com/resources/life-expectancy-t...

Isn't this handled by the fact Lindy-like things have to be non-perishable like ideas or technology?

Not if you are looking at motorcycle accidents data of older and younger people.

And if you are writing a blog, it should be written in a way that it could be read 10 years ago if you want it to be read in 10 years if we're following Taleb's thinking.

Survivorship bias is when you draw conclusions about all members of a certain class of things based only on the surviving examples.

The Lindy effect is a theory about the surviving examples specifically.

I can see that, I think the Lindy effect needs some refinement.

My personal take is that there’s a an apex for a particular generation of technology, and that is good forever. A 1930s Farmall tractor is an example of that... there are improved modern replacements, but the 1930 model still does the job near optimally. I would guess that a non trivial number of those tractors will be in use in 2130.

1980s/early 90s minicomputers are similar. Many of these devices are still in use today, and probably could be kept in use for decades to come.

Modern tech is a little harder because we’ve been in a rapid growth phase and the software services based world is more aligned with production than sustainment. I’d bet that trend will change in 20-30 years.

Yeah I just watched some videos that is extremely related by this modern homesteader (and YouTuber! -- apparently he was on the TV show "Alone").

He says "one of the best pieces of advice I've ever gotten is: Don't trade a gun for a snow machine". This is exactly what you're saying, and it's backed up by a lot of experience living without power and water!

https://www.youtube.com/watch?v=BH15Kua5P1Y&t=918s

He also says "everyone one of us has to decide when to jump ship on a technology"

He says canoes peaked in the 1960's, and you can buy a used one for like $125 that's the same as what you'd buy today for thousands. Same with hand saws. He maintains old saws and chainsaws and uses them:

https://www.youtube.com/watch?v=BH15Kua5P1Y&t=746s

when you look at any kind of manufactured goods a lot of things have reached their peak and are either poorer quality than they used to be or they're just the same quality as their peak

----

I found this channel via a video about building an off grid cabin from scratch for a couple thousand dollars: https://www.youtube.com/watch?v=bOOXmfkXpkM

It's good -- a lot of it is built by hand with a hammer and nails. He even says load bearing screws are too expensive, and nails are better!

All of the advice reminds me of Taleb, because it's not necessarily trying to be "right", but rather distilling rules of thumb from practice.

My favorite example of the quality issue is the “whirly pop”, a stovetop popcorn maker.

The old one my parents had was aluminum with a metal gear. The modern version has been MBAed to death — the gear is plastic, and the lid is so thin that you could probably replicate it with 2 plys of aluminum foil. It costs more and is measurably worse in any dimension.

I recently bought a leaf blower/vac mulcher. Took me way longer than it should to figure out that the difference between the $50 model and the $100 one was that the latter had a metal mulching blade instead of a plastic one, and that the former would likely break as soon as I vacced up a stick that was a bit too big. Thanks random forum poster!

I only somewhat agree with respect to canoes.

Grummans are great and they're still being made (though not by the original company).

However, for recreational/tripping/whitewater, Royalex-based canoes were better for a variety of reasons. Unfortunately the material is no longer being made because its intended use (Go Carts) didn't take off to the degree planned. The company continued to make it on a more or less breakeven basis but upon a change of ownership the new owner decided to scrap it. There have been one or two efforts to make something equivalent, but AFAIK they haven't panned out.

There are still plenty of well-made fiberglass/Kevlar boats being made but they're much more fragile.

The tractor thing is quite an interesting one( not necessarily with just this particular one): the older tractors ended up being so reliable that people often try to get an older one instead of splashing out on a brand new John Deer and this annoys the manufacturers down deep to their bones.

Yeah and it is tragic in some ways as the thing missing from the 1930s gear is safety features.

Many preventable deaths happen every year as a result.

Of course, you'll also see people arguing that you shouldn't be driving a 10 year old car for the same reason. There's some level of tradeoff where using an older product without the latest safety features makes sense.

Actually the same video I referenced above has a section on tractors! https://youtu.be/BH15Kua5P1Y?t=829

He says it cost me $100 to get the best that has ever been made, and it's backed up by a lot of experience living off grid

You may want to specify "lawn tractor" as that's not really what I was picturing when you say tractor.

It's exactly survivorship bias, but the contextual usage is different. Usually you use survivorship bias to discredit the relevance of an observation. You should think of the lindy effect as survivorship bias as a supporting heuristic for a prediction.

Except the Lindy effect does hold even when you use it as a predictor of the future, rather than just analyzing historical data.

