Sometimes I feel his sentiment, like everytime I open up something (that seemingly) is identical to each and every version. Say the FileZilla app, or VLC. (But there are probably bug fixes and new protocols or file formats getting support for each update.) But for Word, come on. There's a ton of new features added through the years. I mean I use none of them, but I suppose people have been asking for database-like features, or simple web layout stuff. As long as its not a subscription model only - just don't buy the new version guy!
Yeah. Even in my graduate math studies I sometimes used Word. Obviously LaTeX is more powerful and expressive, but Word equations are quick and easy for simple tasks.
Some people want ABC, others want ABD, and yet others need ACEFG. Building and maintaining a different app for each of these needs is not only inefficient for the developer but also inconvenient for the user who now has to juggle several different apps. So people just settle on shipping a bloated app that can do everything from A to Z.
Clearly he has no idea how much ancient software remains in use because people have no interest in upgrading. My dentist uses some software to manage her practice - keeps patient records and x-rays and such. It runs in Windows XP. She runs it on a newer Mac in Parallels.
If I bought a new Mac today I wouldn’t buy an M1. Too much old Unix software hasn’t been ported yet. I’d find one of the leftover Intels that are still available.
My Macs run the older MacOS release. Corps run RHEL for a reason. The list goes on and on. No, you don’t have to upgrade anything.
The obvious answer is technological development, though I think it is a bit more subtle than that.
It seems like there is an evolutionary cycle going on: Software becomes too complex, is being abandoned in favour of simpler products to support specific needs. They do in turn also become bloated and complex over time.
My latest think on this is something that has been discussed here earlier in various ways. I think to is because of forced growth. However, I think a lot of projects would benefit from a fixed scope and then doing maintenance really well.
It's the competitive market environment which forces companies to release new stuff in order to stay relevant, coupled with the mental conditioning of most people that newer is almost always better. That's why we see platforms and frameworks and clouds carrying a whole ecosystem around them, instead of sharply-focused tools which get a specific job done.
A sharply-focused tool tackles a specific problem that exists at a certain time, in a certain domain. Once the tool solves the problem well, there is no reason to change it. Once the original problem is gone, the tool has no reason to exist.
Reminds me of how music developed over time. Opera was originally a reaction against the polyphonic vocal music before it. Opera started out as simple, one voice music which they believed was modelled on "ancient Greek style"; then it became progressively more complex until Wagner, then electric guitars, then folk/acoustic, etc..
Even going into maintenance-only mode will not stop companies from having to find creative ways to collect money from their customers long after the initial purchase.
Maintenance costs money. People don't want to pay a lot of money up front. They also don't want to pay for upgrades or subscriptions if they don't feel like they're getting something useful from time to time. Unfortunately, most people who buy consumer-grade proprietary software don't seem to feel that "being stable and well-maintained" is a useful feature on its own even without any new features. (Enterprise software is a different beast, of course. A very large part of RHEL's value proposition is that it doesn't get any new features, only security fixes, for a decade.)
Today a lot of projects looks like simple apps but uses a ton of abstraction layers (that help them to deliver in no time a complete app, but at the cost of performance and optimizations, I'm looking to you electron apps), also there's a C developer that a while ago did a rant about Visual Studio that at in a Pentium 4, more or less 13-15 years ago, it took like 1 or 2 seconds to launch while today it takes more than 10 seconds to launch in a decent computer that is way faster than before, yet it do a lot of maybe unnecessary things or has been built over too many abstraction layers that makes the software slower and slower.
Upgrade cycle will never end. It is just slowly transitioning into same thing that happens with clothes, which is called fashion. People throw clothes away because new clothes become "fashionable" and because they want to show they can expend effort to follow current trends.
For the same reason a lot of people will not be upgrading. Because they have utilitarian look at old clothes or old computers and if the new computers do not offer anything else than to just differentiate from the old, they just don't see any value in it.
