4 ms·

What I'd be really interested in is how tech churn is perceived by people older than me. I'm only 30 years in, so I tend to defend my generation's choices such

What I'd be really interested in is how tech churn is perceived by people older than me. I'm only 30 years in, so I tend to defend my generation's choices such as POSIX, SQL, XML, SOA, Java, and C/C++ before that (plus special-purpose pet peeves of mine such as markup/SGML and logic programming which came before) though I'm also claiming to be proficient in and generally open towards new tech. I consider most of supposedly new tech as rehashes of things past rather than real progress, and to be just lock-in schemes and sucking in another way compared to the tech that it is supposed to replace. But I'm really uncertain if I'm just falling victim to generational effects, like, say, the proverbial COBOL programmer being ridiculed by younger devs to instinctively grow their own tech career. Still, for me, there's a moment around 2006-2012 when the industry went all-in on "the Cloud" and consumerisation of IT, when before I saw canon, progress, and consensus through standardization. I guess this is something that can't be objectively measured; but I can say that initiatives for standardization have almost dropped to zero compared to the 1990s and 2000s.

scsilver 6y ago
As a young blood, 5 years in, got hired by learning react in the great JS hiring of the mid 2010s, I look back towards those standards bodies as something that would be very beneficial now. All we have is chrome creating standards by monopoly. Maybe I have a older mindset aswell. My training in in Civil Engineering and I have my expectations for standards bodies set pretty high because of it. The code is so high level now, going through algo and data structures I see the benefit of that fundamental knowledge, but also see that you can be a very valuable engineer to a company without it(thanks to the rich developer ecosystem for that) If we think about the maturity of software like a biological ecosystem, maybe the zen garden built by past engineers has overgrown into a dense and varietal forrest. Im not sure if its bad or good, maybe there are more niches to move into.
breck 6y ago
I'm only 15 years in but have thought about this a lot. I agree with you that there is nothing new under the sun; that we just cycle back through old ideas; fashions come and go and come back. However, the environment changes so often old ideas go from being "possible" to "they just work" or "possible but really slow" to "instant". These environmental changes can make orders of magnitude differences, and cause old ideas to suddenly feel very different in practice. When I think of programming languages for example, of the languages you mention, a lot of their design (like matched brackets in XML), were decisions that made a lot of sense when we had monochrome editors, but now with extremely fast and advanced IDEs, we can take old PL ideas from the 50's and design languages with IDEs in mind that are a lot less cryptic and concise.
- tannhaeuser 6y ago
  Good points. Though if you've picked up computing from almost the ground up (soldering, hardware hacking, low-level programming, of which I did only a little however), there's that experience when sitting in front of an overwhelming, notebook-melting IDE where you say to yourself "I don't need all those arbitrary abstractions; I've got a pretty good understanding of what I want to achieve, thank you very much" and realize the cognitive overhead can become a net-negative compared to the perceived problems that modern IDEs are attempting to solve. Matter of taste, of course. As to XML, matched end-element tags (if that's what you mean) actually were a simplification compared to SGML from which XML was derived/subset. In SGML, you can omit/infer end-element tags or can type "</>" to make SGML auto-close the most recent element. I agree the verbose-ness of XML looks especially redundant since it's always the most recent element you have to close anyway whereas SGML (in principle, at least) has overlapping ("concurrent") markup. And I can assure you that around 1998 we had 32bit color monitors and IDEs didn't look all that different from today ;) But back to the topic, I believe a lot of material has simply vanished from the 'net, or isn't accessible through search engines anymore.
  - tharkun__ 6y ago
    IDEs and languages that an IDE can exploit are awesome. Especially if you don't really have such a good idea of a project (yet). It helps architects or contractors tremendously (as examples of people that might jump from one code base to another quite frequently). What I mean is that I personally really like statically typed languages like Java and compile time safety. I even like some of 'verboseness' that people always complain about. I can take a modern IDE and a Java project that hasn't replaced everything with runtime magic yet (Spring comes to mind) and I can simply click my way through things to get the info I need and/or to build a mental model. However, I don't need a finished mental model already just to be effective at every task. Also refactorings that are really braindead simple and that you don't even have to think about are possible, precisely because they're simple and guaranteed to be correct. Contrast that with other languages, such as Javascript, Python, Perl (yeah mentioning that because I loved Perl back when I was doing almost exclusively Perl - with a bit of shell scripting and lots of SQL - at my first 'real' job), where you have to have a mental model already and you have to know certain 'magic' to even be able to search for all the right things. Something that stuck in my head in that regard was AngularJS. I forgot the specifics but some type of identifier was use underscores in one place but dashes in another. How the eff am I supposed to find things easily? I have to know that magic conversion and grep for it specifically. If I come to a FE project written in Angular and have never done Angular, I will not find anything whatsoever and I have zero chance but to learn Angular to do basic things. And coming back to refactorings, even a simple rename can be a pain in the rear to do (and you won't do it and be stuck with really bad naming) because you can't be certain that you found all the right places to change until runtime i.e. your 2 a.m. batch run failed and you have users screaming at you. That teaches you to just stick with bad naming real fast. Good languages and frameworks, if you ask me, allow someone that is senior in his role but a total noob with the specific language or framework to look at existing code and easily make certain modifications. I wrote a simple T-SQL (Sybase, not MS-SQL) parser that output graphviz format to get a call graph of some backend job we had that consisted of a gazillion individual stored procedures strewn across a gazillion files and printed that and hung it up on the wall, just to be able to easily navigate that thing. Queue IDE where you just Ctrl-Click for the same end result.