I think the idea is if you randomly sample a range you have weak evidence as to the size of the range. For example if you randomly sampled and got "2" it would be more likely the range had a span of 0 to 4 than 0 to 100,000 though either are possible. On average your random sample will be at the halfway point of the range.

The Lindy effect is the realization that your encounter of something is like a random sample. "How old are chairs when I exist?" "How old are iPhones?"

I think the difference is that survivorship bias applies when the difference between winners and losers is mostly due to chance. I don't think the fact that we use 4-legged chairs and not 5-legged is survivorship bias. I believe the Lindy effect's prediction that 4-legged chairs will be around a long time. Of course, whether it's survivorship or not is case-by-case.

But every standard office chair with wheels is 5-legged.

Yep, and those are a recent development and much less likely to last than 4-legged chairs. So are, for that matter, offices.

There's a good reason why office chairs have five legs/wheels where regular chairs have only four: safety.

Office chairs have a reclining mechanism. If you ever leaned back too far in an old office chair with four wheels, you would find out the hard way that four isn't enough. It's very easy to lean back comfortably and not realize you've reached the tipping point, fall back and hit your head on the floor or have a close call.

My first jobs long ago had four-wheeled office chairs, so you can guess how I found this out.

The fifth wheel goes a long way to preventing this danger.

You're much less likely to be leaning back in a regular four-legged chair so the front legs come off the ground, and if you do you'll probably be more aware that you are doing something outside the chair's normal mode of use.

And even if offices go out of style, office chairs will likely be around long past then.

This logic makes me fear for the metric system

I curse the Imperialists every time I come across the cursed measurements.

This might be obvious, but... the reason we have a special name for things that behave that way (Lindy effect) is because this is usually a terrible heuristic. Most things are "perishable", as the article puts it - past a certain age, the older a human is, the shorter their life expectancy. This is true of most things.

The Lindy effect talks about the rare cases where this isn't true.

It is just a natural case of the exponential distribution, which is widely used to model expected waiting time. I wouldn't call it 'rare'.

Absent other observations and/or priors, the best estimate of your expected waiting time is the amount of time you have already waited.

So if you've been waiting 10 minutes for something to arrive, your best estimate of how much longer you have to wait is 10 minutes.

If you've been waiting 10 years for a tech to become outdated, your best estimate for how much longer it will take is 10 years.

People usually use buses as an example. So it makes sense: if the bus is supposed to come every few minutes, and it hasn't come in the last 15 minutes, then odds are something is wrong thus increasing the prob that the additional wait will also be 15 minutes.

> The Lindy effect applies to "non-perishable" items

It's the 2nd sentence in the article. So not sure using examples of perishables shows it is a terrible heuristic.

It applies to ideas/cultures/technologies.

Well yes, but not only. I think there are examples where it applies to living organisms too. E.g. a few hundred years ago, a human who lived past the age of 5 would be more likely to grow to be 40, afaik.

I was mostly pointing this out because a lot of people in the comments seem to be talking about this as if it's some kind of universal law, or a proposed universal law, when really this applies to only exceptional circumstances.

> Land lines have been around a lot longer than fax machines (both in the article), so they will likely outlive fax machines.

I'm less sure about this one, because of the relative difficulties of replacement. You can replace someone's POTS line with VOIP and they'll likely never notice (and this is underway). They'll notice if you take your fax machine.

>You can replace someone's POTS line with VOIP and they'll likely never notice (and this is underway). They'll notice if you take your fax machine.

The question is if fax on VoIP is still fax in that sense. At least the wikipedia article on fax mentions transmission through audio-frequency tones [1]. However if you count fax over VoIP as fax, I guess you should also count phone on VoIP as "landline".

[1] https://en.wikipedia.org/wiki/Fax

I haven't used fax for at least 15 years. Might be 20, anyway so long ago that I can't remember when it was.

Fax machines seem long extinct everywhere except in Japan and in some hotels. Landlines probably only survive in business environments, I hardly know anybody who would have and use a landline at home. Also, both businesses and consumers mostly use VoIP-based "landlines" employing codecs which can't support fax.

Faxes have become a largely digital thing, as there are various digital fax services. Neither side needs to have a phone connection that can support it. The fax services just need to be able to communicate to each other.

I was in a radiology startup for a bit, and getting people to fill out forms on something like an iPad is still a problem. You need staff able to help them, and people damage or try to steal them. So then you end up with paperwork, and if that paperwork needs to move somewhere else, people fax it.

Fax machines are still widely used in healthcare, at least in the U.S. Can't speak for other countries, but I'd be surprised if that weren't also the case elsewhere.

