5 ms·

The problem is with the idea of "continuous delivery". Many people fail to understand that technological advances only increase productivity if the innovations,

by _qulr 6y ago

The problem is with the idea of "continuous delivery". Many people fail to understand that technological advances only increase productivity if the innovations, the great leaps forward, are relatively rare, with long periods of stability and refinement in between. There's always an adjustment period, where people have to spend time learning a new technology, and any issues with the new technology need to be resolved. The gains in productivity happen mainly after the adjustment period. But we've eliminated the periods of stability and are constantly pushing for more "innovation", which means we're in constant periods of adjustment and resolving problems, where the promise of increased productivity is never fully met. The worst idea ever in technology is regularly scheduled updates. Innovation has never and will never happen on a schedule. This is simply greed-driven, promotion-driven, pointy-haired-boss-driven development. Produce something new and great... but then let us all enjoy the new thing for a while. Novelty for its own sake is not productive.

leetcrew 6y ago
> The worst idea ever in technology is regularly scheduled updates. Innovation has never and will never happen on a schedule. This is simply greed-driven, promotion-driven, pointy-haired-boss-driven development. this is sort of uncharitable. the development/maintenance cycle for software is incompatible with the traditional way of monetizing a product (ie, design it up front, manufacture at scale, and then the buyer gets what they get, barring severe safety defects). buyers of software expect the product to at least mostly work in the first place, but they also expect bugs to continue to be fixed after the sale, even if the bugs are introduced through unforeseeable interactions with other software. imo, subscriptions are actually the ideal way of aligning incentives for products that involve ongoing maintenance. but buyers tend to consider this a ripoff if they don't actually see a stream of new features in development. while it introduces some unfortunate constraints in the dev cycle, bundling up features in a scheduled update is a good way to make it visible to users that their subscription dollars aren't just falling into a black hole. trickling out new features "when they're done" earns the respect of engineers, but results in the average user simply not noticing that progress is being made.
- _qulr 6y ago
  > they also expect bugs to continue to be fixed after the sale I wasn't disagreeing with that. I mentioned "long periods of stability and refinement" — refinement including bug fixes — and "any issues with the new technology need to be resolved". But again, bug fixes don't magically happen on a schedule either. Maybe fixes are easy, maybe they're hard, you never know in advance. > bundling up features in a scheduled update is a good way to make it visible to users that their subscription dollars aren't just falling into a black hole This is exactly why it's not true that "subscriptions are actually the ideal way of aligning incentives for products that involve ongoing maintenance". Instead of maintenance, subscriptions incentivize continuous delivery of new features, and consequently continuous delivery of new bugs.
- TeMPOraL 6y ago
  > buyers of software expect the product to at least mostly work in the first place, but they also expect bugs to continue to be fixed after the sale, Contrast that to traditional physical goods, where buyers expect the product to work as advertised, right out of the box, or their money back. Software in the Internet era has it easy, because it gets to release shitty half-finished versions, and then keep charging money while never quite finishing the software. > subscriptions are actually the ideal way of aligning incentives for products that involve ongoing maintenance. but buyers tend to consider this a ripoff if they don't actually see a stream of new features in development Because software does not decay on its own (despite the misleading term "bitrot" being popular in tech circles). That's literally why digitizing data has taken the world by storm: digital data does not decay; as long as the physical medium is readable, you can make a perfect copy of the data it contains. As a buyer, I don't expect my software to need maintenance. I expect it to work out of the box (just like I expect every physical product to work out of the box), and once I find software that fulfills my needs, I expect it to work all the way until computing technology moves forward so much that it's no longer possible to run the software. Which, in the era of virtual machines, may take decades. So yeah, there's a need to clearly justify to the customers why you're charging subscription, because software in its natural state does not need maintenance.