waldrews 6y ago
Curious how logic programming fits in with the rest of that list?
- tannhaeuser 6y ago
  Why wouldn't it? I've always been a fan of polyglot programming and use of languages oriented towards specific use cases, as opposed to language-centric ecosystems. Now Java doesn't exactly fit that criterion ;( but makes up for it by being portable, relatively open, and cross-platform, and being instrumental in having prevented Windows dominance on the server-side in the 1990s. And it pays the bills. Haven't used it for new personal projects since 2008 or so, though.
twh270 6y ago
I'm 30 years into this industry also. We all have a tendency to defend and use what we're comfortable with, which often is what we learned ages ago. I try to challenge myself regularly by looking for my biases and assumptions. (Easier said than done, of course.) However, I'd say the rate of "tech churn" has become impossible to keep up with in the most popular stacks (Java[Spring], .NET, front-end, Node). If you are in other areas, things tend to move slower. I think the initiatives for standardization move at about the same rate as they used to, but the increasing amount of industry churn makes it seem like standardization has slowed down.
gen220 6y ago
> initiatives for standardization have almost dropped to zero compared to the 1990s and 2000s I think OSS adoption explains this. Standards were all about advocating for common interfaces, even if the implementations were proprietary. Proprietary software was more a thing in the 90s and 00s. Nowadays, common practice is to use open implementations, not just open interfaces. Nowadays, we only see widespread initiatives for standardization when people are trying to cement the long-term survival of their implementation's interface. See, for example, the Open Container Initiative, put forward by Docker (a company that engineers were concerned about betting the barn on, in 2015). Contrast that with S3, which does not have a corresponding commitment to an open interface standard. Their interface has been copied nonetheless by several vendors (DO, Linode), making it a de facto standard, albeit a fragile and autocratic one. Basically, I'm not sure if standardization has evaporated, I think it's diffused, and OSS is (ironically?) a contributing factor.
- dodobirdlord 6y ago
  Before the triumph of OSS standardization required multiple parties to come to agreement and then each produce their own implementation, so implementers had an incentive to fight for their own objectives so that they didn't get backed into a corner of having to do unprofitable work. Lots of bickering about was was and wasn't worth including. To navigate this environment requires committees and formal standards documents. But you know what's a great consolation prize for not getting exactly what you want? Not having to do any work! So when Docker comes along with Docker Containers or Google comes along with Kubernetes, the consolation prize of not having to do any work is a lot larger than the downside of not getting to have any input. Working code is working code, especially if someone reputable has also promised to maintain it for free.
  - gen220 6y ago
    Adding on to your point, the phrase “designed by committee” carries a strong negative connotations for a reason. People may have some gripes with the implementation of some subsystems, but they are free to fork and submit patches for review. This opportunity for grievance remediation did not exist in the old world we’re describing, and contributed to the outcome that standard interfaces were the unholy logical union of each member’s opinions and beliefs. I agree that the benefit of not doing work far outweighs the downsides, for most time horizons (<10 years). If you’re thinking longer term than that, interface standards might be better for you. But few people are.
tharkun__ 6y ago
Very interesting indeed and I had some glimpse of that in my dad, who all his life worked for only one company. And especially as I got older and interested in computers he started talking about some work stuff sometimes. His first programming language at work was 360 assembler. For the uninitiated/too young to remember: https://en.wikipedia.org/wiki/IBM_System/360 https://en.wikipedia.org/wiki/IBM_System/360. When I started working at the same company for my first job and I had to log into 'the host' - meaning the IBM mainframe they had (I believe by that time a Z series https://en.wikipedia.org/wiki/IBM_Z https://en.wikipedia.org/wiki/IBM_Z) I noticed how the custom login screen for their entire system had actually been written by my dad. So all this just as a little context. But when I was talking about SQL like 15+ years ago he would always tell me that sure that's nice and all, but all those joins. They're painfully slow. Why would you do that? They had a table that had exactly all the information that they needed to have and they accessed it via a key. Fast and easy. Need a different view of the data? Make a new table that's organized the way you need it. Done and fast. Ring a bell? (NoSQL :)) I also started playing with Linux and system administration and obviously virtualization (vmware at the time, qemu stuff etc.). So I go talk to my dad about it and how I love the concept and how it does X and Y and Z cool thing. Yeah well, for him that was really old stuff, coz they had that on their mainframe since like forever (now called z/VM, then called VM/370 - in 1972). We had to learn and use Java at university, which my dad always made fun of. Especially with articles like "The state of Java application middleware" and such that I was looking at. All that new fangled stuff, pah! He had been working with IBM CICS as the middleware for forever. Initial release of CICS July 8 1969, latest release June 12, 2020. (https://en.wikipedia.org/wiki/CICS https://en.wikipedia.org/wiki/CICS) I get it. I say the same thing(s) about some of the new fangled stuff I have to work with nowadays. With the kids. Some of it is great. Other things I'm just asked myself why the wheel had to be re-invented just to come back to square one. Like NoSQL databases that add features you'd expect from relational databases.
bananaface 6y ago
I think one source of tech churn people don't talk about is just growth. If you're 5x size you were last year, an entire system rewrite is relatively cheap, and if you expect to grow in the future, you can make riskier bets knowing they'll be cheaper to replace down the line. It's not ipso facto irrational. I'd be interested in seeing a breakdown of the correlations.