My employer was moving us to a new office, right before the pandemic, it's not seen much action, but... there's a fax machine. I doubt any of us would use it, but it comes with our corporate really estate package. Just in case you need to send a fax to Japan... or some hotel?

15 years ago, the office I worked in received daily menus via fax from local delis and restaurants with their specials, in case we wanted to get lunch. I have no idea if this is still a common practice but it seemed to be at the time.

This even had the slight benefit over sending email to a random address because a fax can just be posted on a common board in the office space, rather than someone having to take the step of printing the email first.

Perhaps. Whatever, it sounds comforting and fun. I've heard they have a colour (!) fax machine at almost every home Japan. If I moved into an office which had one, I would actually contact somebody there and have fun sending hand drawn pictures to each other :-)

Landlines are probably more common than you think. I only got rid of mine last year. I would have kept it for backup but it just cost more than I was willing to pay. Many of us don't get great cell reception and WiFi assist isn't always perfect.

Where are the Trinitron displays and ICs made with 1 µm process?

Adding to my first reply: if Trinitron displays were gone (it appears they aren't), and some newer tech isn't, that would NOT contradict the Lindy effect.

If you already KNOW that Trinitron displays are gone, then there's no uncertainty. The Lindy effect is a heuristic for making decisions under uncertainty.

The relevant situation is if you have two things that still exist, and you want to guess which one will last longer. This doesn't override other facts about the domain -- it's SOME information in the absence of any other knowledge. It's for poker players, not scientists.

The example I gave was the new cloud blog startup vs. the old hosting platform. Which one would you put your blog on if you wanted people to read it 5 years from now? All other things being equal, I'd take the old hosting platform. But if you think the startup has a really good business model or you like the founders, maybe you choose that one. It's just common sense.

Another example might be cold-blooded crocodiles vs. a warm-blooded mouse as an example. All things being equal, you could guess that the crocodiles will survive further into the future, since they were here first. But someone with a specific theory or expertise could also argue that the bigger animal is less likely to survive, etc.

Similarly, we already know that dinosaurs are gone. This isn't a situation where you need to act or make a prediction.

Someone who understands Bayesian statistics can probably explain it better than me, but you have to take into account existing knowledge, and update it with new knowledge. Picking out something that you know is obsolete isn't relevant.

It's a heuristic. Although I bet you can find those things in use somewhere.

This article is a little different -- "things my son would use" implies that they're still popular, not just extant. Both questions are interesting, and influenced by the same principles.

The Lindy effect is one reason I'm working on https://www.oilshell.org/, because shell is now more than 50 year old, much older than Python/JS/Ruby, etc.

Concrete example from the last few days: https://news.ycombinator.com/item?id=26746280

i.e. When people want to explain a modern cloud platform, they use shell. Go would have been more obvious, but shell is clearer. Lindy prediction: shell will outlive Go :)

I think you can argue that its not true for things in which item A and item B are members of a broader class of things wherein changing from A -> B incurs no or trivial costs or little fundamental changes in fulfilling the purpose of the class of items and there exists no immediate need to stop using A.

For example nobody expects the 2004 Toyota Carolla to be forever but the gas powered car will be far harder to kill.

I run a retro games business.

Trinitrons are still to this day the best tech for displaying old games, and people will pay over $100 for even a consumer-grade unit without SCART.

For the professional grade PVM and BVM monitors, people will pay thousands for a large (21"+) one and they sell like hotcakes.

Big CRTs are pretty much gone as far as I know, but 1-micron ICs are all over the place. They're just not interesting any more.

> The rough heuristic is: the longer a technology has been around, the longer it will last into the future.

"The term Lindy refers to Lindy's delicatessen in New York, where comedians "foregather every night at Lindy's, where ... they conduct post-mortems on recent show business 'action'"."

And no more should be read into that. There are solutions to problems which are adequate, e.g. "chair", where further changes can be expected to be modest. And since the problem isn't going away (unless someday we're told that sitting kills us and that we need to stand or lay instead), the solution won't either.

Otoh, there are technologies which simply supersede and obsolete others. E.g. UTF8 has ASCII as subset and hence I don't expect to see the latter around for long.

That example proves the point. If UTF-8 exists then ASCII will exist.

It could have gone the other way: if UTF-16 was the ONLY encoding, then ASCII would be obsolete. But that didn't happen.

UTF-8 is backwards compatible with ASCII “as she is spake” but not strictly speaking ASCII as any ASCII control characters will break UTF-8. It also breaks any 8-but extensions/code pages. ASCII vs HTML is a bad example though because HTML is used globally and although ASCII is too this is more a historical artefact. It’s not hard to imagine ASCII dying out over the next few years while HTML continues to adapt to every encoding under the sun and pure ASCII becomes used less and less ...