Computers (other than smart phones) aren't really fashion accessories, though. They're tools. Just like you don't see people upgrading their vacuum cleaner every year, I'm sure most people won't be upgrading their computers every year, unless the old ones just don't run some software they need or want to run.
Only a very small proportion of computer owners use these kind of kits. And it's of pretty much every product category that there are a small proportion of users who treat them as a fashion accessory.
Arguably cars are tools too, yet people care a great deal about how they look. Same with wristwatches, or glasses, or a host of other things we use every day.
I don't think this is a valid argument. People do not "upgrade their cars" by buying a different looking one every year - rather, when they buy a new car, looks matter then.
> Same with wristwatches, or glasses, or a host of other things we use every day.
It's hard to consider watches a tool these days and both of these examples are fashion related. Most people do not wear glasses by choice.
But they do upgrade cars because they look old or because they can afford better car, even if the old one is still in good working condition.
Soon all phones will offer basically same functionality with only minor differences. After some point there will be no more gains from more camera or screen pixels, more cores, more gigabytes of storage. The only reason to buy a new phone will be if the old one broke or if you feel to switch to a new phone to follow trends.
The average length of time people own cars is 6 years. Sure, rich people lease every 3, but most people aren't rich. It's still an unfair comparison imo.
> Soon all phones will offer basically same functionality with only minor differences. After some point there will be no more gains from more camera or screen pixels, more cores, more gigabytes of storage. The only reason to buy a new phone will be if the old one broke or if you feel to switch to a new phone to follow trends.
We have essentially been at this point for a few years now. Most people I know who don't have to have the latest and greatest don't want to upgrade - but are forced to because of performance degradation, un-repairability, and bad faith phone contracts.
The parent claims phones are fashion accessories even though they are also useful. I'm sure that means cars are fashion in that view, too. And everything else you mention. They're public displays.
So, that’s you. There’s a lot of people who will upgrade every year, and there’s an obvious signal that they’ve done so - that camera array on the back. People can tell at a glance whether you have the latest (and highest spec) version of the iPhone.
Yes, but with fashion you are not going to find yourself locked into a completely different business model overnight, your old clothes still work and nobody is going to slip in a bunch of malware in your shorts.
I think I was happiest as a developer, as far as how we did software, when I was working at a company that made utility software for DOS, Windows (3 and 95), and Mac back when most computer users did not have internet.
The way people got our software was by buying a box containing a CD-ROM or floppies and a manual at their local Egghead, CompUSA, Best Buy, or similar or ordering it from a mail order place like MacMall, CDW, or similar.
There was no good mechanism to widely distribute updates, so if we wanted to get more money from a customer after that initial sale we had to do it by making the next version of the product compelling enough that they would go the store and buy another box.
Furthermore, our cut on a sale after the store and the distributor had taken their cuts was pretty small. If someone called our toll free support number with a problem that wasn't trivial for support to handle, it would wipe out all that we had made from the sale to that customer and more.
Without internet, malware only got onto a system via physical media. It wasn't like today where almost everything we do involves the internet and a potential attack. If you were happy with your OS and the programs you had, there was no reason you could not continue with them for many years.
To make money in that environment, what management wanted out of us developers was software the did what it box said it would do, reliably and without the need for the user to work around bugs and glitches, with good enough documentation that the user would never call support, and that would leave the user happy enough that when they need software for something else and see a box with that kind of software with our name on it, they would buy it.
If we did a new version of an existing product, management wanted all of that and they wanted the new version to have enough new features so that people with the prior version would be willing to pay full price for the new version.
There was no "ship it...we can fix that with an update later" attitude. It was more "we've got one shot at this...so let's do the best we possibly can".
> There was no good mechanism to widely distribute updates, so if we wanted to get more money from a customer after that initial sale we had to do it by making the next version of the product compelling enough that they would go the store and buy another box.
This is also something the article points out though...