scsilver 6y ago

As a young blood, 5 years in, got hired by learning react in the great JS hiring of the mid 2010s, I look back towards those standards bodies as something that would be very beneficial now. All we have is chrome creating standards by monopoly. Maybe I have a older mindset aswell. My training in in Civil Engineering and I have my expectations for standards bodies set pretty high because of it. The code is so high level now, going through algo and data structures I see the benefit of that fundamental knowledge, but also see that you can be a very valuable engineer to a company without it(thanks to the rich developer ecosystem for that) If we think about the maturity of software like a biological ecosystem, maybe the zen garden built by past engineers has overgrown into a dense and varietal forrest. Im not sure if its bad or good, maybe there are more niches to move into.

breck 6y ago

I'm only 15 years in but have thought about this a lot. I agree with you that there is nothing new under the sun; that we just cycle back through old ideas; fashions come and go and come back. However, the environment changes so often old ideas go from being "possible" to "they just work" or "possible but really slow" to "instant". These environmental changes can make orders of magnitude differences, and cause old ideas to suddenly feel very different in practice. When I think of programming languages for example, of the languages you mention, a lot of their design (like matched brackets in XML), were decisions that made a lot of sense when we had monochrome editors, but now with extremely fast and advanced IDEs, we can take old PL ideas from the 50's and design languages with IDEs in mind that are a lot less cryptic and concise.

