4 ms·

My response to "No Code" solutions: If there was a faster and easier way to develop software, we would be doing it. We are already using the easiest to underst

My response to "No Code" solutions: If there was a faster and easier way to develop software, we would be doing it. We are already using the easiest to understand tools we can find to develop the best software possible. There is no secret programmer cabal where we are holding hostage the REAL tools we use to develop software, and we keep them secret and hidden while we fuck around every day. For someone to develop a "no-code" solution that fits the huge variety of tasks developers are doing every day, they would be a genius in the same vein as Turing. They would be billionaires.

coffee 7y ago
> If there was a faster and easier way to develop software, we would be doing it. We are already using the easiest to understand tools we can find to develop the best software possible. Not really. As we evolve software development, historically, we've done this through adding new layers of abstraction. Assembly --> C --> Interpreted Code (Java, Ruby, etc...) Why isn't the next layer of abstraction simply what is being termed as "No Code" today?
- Bartweiss 7y ago
  > Why isn't the next layer of abstraction simply what is being termed as "No Code" today? Cynically? Because people were also selling "no code" ten years ago, or even further back. I know that sort of inductive argument doesn't actually work. We went through a dozen nonsense models of infections, but germ theory was actually correct in the end. AI has seen frauds and baseless hype since the mechanical Turk, but there still came a point when computers won at chess and Jeopardy. But it's a pretty good warning sign. The gradual progression of abstractions still caps out somewhere around Python and Rails. We have drag-and-drop website makers, and visual tools for "plumbing" ML flows backed by pandas, or engineering calculations backed by Matlab. They mostly work, and people who know the underlying tools sometimes use them for speed while mixing in actual code as needed. Meanwhile, promises of codeless business apps don't seem rooted in any of those advances, and continue on from their dysfunctional predecessors. And people actually making business apps in a hurry pull from NPM or PyPI without ever touching "no code" tools. It seems like we might get much closer to that layer within a decade or so, but I don't think we're there yet.
  - coffee 7y ago
    > Cynically? No, not cynically :-) > Because people were also selling "no code" ten years ago, or even further back. Not really sure it matters when it starts. Sometimes markets aren't ready and products are too early. What matters is are these solutions providing value to people today, and is it enough value for them to buy/spend time on/etc... Seems there is enough value today, with a mature enough market, based on the results these platforms are experiencing. > I know that sort of inductive argument doesn't actually work. Then why waste time making it in the first place? :-)
- goatlover 7y ago
  Smalltalk was that next level, but for whatever reasons, the programming community never fully embraced that approach, although some modern environments and IDEs come close.
  - coffee 7y ago
    It doesn't matter. I'm not arguing the chronological order in which layers of abstraction have been created and when, just that this is fact. And if this is fact, why not this as the next layer?
jakelazaroff 7y ago
Conversely, we have not in 2020 reached the peak of software development. Presumably there are faster and easier ways to develop software that haven’t yet been invented. Of course, it’s probably more likely that a no-code solution like you’re describing would be the result of decades of iterative development rather than one genius project. Put that way, it seems like we’re heading in the right direction, no?
Bartweiss 7y ago
It's a bit like the conspiracy that pharma companies suppress cures because ongoing treatments make more money, and it has the same main flaws. First, without a huge amount of collusion, somebody would release the cure to beat everyone else's treatment. (Which totally does happen.) And in software, people constantly build and release tools for free while decrying paid software as evil; if neither Stallman nor Oracle is offering something, you can't bet they don't have it. Second, making these things is hard. If you talk to programmers, we link to Programming Sucks and complain about how bad our tools are and how most software is terrible. If you talk to biochemists, you'll hear all about FDA hurdles, the rising cost and difficulty of finding new drugs worth bringing to market, and how they want to find real breakthroughs but companies can't afford speculative research. So it can't really be the case that people are just casually choosing not to produce these things. Maybe there's a vast conspiracy that nobody's leaking, even when they're drunk and complaining about work at 2AM. But if it's not that, it has to be that we just don't know how to make this stuff happen.
- TuringNYC 7y ago
  >> First, without a huge amount of collusion, somebody would release the cure to beat everyone else's treatment. Unless the same party owned the treatment and cure. Dont you see this everyday with x-AAS subscriptions? I mean, they could sell you the software once, but instead it is a service. I've already paid for MS Office three times over because of the service cost