  - leetcrew 6y ago
rch 6y ago
Developers get paid every couple of weeks, so an agile business tries to track the value gained from that expense on a similar cadence. The process has become more about corporate governance than delivering customer value.
bena 6y ago
This criticism is based on the assumption that all updates are created equal. But they aren't. One week, the update could be a relatively minor bug fix. The next week, a major feature upgrade that's been in the pipeline for months. You also remove the ambiguity of "Is this worth pushing out? When should I push this out? Should I do some more fixes or push this one out first?". You got fixes, push them out in the next update. Your criticism also assumes a small team. If you have a large enough team where you can split them into new feature development and current bug fixing, they're going to work at different rates and be ready at different times. If instead your entire team just works on "the product", then there is no effective difference between fixing issues and creating functionality.
com2kid 6y ago
> Innovation has never and will never happen on a schedule. This is simply greed-driven, promotion-driven, pointy-haired-boss-driven development. For consumer goods, it is the hype cycle. New updates and releases get press. They also restore consumer confidence. If Samsung announced that next year they weren't releasing a new Galaxy phone, the entire industry would freak out. Consumers would lose confidence in buying Samsung phones, journalists would write articles questioning if Samsung was pulling out of the market, a lot of bad things would happen. Give it 18 months without a release and people would start to think of Samsung as "that company that used to make phones." They would have to fight like heck to restore their image. Software is the same way. In the Vista/7/8 era, Microsoft looked like they were falling behind because their competitors started releasing yearly, or even bi-yearly, feature updates. Sure every Android version up until 7 was kinda-sorta-terrible, but it kept Android in the news. Likewise, Apple got huge free press every time they announced a new revision of OS X (now MacOS), and every time they came out an announced a new version of iOS. The result? "The desktop is dying, phones are where the real innovation is at!" articles being published even faster than those software updates came out. You can of course release too fast, rarely do Chrome or Firefox's releases get any press (unless there is a controversial UI change), but in general frequent updates are free advertising. Tesla is also great at this, I'm nowhere near being in the market for a Tesla, but at least a couple times a year I still end up hearing about software features they are rolling out!
- _qulr 6y ago
  > If Samsung announced that next year they weren't releasing a new Galaxy phone, the entire industry would freak out. The smartphone industry has only themselves to blame for setting up this expectation. But it is possible to get off the train. I remember when Apple announced they were dropping out of the annual MacWorld San Francisco conference, because they didn't want to constrain their product release cycle. Apple survived that just fine. > In the Vista/7/8 era, Microsoft looked like they were falling behind because their competitors started releasing yearly, or even bi-yearly, feature updates. Competitors? Windows and Mac had near 100% market share on desktop. There was only 1 competitor. Vista was released in January 2007, but Mac OS X 10.5 Leopard was infamously delayed until October 2007 because of iPhone, so this telling of history doesn't seem entirely accurate. Moreover, Mac OS X releases were already slowing. Here's a list of months since the previous major .0 release: 10.1.0 6 10.2.0 11 10.3.0 14 10.4.0 18 10.5.0 18 10.6.0 22 10.7.0 23 10.8.0 12 10.9.0 15 10.10.0 12 10.11.0 11 10.12.0 12 10.13.0 12 10.14.0 12 10.15.0 12 Thus, major Mac OS releases were slowing down year by year — which is totally sensible — but then Steve Jobs died after 10.7 was released in 2011, and only then did they switch to a yearly schedule.