tannhaeuser 6y ago
Good points. Though if you've picked up computing from almost the ground up (soldering, hardware hacking, low-level programming, of which I did only a little however), there's that experience when sitting in front of an overwhelming, notebook-melting IDE where you say to yourself "I don't need all those arbitrary abstractions; I've got a pretty good understanding of what I want to achieve, thank you very much" and realize the cognitive overhead can become a net-negative compared to the perceived problems that modern IDEs are attempting to solve. Matter of taste, of course. As to XML, matched end-element tags (if that's what you mean) actually were a simplification compared to SGML from which XML was derived/subset. In SGML, you can omit/infer end-element tags or can type "</>" to make SGML auto-close the most recent element. I agree the verbose-ness of XML looks especially redundant since it's always the most recent element you have to close anyway whereas SGML (in principle, at least) has overlapping ("concurrent") markup. And I can assure you that around 1998 we had 32bit color monitors and IDEs didn't look all that different from today ;) But back to the topic, I believe a lot of material has simply vanished from the 'net, or isn't accessible through search engines anymore.
- tharkun__ 6y ago
  IDEs and languages that an IDE can exploit are awesome. Especially if you don't really have such a good idea of a project (yet). It helps architects or contractors tremendously (as examples of people that might jump from one code base to another quite frequently). What I mean is that I personally really like statically typed languages like Java and compile time safety. I even like some of 'verboseness' that people always complain about. I can take a modern IDE and a Java project that hasn't replaced everything with runtime magic yet (Spring comes to mind) and I can simply click my way through things to get the info I need and/or to build a mental model. However, I don't need a finished mental model already just to be effective at every task. Also refactorings that are really braindead simple and that you don't even have to think about are possible, precisely because they're simple and guaranteed to be correct. Contrast that with other languages, such as Javascript, Python, Perl (yeah mentioning that because I loved Perl back when I was doing almost exclusively Perl - with a bit of shell scripting and lots of SQL - at my first 'real' job), where you have to have a mental model already and you have to know certain 'magic' to even be able to search for all the right things. Something that stuck in my head in that regard was AngularJS. I forgot the specifics but some type of identifier was use underscores in one place but dashes in another. How the eff am I supposed to find things easily? I have to know that magic conversion and grep for it specifically. If I come to a FE project written in Angular and have never done Angular, I will not find anything whatsoever and I have zero chance but to learn Angular to do basic things. And coming back to refactorings, even a simple rename can be a pain in the rear to do (and you won't do it and be stuck with really bad naming) because you can't be certain that you found all the right places to change until runtime i.e. your 2 a.m. batch run failed and you have users screaming at you. That teaches you to just stick with bad naming real fast. Good languages and frameworks, if you ask me, allow someone that is senior in his role but a total noob with the specific language or framework to look at existing code and easily make certain modifications. I wrote a simple T-SQL (Sybase, not MS-SQL) parser that output graphviz format to get a call graph of some backend job we had that consisted of a gazillion individual stored procedures strewn across a gazillion files and printed that and hung it up on the wall, just to be able to easily navigate that thing. Queue IDE where you just Ctrl-Click for the same end result.

waldrews 6y ago

Curious how logic programming fits in with the rest of that list?

tannhaeuser 6y ago
Why wouldn't it? I've always been a fan of polyglot programming and use of languages oriented towards specific use cases, as opposed to language-centric ecosystems. Now Java doesn't exactly fit that criterion ;( but makes up for it by being portable, relatively open, and cross-platform, and being instrumental in having prevented Windows dominance on the server-side in the 1990s. And it pays the bills. Haven't used it for new personal projects since 2008 or so, though.

twh270 6y ago

I'm 30 years into this industry also. We all have a tendency to defend and use what we're comfortable with, which often is what we learned ages ago. I try to challenge myself regularly by looking for my biases and assumptions. (Easier said than done, of course.) However, I'd say the rate of "tech churn" has become impossible to keep up with in the most popular stacks (Java[Spring], .NET, front-end, Node). If you are in other areas, things tend to move slower. I think the initiatives for standardization move at about the same rate as they used to, but the increasing amount of industry churn makes it seem like standardization has slowed down.