> The problem is that the tech companies have only one big tool to entice you to upgrade each year: piling on new features. More, more features. Microsoft Word was once a word processor. Today it's a database, and a Web-layout program, and a floor wax.
> Eventually these companies have no choice but to add features that nobody asked for. Meanwhile bloated, overwhelming technology has a very real emotional effect on us; we feel like idiots when we can't master it.
Though overall i agree with you that i prefer the way you mention. If nothing else i felt more in control on what goes into my PC.
You can blunt this effect if you list performance as a feature. And reimagined features with fewer corner cases aka bugs. You don’t have to go feature factory if you don’t want to.
Trouble is, too many people want to.
I think it’s Conway’s Law after a fashion. The organization hasn’t set me up to be able to integrate closely with the existing code, so I write new code. Or I put other people in that position because the company ties my worth to old code too tightly so I feel threatened by changes.
This mirrors my own experience perfectly. Stuff I wrote three decades ago is still used unchanged in production in plenty of places. It doesn't need to be upgraded because it was built right the first time around. The fact that people would lose limb or life if I f'd up also really helped to focus and to get management to understand that I simply wouldn't ship if I felt it couldn't be trusted.
I don't recall feeling any higher level of work-related stress than when about to send out the golden master for our Sony PSX (original, CD-based) title. One shot, no patches ever.
Interestingly enough, for those of us mature enough to have experienced the former "ship it" attitude second-hand as consumers know these companies often only really have "one shot," because I'm sure as hell not going to pay to be QA at my age.
Reading articles about products that crash day one tells me all I need to know. I don't consider the product even after it's been patched.
This whole discussion seems to leave out a point that I've been thinking about...
"New features" isn't the ONLY reason to keep making incremental improvements to software. We also need to make adjustments as hardware paradigms shift over time. Hardware improvements are meaningful, and we often SHOULD upgrade software to take advantage of that.
My question is, will hardware ever stop improving? Will we reach a point where an M1-like full-desktop SOC includes enough processing, storage and graphics power that it satisfies any user's desktop/laptop needs? If/when that happens, can software stop evolving in parallel?
And a related thought, when we finally start getting significant amounts of people off the planet (space travel), will all networking software need to undergo a major paradigm shift to manage the extremely-high-latency-low-bitrate situation as a normal case?
> Will we reach a point where an M1-like full-desktop SOC includes enough processing, storage and graphics power that it satisfies any user's desktop/laptop needs?
I doubt it. It used to be that things requiring a cluster of computers 20 years ago now runs on desktops while clusters and HPC systems run even bigger calculations.
With more power, memory, and storage, things will trickle down and new approaches will take their place.
In my research, it would be great if I could use an algorithm scaling at N^5 which gives high accuracy, but it’s too expensive for most N, so pretty much everyone uses an N^3 scaling approach which is less accurate.
I think it has only gotten worse in the sense that now most software vendors want to sell you a subscription that spreads your payments over time and literally never ends as long as you still want to use the software.
I think it has gotten worse since most software vendors want to sell you a subscription to spread your payment out over time. As long as you want to use the software you have to keep paying. In the long run the software will cost you more.
I would’ve thought that by this time, alternative software would replace Word, Excel, and PowerPoint.
That maybe Google would make a good spreadsheet. Or Libre Office would be a contender. But nope.
Libre Office is terrible. Buggy. And just plain janky. And it has no good PowerPoint contender.
I tried to use its word processor, but that was a terrible mess. It made me want to reach out my wallet and drop another $150 to Microsoft for another license. Even though I already bought 3 licenses, for computers that I barely use anymore, but whose license is for that computer.
I gave up and settled for Google’s online office products. It was janky, but it sorta worked. And it works across different operating systems.
But if I have to do any real work, then I’ll have to use Microsoft Office apps again.
In terms of features creep, I think some of them turned to Table Stake. Benedict Evans has a great article on it. [1]
It is not that the Cycle never end. For example no one look forward to upgrade their Microsoft Office or Word anymore. And even without support they will continue to use it. The Author may argue your OS cant last forever to support old Apps. But that is true with everything else. Your TV, Fridge, Clothes etc Nothing we made last forever.