by _qulr 6y ago

leetcrew 6y ago

> The worst idea ever in technology is regularly scheduled updates. Innovation has never and will never happen on a schedule. This is simply greed-driven, promotion-driven, pointy-haired-boss-driven development. this is sort of uncharitable. the development/maintenance cycle for software is incompatible with the traditional way of monetizing a product (ie, design it up front, manufacture at scale, and then the buyer gets what they get, barring severe safety defects). buyers of software expect the product to at least mostly work in the first place, but they also expect bugs to continue to be fixed after the sale, even if the bugs are introduced through unforeseeable interactions with other software. imo, subscriptions are actually the ideal way of aligning incentives for products that involve ongoing maintenance. but buyers tend to consider this a ripoff if they don't actually see a stream of new features in development. while it introduces some unfortunate constraints in the dev cycle, bundling up features in a scheduled update is a good way to make it visible to users that their subscription dollars aren't just falling into a black hole. trickling out new features "when they're done" earns the respect of engineers, but results in the average user simply not noticing that progress is being made.

_qulr 6y ago
> they also expect bugs to continue to be fixed after the sale I wasn't disagreeing with that. I mentioned "long periods of stability and refinement" — refinement including bug fixes — and "any issues with the new technology need to be resolved". But again, bug fixes don't magically happen on a schedule either. Maybe fixes are easy, maybe they're hard, you never know in advance. > bundling up features in a scheduled update is a good way to make it visible to users that their subscription dollars aren't just falling into a black hole This is exactly why it's not true that "subscriptions are actually the ideal way of aligning incentives for products that involve ongoing maintenance". Instead of maintenance, subscriptions incentivize continuous delivery of new features, and consequently continuous delivery of new bugs.
TeMPOraL 6y ago
> buyers of software expect the product to at least mostly work in the first place, but they also expect bugs to continue to be fixed after the sale, Contrast that to traditional physical goods, where buyers expect the product to work as advertised, right out of the box, or their money back. Software in the Internet era has it easy, because it gets to release shitty half-finished versions, and then keep charging money while never quite finishing the software. > subscriptions are actually the ideal way of aligning incentives for products that involve ongoing maintenance. but buyers tend to consider this a ripoff if they don't actually see a stream of new features in development Because software does not decay on its own (despite the misleading term "bitrot" being popular in tech circles). That's literally why digitizing data has taken the world by storm: digital data does not decay; as long as the physical medium is readable, you can make a perfect copy of the data it contains. As a buyer, I don't expect my software to need maintenance. I expect it to work out of the box (just like I expect every physical product to work out of the box), and once I find software that fulfills my needs, I expect it to work all the way until computing technology moves forward so much that it's no longer possible to run the software. Which, in the era of virtual machines, may take decades. So yeah, there's a need to clearly justify to the customers why you're charging subscription, because software in its natural state does not need maintenance.
- leetcrew 6y ago

rch 6y ago

Developers get paid every couple of weeks, so an agile business tries to track the value gained from that expense on a similar cadence. The process has become more about corporate governance than delivering customer value.

bena 6y ago

This criticism is based on the assumption that all updates are created equal. But they aren't. One week, the update could be a relatively minor bug fix. The next week, a major feature upgrade that's been in the pipeline for months. You also remove the ambiguity of "Is this worth pushing out? When should I push this out? Should I do some more fixes or push this one out first?". You got fixes, push them out in the next update. Your criticism also assumes a small team. If you have a large enough team where you can split them into new feature development and current bug fixing, they're going to work at different rates and be ready at different times. If instead your entire team just works on "the product", then there is no effective difference between fixing issues and creating functionality.

com2kid 6y ago

> Innovation has never and will never happen on a schedule. This is simply greed-driven, promotion-driven, pointy-haired-boss-driven development. For consumer goods, it is the hype cycle. New updates and releases get press. They also restore consumer confidence. If Samsung announced that next year they weren't releasing a new Galaxy phone, the entire industry would freak out. Consumers would lose confidence in buying Samsung phones, journalists would write articles questioning if Samsung was pulling out of the market, a lot of bad things would happen. Give it 18 months without a release and people would start to think of Samsung as "that company that used to make phones." They would have to fight like heck to restore their image. Software is the same way. In the Vista/7/8 era, Microsoft looked like they were falling behind because their competitors started releasing yearly, or even bi-yearly, feature updates. Sure every Android version up until 7 was kinda-sorta-terrible, but it kept Android in the news. Likewise, Apple got huge free press every time they announced a new revision of OS X (now MacOS), and every time they came out an announced a new version of iOS. The result? "The desktop is dying, phones are where the real innovation is at!" articles being published even faster than those software updates came out. You can of course release too fast, rarely do Chrome or Firefox's releases get any press (unless there is a controversial UI change), but in general frequent updates are free advertising. Tesla is also great at this, I'm nowhere near being in the market for a Tesla, but at least a couple times a year I still end up hearing about software features they are rolling out!

_qulr 6y ago
> If Samsung announced that next year they weren't releasing a new Galaxy phone, the entire industry would freak out. The smartphone industry has only themselves to blame for setting up this expectation. But it is possible to get off the train. I remember when Apple announced they were dropping out of the annual MacWorld San Francisco conference, because they didn't want to constrain their product release cycle. Apple survived that just fine. > In the Vista/7/8 era, Microsoft looked like they were falling behind because their competitors started releasing yearly, or even bi-yearly, feature updates. Competitors? Windows and Mac had near 100% market share on desktop. There was only 1 competitor. Vista was released in January 2007, but Mac OS X 10.5 Leopard was infamously delayed until October 2007 because of iPhone, so this telling of history doesn't seem entirely accurate. Moreover, Mac OS X releases were already slowing. Here's a list of months since the previous major .0 release: 10.1.0 6 10.2.0 11 10.3.0 14 10.4.0 18 10.5.0 18 10.6.0 22 10.7.0 23 10.8.0 12 10.9.0 15 10.10.0 12 10.11.0 11 10.12.0 12 10.13.0 12 10.14.0 12 10.15.0 12 Thus, major Mac OS releases were slowing down year by year — which is totally sensible — but then Steve Jobs died after 10.7 was released in 2011, and only then did they switch to a yearly schedule.