gen220 6y ago

> initiatives for standardization have almost dropped to zero compared to the 1990s and 2000s I think OSS adoption explains this. Standards were all about advocating for common interfaces, even if the implementations were proprietary. Proprietary software was more a thing in the 90s and 00s. Nowadays, common practice is to use open implementations, not just open interfaces. Nowadays, we only see widespread initiatives for standardization when people are trying to cement the long-term survival of their implementation's interface. See, for example, the Open Container Initiative, put forward by Docker (a company that engineers were concerned about betting the barn on, in 2015). Contrast that with S3, which does not have a corresponding commitment to an open interface standard. Their interface has been copied nonetheless by several vendors (DO, Linode), making it a de facto standard, albeit a fragile and autocratic one. Basically, I'm not sure if standardization has evaporated, I think it's diffused, and OSS is (ironically?) a contributing factor.

dodobirdlord 6y ago
Before the triumph of OSS standardization required multiple parties to come to agreement and then each produce their own implementation, so implementers had an incentive to fight for their own objectives so that they didn't get backed into a corner of having to do unprofitable work. Lots of bickering about was was and wasn't worth including. To navigate this environment requires committees and formal standards documents. But you know what's a great consolation prize for not getting exactly what you want? Not having to do any work! So when Docker comes along with Docker Containers or Google comes along with Kubernetes, the consolation prize of not having to do any work is a lot larger than the downside of not getting to have any input. Working code is working code, especially if someone reputable has also promised to maintain it for free.
- gen220 6y ago
  Adding on to your point, the phrase “designed by committee” carries a strong negative connotations for a reason. People may have some gripes with the implementation of some subsystems, but they are free to fork and submit patches for review. This opportunity for grievance remediation did not exist in the old world we’re describing, and contributed to the outcome that standard interfaces were the unholy logical union of each member’s opinions and beliefs. I agree that the benefit of not doing work far outweighs the downsides, for most time horizons (<10 years). If you’re thinking longer term than that, interface standards might be better for you. But few people are.

tharkun__ 6y ago

Very interesting indeed and I had some glimpse of that in my dad, who all his life worked for only one company. And especially as I got older and interested in computers he started talking about some work stuff sometimes. His first programming language at work was 360 assembler. For the uninitiated/too young to remember: https://en.wikipedia.org/wiki/IBM_System/360 https://en.wikipedia.org/wiki/IBM_System/360. When I started working at the same company for my first job and I had to log into 'the host' - meaning the IBM mainframe they had (I believe by that time a Z series https://en.wikipedia.org/wiki/IBM_Z https://en.wikipedia.org/wiki/IBM_Z) I noticed how the custom login screen for their entire system had actually been written by my dad. So all this just as a little context. But when I was talking about SQL like 15+ years ago he would always tell me that sure that's nice and all, but all those joins. They're painfully slow. Why would you do that? They had a table that had exactly all the information that they needed to have and they accessed it via a key. Fast and easy. Need a different view of the data? Make a new table that's organized the way you need it. Done and fast. Ring a bell? (NoSQL :)) I also started playing with Linux and system administration and obviously virtualization (vmware at the time, qemu stuff etc.). So I go talk to my dad about it and how I love the concept and how it does X and Y and Z cool thing. Yeah well, for him that was really old stuff, coz they had that on their mainframe since like forever (now called z/VM, then called VM/370 - in 1972). We had to learn and use Java at university, which my dad always made fun of. Especially with articles like "The state of Java application middleware" and such that I was looking at. All that new fangled stuff, pah! He had been working with IBM CICS as the middleware for forever. Initial release of CICS July 8 1969, latest release June 12, 2020. (https://en.wikipedia.org/wiki/CICS https://en.wikipedia.org/wiki/CICS) I get it. I say the same thing(s) about some of the new fangled stuff I have to work with nowadays. With the kids. Some of it is great. Other things I'm just asked myself why the wheel had to be re-invented just to come back to square one. Like NoSQL databases that add features you'd expect from relational databases.

bananaface 6y ago

I think one source of tech churn people don't talk about is just growth. If you're 5x size you were last year, an entire system rewrite is relatively cheap, and if you expect to grow in the future, you can make riskier bets knowing they'll be cheaper to replace down the line. It's not ipso facto irrational. I'd be interested in seeing a breakdown of the correlations.