At the same time, Adobe is making some great improvement with each Photoshop with AI and Machine Learning. Shortening the time to do some repetitive task. And while they may not be wroth the money for everyone, professional tends to love it. And there are still some ways to go.
And just happened Microsoft Fox Pro to be on HN Front Page. I know thousands of small companies are still relying on Visual FoxPro. When the 9.0 release was 17 years ago and its patch SP2 was 14 years.
49 comments
[ 3.4 ms ] story [ 107 ms ] threadI stopped reading at that point. I was hoping for some thoughtful analysis, not Andy Rooney.
I dont know what outher mainstream software lets you write an essay and math homework.
If I bought a new Mac today I wouldn’t buy an M1. Too much old Unix software hasn’t been ported yet. I’d find one of the leftover Intels that are still available.
My Macs run the older MacOS release. Corps run RHEL for a reason. The list goes on and on. No, you don’t have to upgrade anything.
It seems like there is an evolutionary cycle going on: Software becomes too complex, is being abandoned in favour of simpler products to support specific needs. They do in turn also become bloated and complex over time.
My latest think on this is something that has been discussed here earlier in various ways. I think to is because of forced growth. However, I think a lot of projects would benefit from a fixed scope and then doing maintenance really well.
A sharply-focused tool tackles a specific problem that exists at a certain time, in a certain domain. Once the tool solves the problem well, there is no reason to change it. Once the original problem is gone, the tool has no reason to exist.
Maintenance costs money. People don't want to pay a lot of money up front. They also don't want to pay for upgrades or subscriptions if they don't feel like they're getting something useful from time to time. Unfortunately, most people who buy consumer-grade proprietary software don't seem to feel that "being stable and well-maintained" is a useful feature on its own even without any new features. (Enterprise software is a different beast, of course. A very large part of RHEL's value proposition is that it doesn't get any new features, only security fixes, for a decade.)
Upgrade cycle will never end. It is just slowly transitioning into same thing that happens with clothes, which is called fashion. People throw clothes away because new clothes become "fashionable" and because they want to show they can expend effort to follow current trends.
For the same reason a lot of people will not be upgrading. Because they have utilitarian look at old clothes or old computers and if the new computers do not offer anything else than to just differentiate from the old, they just don't see any value in it.
Really? Then what is with all the RGB kit?
> Same with wristwatches, or glasses, or a host of other things we use every day.
It's hard to consider watches a tool these days and both of these examples are fashion related. Most people do not wear glasses by choice.
Soon all phones will offer basically same functionality with only minor differences. After some point there will be no more gains from more camera or screen pixels, more cores, more gigabytes of storage. The only reason to buy a new phone will be if the old one broke or if you feel to switch to a new phone to follow trends.
> Soon all phones will offer basically same functionality with only minor differences. After some point there will be no more gains from more camera or screen pixels, more cores, more gigabytes of storage. The only reason to buy a new phone will be if the old one broke or if you feel to switch to a new phone to follow trends.
We have essentially been at this point for a few years now. Most people I know who don't have to have the latest and greatest don't want to upgrade - but are forced to because of performance degradation, un-repairability, and bad faith phone contracts.
That is, the minority of currently owned computing devices.
The way people got our software was by buying a box containing a CD-ROM or floppies and a manual at their local Egghead, CompUSA, Best Buy, or similar or ordering it from a mail order place like MacMall, CDW, or similar.
There was no good mechanism to widely distribute updates, so if we wanted to get more money from a customer after that initial sale we had to do it by making the next version of the product compelling enough that they would go the store and buy another box.
Furthermore, our cut on a sale after the store and the distributor had taken their cuts was pretty small. If someone called our toll free support number with a problem that wasn't trivial for support to handle, it would wipe out all that we had made from the sale to that customer and more.