The C1 block isn't ASCII. UTF8 is a perfect superset of 7-bit ASCII.

Nope. If you read an ASCII file with control characters in Java you’ll get an exception. Also it won’t work with the 8-bit ASCII variants. Neither are “true Scotsmen” of course, but the point still stands that HTML could yet be more durable.

I think you’re confusing Unicode and utf8. java uses Unicode but not utf8; it uses a 2 byte encoding with surrogate pairs by default.

ASCII is utf8, but it’s not utf16. ASCII will be around for as long as utf8 is.

I’m almost certain the default encoding for reading/writing files in Java is UTF-8 and similarly for the source files. I don’t think I encounter wide char data much really at all day to day ...

> Will HTML or JavaScript last longer? Probably HTML, since it came first.

Well, barely. I’d bet on JavaScript for this one; programming languages are almost immortal once they reach the popularity of JS, while HTML would be easier to replace.

Is ASCII really stîll âlîve ?

ASCII is still alive in UTF-8 and other extended encoding systems. :)

I wonder if there is any systems using 7-bit ascii in production... Or extended code pages...

EBCDIC is still in use in production.

Base64 for the win....

Technologies never truly die due to outdatedness they just become rare. Horse buggies are still used, though rarely, steam engines are more hobbies now but still around.

What truly causes a technology to die is when it's kept a secret and all those who know the secret pass away without passing it on (Greek fire, ancient Babylonian batteries, Roman stainless steel). It's the one good argument for having a patent system, to keep that knowledge from being lost.

Land lines are mostly gone where I live, Sweden. I don't really know anyone who uses then, of course they do still exist due to some alarm systems and so on.

As for ripping out the physical service entirely, I don't have any numbers but I saw the lists from 2020 and 2021 and they are huge. So it is clearly happening. And they have been doing this for quite a few years.

Where I live (the Netherlands) all landlines (copper) have been replaced by fiber, I still call them landlines. I guess OP does too?

If you mean landlines for Phones: The fiber can handle that, and does, but people are indeed dropping their telephone-number-for-a-house subscription in favor of individual cell phones (and my Phone is generally on WLAN-call when I'm at home, so it use the fiber as well). My parents in law were the last ones I knew with a house-based phone number and they dropped it this year. That said, I think you still get a house-telephone-number for free with many internet subscriptions and the ISP's modems have ports for Phones on them, so indeed the landlines as defined still exists, but it's a matter of definition.

Where I live (the Netherlands) all landlines (copper) have been replaced by fiber

Do you have a source for this? All I can find is this quote [0], which is a long way off from "all landlines":

> nearly 1 in 2 homes in the 25 largest cities will have KPN FTTH access by the end of 2021.

[0] https://glasvezelnieuws.blogspot.com/2021/04/kpn-op-schema-m...

Maybe I should have specified, "in my city", indeed, when the concentration of people drops, people are still on copper or on satellite or other wireless alternatives. The point is, the country is investing in fiber: Landlines. Allthough this may change when we blanked the country with 5G towers. Already I have 60/60 mbit in my home via 4G, approaching my current 100/100 fiber subscription. Not sure which is going to scale better in the future.

I guess time averages the satisfaction humans have around a thing. Fads come and go and attracts towards new sensations but over time.. that old thing might be the only one that has the right blend.

Let’s apply this to the future: so just like fiber came before 5G so we are going to lay way more fiber after 5G fizzles out.

Is it also true for copper, though?

Yes, because that's how to transmit electricity.

Even for local network cabling, PoE is working to keep copper around.

A sextant (1731) vs. GPS (1973, fully operational 1993). Most large ships still have sextants as backup.

And the US Navy, after a hiatus has started training officers and crew on how to use a sextant since modern warfare will probably result in GPS being either jammed to oblivion or shot down.

Calculus will die before algebra, will die before geometry.

> The rough heuristic is: the longer a technology has been around, the longer it will last into the future.

This doesn't make any sense because it maps trivially to "any technology still around will last forever."

When can we expect horse-drawn carriages to replace automobiles again, then?

It's a heuristic. A single counterargument does not disprove it. It just shows that it's not a law, which was obvious to begin with.

When? If we experience a civilisation collapse.

It's the same thing. horse-drawn carriages are automobile. Today's car is a more advanced version. In a sense, today's laptop is a spaceship comparing to the first laptop.

Technically, a car is a spaceship compared to a carriage.. one Tesla in particular.

Now I’m wondering if we should differentiate between “spaceships” and “space boats” by whether or not they hold air…

Reply on news.ycombinator.com