coffee 7y ago

> If there was a faster and easier way to develop software, we would be doing it. We are already using the easiest to understand tools we can find to develop the best software possible. Not really. As we evolve software development, historically, we've done this through adding new layers of abstraction. Assembly --> C --> Interpreted Code (Java, Ruby, etc...) Why isn't the next layer of abstraction simply what is being termed as "No Code" today?

Bartweiss 7y ago
> Why isn't the next layer of abstraction simply what is being termed as "No Code" today? Cynically? Because people were also selling "no code" ten years ago, or even further back. I know that sort of inductive argument doesn't actually work. We went through a dozen nonsense models of infections, but germ theory was actually correct in the end. AI has seen frauds and baseless hype since the mechanical Turk, but there still came a point when computers won at chess and Jeopardy. But it's a pretty good warning sign. The gradual progression of abstractions still caps out somewhere around Python and Rails. We have drag-and-drop website makers, and visual tools for "plumbing" ML flows backed by pandas, or engineering calculations backed by Matlab. They mostly work, and people who know the underlying tools sometimes use them for speed while mixing in actual code as needed. Meanwhile, promises of codeless business apps don't seem rooted in any of those advances, and continue on from their dysfunctional predecessors. And people actually making business apps in a hurry pull from NPM or PyPI without ever touching "no code" tools. It seems like we might get much closer to that layer within a decade or so, but I don't think we're there yet.
- coffee 7y ago
  > Cynically? No, not cynically :-) > Because people were also selling "no code" ten years ago, or even further back. Not really sure it matters when it starts. Sometimes markets aren't ready and products are too early. What matters is are these solutions providing value to people today, and is it enough value for them to buy/spend time on/etc... Seems there is enough value today, with a mature enough market, based on the results these platforms are experiencing. > I know that sort of inductive argument doesn't actually work. Then why waste time making it in the first place? :-)
goatlover 7y ago
Smalltalk was that next level, but for whatever reasons, the programming community never fully embraced that approach, although some modern environments and IDEs come close.
- coffee 7y ago
  It doesn't matter. I'm not arguing the chronological order in which layers of abstraction have been created and when, just that this is fact. And if this is fact, why not this as the next layer?

jakelazaroff 7y ago

Conversely, we have not in 2020 reached the peak of software development. Presumably there are faster and easier ways to develop software that haven’t yet been invented. Of course, it’s probably more likely that a no-code solution like you’re describing would be the result of decades of iterative development rather than one genius project. Put that way, it seems like we’re heading in the right direction, no?

Bartweiss 7y ago

It's a bit like the conspiracy that pharma companies suppress cures because ongoing treatments make more money, and it has the same main flaws. First, without a huge amount of collusion, somebody would release the cure to beat everyone else's treatment. (Which totally does happen.) And in software, people constantly build and release tools for free while decrying paid software as evil; if neither Stallman nor Oracle is offering something, you can't bet they don't have it. Second, making these things is hard. If you talk to programmers, we link to Programming Sucks and complain about how bad our tools are and how most software is terrible. If you talk to biochemists, you'll hear all about FDA hurdles, the rising cost and difficulty of finding new drugs worth bringing to market, and how they want to find real breakthroughs but companies can't afford speculative research. So it can't really be the case that people are just casually choosing not to produce these things. Maybe there's a vast conspiracy that nobody's leaking, even when they're drunk and complaining about work at 2AM. But if it's not that, it has to be that we just don't know how to make this stuff happen.

TuringNYC 7y ago
>> First, without a huge amount of collusion, somebody would release the cure to beat everyone else's treatment. Unless the same party owned the treatment and cure. Dont you see this everyday with x-AAS subscriptions? I mean, they could sell you the software once, but instead it is a service. I've already paid for MS Office three times over because of the service cost