š¾ Archived View for dioskouroi.xyz āŗ thread āŗ 24914062 captured on 2020-10-31 at 00:49:24. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
________________________________________________________________________________
The ultimate no-code tool is Excel. It is the beach to code ocean. Itās about lowering the threshold of results possible to the widest possible group of people. You can have extra modules and plugins and you will always see people who have never dabbled with code before snorkel down to more and more complex functionality. No code done right is a gym - you get users willing to do something, and you enable them to do whatever they are willing to work for. I was one of those users - started with Flash, then trickled into action script, then trigonometry and animation, then JS, PHP. Stereotypes and preconceptions about code are the biggest hurdle for many. All you have to do is let them play and see whose brain naturally carries the load and asks for more. So Id say no code doesnāt replace engineers, but makes more of them.
> No code done right is a gym - you get users willing to do something, and you enable them to do whatever they are willing to work for.
Exactly! Nearly every self-taught programmer started this way. They had something they wanted to achieve and learned enough to get that done. The real opportunities here are in creating tools that do have depth behind the scenes which users can learn gradually as they need to.
No-code is the on-ramp to the yellow brick road - people who have challenges that take them further than that should be able to grapple with the code abstractions behind the wizard's curtain.
Excel is one of the most powerful tools for business ever created and the vast majority of SaaS is providing some specialized behavior of it. That said, Iād hesitate call it the ultimate no-code tool mostly for mundane reasons: no version control, no comments, unconstrained with multiple ways to do something, limited scalability past a certain size or data, no user facing āfront endā (e.g. an online form). Iām therefore very hesitant to say that excel is the best we can do. In fact Iām quite sure thereās a multi hundred billion dollar low code product waiting to be built and no shortage of people trying to figure out what it is.
> no version control, no comments, unconstrained with multiple ways to do something, limited scalability past a certain size or data, no user facing āfront endā
I donāt think you and I are using the same Excel...
I've worked in a Forture 20 company. Literally no one has ever used version control for Excel.
meaningless_roi_calculator.xlsx
meaningless_roi_calculator_v1.xlsx
meaningless_roi_calculator_with_fabricated_hours_saved_v1.xlsx
meaningless_roi_calculator_v2.xlsx
meaningless_roi_calculator_v2.xls
~meaningless_roi_calculator_v2.temp.xls
meaningless_roi_calculator_jerry_bullshit_numbers_v3.xlsx
meaningless_roi_caluclator_v3.xlsx
Which feature have I mischaracterized?
I canāt speak for OP but doesnāt office365 have some sort of versioning mechanism for documents? Through share point or teams or something?
No where near as useful or powerful as Git
Most low code tools I've seen companies drop 6-7 figures on would be worlds better if they had Excel's level of version control.
I've seen low code vendors claim you can do reviews in git... but its an experience not unlike reviewing changes to a generated hibernate or entity framework xml. Its not really reviewable
not having something as powerful as what is generally considered the best is not the same as not having something.
There is a "user facing front-end" as well. UserForms give you customizable dialogs for displaying/prompting the user.
I frankly haven't used vba for a couple decades at this point, but if you dig deep enough its pretty much all there. Combined with sharepoint/etc you can probably get a lot of the UI to be web facing as well.
If there's a good way to version control Excel I'm intrigued.
With sharepoint you can check the file in/out and see version history, possibly with a comment on each versio. But itās not very similar to version control system. Particularly since formulas and data live together. Better than nothing. (As soon as it gets harder to use, people revert to saving files locally and emailing them to each other... so itās about as good as it can be really)
āRevision Trackingā is the built-in tool for this, which is about as good as most no-code solutions Iāve seen.
Thereās a git extension, git-xltrail, if you want to use Excel with git.
I actually built an external version control system for Excel a few years ago. It is actually a hard problem to solve, as you have to figure out when a user cuts and pastes between worksheets to be useful.
> The ultimate no-code tool is Excel.
What? How can you sum the numbers in a row without writing a line of code? I've seen lots of people using Excel for simple and complex tasks, and every single one of them wrote code to accomplish their task.
> How can you sum the numbers in a row without writing a line of code?
Drag your mouse over the cells and press the sum button.
As a bonus, you'll see Excel filling the next cell with a formula for summing, which you can ignore if you don't care, but which provides a seamless opportunity for you to learn a bit of formula language. Step by step, these opportunities will let you become a more proficient Excel user.
Sounds cool. I never used the "buttons" on spreadsheets, but they are definitely a nice learning resource for some people. Still, it's easier to just look at a list of functions described in plain English, than try to interpret the hyeroglyphs that appear on the buttons.
Actually the functions are always shown in the user's language. The formulas are translated even for files created in other languages, which makes it a lot easier for beginners.
By selecting all the cells in the row and clicking the sum button?
I get your point though, many excel users are writing bits of code here and there, especially if you consider math to be code.
A client of mine does web page mockups in Excel.
I had a civil engineer once send me a dense, complicated flowchart as PDF that I thought they had done in Visio but when I asked for the source file it turns out it was "drawn" in Excel.