Without internet, malware only got onto a system via physical media. It wasn't like today where almost everything we do involves the internet and a potential attack. If you were happy with your OS and the programs you had, there was no reason you could not continue with them for many years.
To make money in that environment, what management wanted out of us developers was software the did what it box said it would do, reliably and without the need for the user to work around bugs and glitches, with good enough documentation that the user would never call support, and that would leave the user happy enough that when they need software for something else and see a box with that kind of software with our name on it, they would buy it.
If we did a new version of an existing product, management wanted all of that and they wanted the new version to have enough new features so that people with the prior version would be willing to pay full price for the new version.
There was no "ship it...we can fix that with an update later" attitude. It was more "we've got one shot at this...so let's do the best we possibly can".
This is also something the article points out though...
> The problem is that the tech companies have only one big tool to entice you to upgrade each year: piling on new features. More, more features. Microsoft Word was once a word processor. Today it's a database, and a Web-layout program, and a floor wax.
> Eventually these companies have no choice but to add features that nobody asked for. Meanwhile bloated, overwhelming technology has a very real emotional effect on us; we feel like idiots when we can't master it.
Though overall i agree with you that i prefer the way you mention. If nothing else i felt more in control on what goes into my PC.
Trouble is, too many people want to.
I think it’s Conway’s Law after a fashion. The organization hasn’t set me up to be able to integrate closely with the existing code, so I write new code. Or I put other people in that position because the company ties my worth to old code too tightly so I feel threatened by changes.
Reading articles about products that crash day one tells me all I need to know. I don't consider the product even after it's been patched.
"New features" isn't the ONLY reason to keep making incremental improvements to software. We also need to make adjustments as hardware paradigms shift over time. Hardware improvements are meaningful, and we often SHOULD upgrade software to take advantage of that.
My question is, will hardware ever stop improving? Will we reach a point where an M1-like full-desktop SOC includes enough processing, storage and graphics power that it satisfies any user's desktop/laptop needs? If/when that happens, can software stop evolving in parallel?
And a related thought, when we finally start getting significant amounts of people off the planet (space travel), will all networking software need to undergo a major paradigm shift to manage the extremely-high-latency-low-bitrate situation as a normal case?
I doubt it. It used to be that things requiring a cluster of computers 20 years ago now runs on desktops while clusters and HPC systems run even bigger calculations.
With more power, memory, and storage, things will trickle down and new approaches will take their place.
In my research, it would be great if I could use an algorithm scaling at N^5 which gives high accuracy, but it’s too expensive for most N, so pretty much everyone uses an N^3 scaling approach which is less accurate.
That maybe Google would make a good spreadsheet. Or Libre Office would be a contender. But nope.
Libre Office is terrible. Buggy. And just plain janky. And it has no good PowerPoint contender.
I tried to use its word processor, but that was a terrible mess. It made me want to reach out my wallet and drop another $150 to Microsoft for another license. Even though I already bought 3 licenses, for computers that I barely use anymore, but whose license is for that computer.
I gave up and settled for Google’s online office products. It was janky, but it sorta worked. And it works across different operating systems.
But if I have to do any real work, then I’ll have to use Microsoft Office apps again.
It is not that the Cycle never end. For example no one look forward to upgrade their Microsoft Office or Word anymore. And even without support they will continue to use it. The Author may argue your OS cant last forever to support old Apps. But that is true with everything else. Your TV, Fridge, Clothes etc Nothing we made last forever.
At the same time, Adobe is making some great improvement with each Photoshop with AI and Machine Learning. Shortening the time to do some repetitive task. And while they may not be wroth the money for everyone, professional tends to love it. And there are still some ways to go.
And just happened Microsoft Fox Pro to be on HN Front Page. I know thousands of small companies are still relying on Visual FoxPro. When the 9.0 release was 17 years ago and its patch SP2 was 14 years.
[1] https://www.ben-evans.com/benedictevans/2020/12/21/google-bu...