Iām still on the fence. He claims no code has a niche āconcretizing workflowsā. The problem is vendor lock-in, as well as the myth of the non coding business analyst. Iāve seen enterprises sink millions into BPM software to crank out a few forms a Dev could do in open source code overnight. Inevitably the ābusiness analystsā canāt use the tool either, so now the company gets 300/hr specialist Dev who only knows this bpm tool. Finally they realize the costs and absurdity, but are faced with upgrading or scrapping and rewriting these processes in custom code or the new hot tool.
Concretizing workflow is good, but take care what concrete you use and who pours it. Could end up pouring your company a new pair of boots if you know what I mean.
In general we have started to figure out how to make the DNA of software systems resilient against the changing tides of time. No-code seems to reject a lot of those learnings, for better or worse.
When I was a kid we had Microsoft front page, a no-code solution that allowed you to create a web page. A few years into college my brother-in-law started making money writing web pages in plain CSS and HTML and a little jQuery. I asked him why he could make money when I could make the web page in front page. he said it was much as easier to change the web page over time when it was written in code than changing the auto-generated garbage HTML that front page produced. No code, then, stood up poorly over time back then and I doubt it will today except as a way to make standard parts out of modular pieces of software, sort of like Unix did back in the '80s with its modular command line tools.
> _it was much as easier to change the web page over time when it was written in code than changing the auto-generated garbage HTML that front page produced. No code stood up poorly over time back then and I doubt it will today_
You're in for a rude awakening then!
Webflow (the modern equivalent of Frontpage/dreamweaver) produces portable, semantic HTML and sane CSS. Like half of all YC companies are using it to build their marketing sites now.
I've found it's _much_ faster to completely redesign your site over time with Webflow than with some custom built static site. In fact, I'd say things have flipped. I think whatever esoteric Node Package Mess, shit hot CSS framework and clunky headless CMS you're using today will be the thing that ages poorly over time.
> Webflow (the modern equivalent of Frontpage/dreamweaver) produces portable, semantic HTML and sane CSS.
Thatās assuming you already know how to use HTML and CSS. The tool does _not_ help you learn that. I worked on a project in Webflow with talented designers who didn't know diddly squat about HTML and CSS, and it was a mess. I agree that itās amazing that the tool allowed them to do anything at all, but the output was a big ball of tech debt.
No code isn't going to fix bad project management and organization skills.
I'd say 90% of the companies I've worked at who built their sites "from scratch" also had complex messes of duplicate/unused CSS styles and non-semantic HTML elements.
All this built in by so-called "professional" front-end developers!
Your response almost proves my point. All those YC companies are super young, and are incentived to go into technical debt in the interest of first becoming profitable. They do all sorts of things the industry has decided are bad ideas in general, in the important (but short lived) interest of speed. They are in the experimentation phase, trying to figure out what will work long term while also not being held back by history with a working product.
Matlab SimuLink is a no code solution that works well. It does so because its users, mostly academics, fit a similar profile to YC startups: both are in the business of making stuff that doesn't need to last long/"writing throwaway code". In both cases, experimentation is important. After all, the YC startups don't know what is profitable yet.
Once profitable, they will want to make much more minor changes. After all, why change something that makes money?
It is this field of fine grained changes where code thrives.
I'll be convinced of no code when banks start using it.
I don't know whether that has changed, but in an old startup I worked at(20 years ago) a tool similar to simulink was used to simulate a wireless chip design over it's development life, a few years.
And in simulation you try/change things often.
It seems to work well.
There, change happen over the many system parameters, small blocks(some written in code), or maybe adding a new block(but those systems are dataflow like, and without that many unique blocks, so it's the right abstraction)
You're probably right that the code Webflow produces is decent for maintainability, but the Webflow website (made in Webflow) does not inspire confidence. At 100% zoom on Firefox/MacOS, content does not actually fit the entire width of the page. Even the headers don't stretch from edge to edge, leaving white bars on the side of most pages. Maybe this is what they intended, but it makes it look like the CSS is broken.
it's intended
Isn't Front Page written in _code_ ? Seems like we are just moving code around between common tooling and tooling we generate ourselves? What is "no code" really about? I'm genuinely asking.
From what I understand, "No code" input is almost fully declarative. Examples are Lotus 123, Oracle forms, Excel without macros. There are no loops, no dictionaries, classes are simple objects mapped to a database table. Functions are inbuilt and most you can do is to compose them. The premise of "No code" is that there will come a time where people will simply declare and compose what exists instead of creating some new code. Take from that what you will.
The intuition that declarative programming fits the human brain much better is correct, but I'd hesitate to call this "no code". You're just moving towards a more mathematical code formalism.
> From what I understand, "No code" input is almost fully declarative.
So is pure functional programming, as well as idealized logic programming (actual Prolog falls short of this in important cases), and some forms of relational programming.
> There are no loops
Often there are via recursive/circular references, for which the platform often has bailout logic to prevent infinite loops.
> no dictionaries,
How is an Excel table different from a dictionary?
> How is an Excel table different from a dictionary?
Dictionary is an indexed key-value store, tables are not?
"No code" means the humans is liberated from creating and maintaining code by hand, because there is now a machine doing it. More specifically it means to generate highly abstracted interfaces for common tasks and hiding all the boilerplate and problems from the humans.
Basically it's "decrading" a developer to be a normal user who is just using some software. Which has the benefit that now every non-developer can do the same work as the developer, and they both have the same output and qualification (theoretically).
Thinking about, the endgoal of this is similar to a full self-driving car. You just name the target and the software drives you there.
Yes, yes, and needing to generate less yourself to get things done. That's the pitch at least.
Wordpress seems to be relatively adaptable. And it's very sucsessful, in the field of internet marketing, a very dynamic field.
Especially if you look at it as a low-code solution(instead of no-code): most changes can be done using plugins and visual builders, while some require writing a new plugin.
We both agree that Ma and Pa's and startups alike are well served by WordPress. However, this example also reveals that no/low code is best used when starting out or staying small.
I've heard soooo many horror stories around when businesses start to pick up and they have to take this frankenstein of a WordPress site, with custom code atop poorly supported plugins, and must either support or rewrite it.
I think anyone knowledgeable using Wordpress realizes quickly that a rewrite will be part of the plan once the MVP is proved, unless you're literally just making a blog or some very simple site.
Maybe we should distinguish between "no code" and "no knowledge."
Even Excel can excel (!) at, say, a small statistical problem, but scale it and you want something dedicated.
> sort of like Linux did back in the '80s with its modular command line tools.
I believe you mean Unix, as Linux didn't exist in the 80s.
Right! Editing now.
I find it a bit weird that people are treating all this as a new thing. There have been no-code/visual programming tools for ages, especially in creative industries, e.g MAX & Reaktor (music DSP), Grasshopper (architecture), visual shader editors (most 3d software + game engines), visual game logic scripting (e.g. Unreal), etc...
IMHO in these fields people care about getting results without all the software dev ceremony (which probably would add very little in this context).
Obviously there's a ton of code machinery behind the scenes to make it work, but the same is true Javascript so that's neither here nor there.
That. It makes me confused what no-code actually refers to. Sometimes, they call a software service "no-code" tool, well it is!
> IMHO in these fields people care about getting results without all the software dev ceremony
It's just called "software" you buy from someone, it helps you produce result, even produce another software.
You don't want to code, buy a software.
You don't want to code all the things, use library, framework, services that have API.
You just want to hook a bunch of things together with business idea with zero code. Like hooks up all supply chain? You are not programming anyway you are a manager sort of thing.
What's no-code?
Indeed. Is Wordpress "no code?" I think of it as legos.
Also on the more VFX side, Houdini is super powerful in the way it puts it's visual scripting system/node graph front and center - while also making it really easy to embed "actual" code in the form of VEX (or Python) directly into the node graph.
No-code seems to reject a lot of those learnings, for better or worse. I havenāt seen any no-code company or product that allows source control
Bubble.io offers source control in a somewhat limited sense. Although the utility is quite limited. You can create save-points and revert to them as required.
You can also export versions to json and commit them to a git repo. I do this for fun and some code hacking. It isn't as useful as traditional code maintenance but can come in handy.
Coming to where nocode is useful? Few points based on my experience:
a) We had to create a prototype with a single member techno-business team (me) and few business guys on a very undefined problem. I managed to get it done with Bubble. In fact our business guys tinkered with the layout and design while I was asleep (I gave them not to mess anything up from a functionality though :-))... for e.g. I could spend days/weeks discussing colors, fonts, layouts etc or they could just do it themselves to their visual satisfaction
b) Limited use-cases. I am working on a side project for freelance recruiters. Goal is to help them launch their own indie-business. I can quickly put the system together with visual feedback and perhaps finish it in less than a week during my side-project hours. I think it would take me week to do it the first time. Perhaps 2 days for similar apps from then on.
I plan to supplement it with more involved sophisticated data mining hacks from the backend. Works neatly!
In essence, many businesses are not technologically sophisticated (in the coder sense) and need not be. They need to deliver value in utility, organization, reduced friction, etc.
no-code can shine in these areas. Coders can hack on top of it to try to reduce the gaps.
The ideal mid-way frameworks offer both, a library of patterns that work for internal applications that are almost no-code, and ability to write code where business logic is required. There are some things that are still best expressed as code. I am not sure if a truly "no-code" framework is possible for real-world applications.
We have built Frappe Framework [1] as a platform to re-use code for our massively complex application (ERPNext with 900+ models). While it does not have a drag-and-drop UI builder, object creation and migration is fully automated via configuration. The key reason why this works as low-code is:
1. Automated UI generation (via metadata)
2. Automated database management (schema and migrations)
3. Abstracted functionality that can run on any object (like permission rules, workflows, assignments, file attachments, PDF generation etc)
The configurations are stored as JSON files that can be used for source code management [2]. Frappe Framework is still under the radar, but of late we have seen many teams using it massively for internal tooling.
[1]
[2]
https://github.com/frappe/erpnext/tree/develop/erpnext/selli...
FWIW ...Frappe and ERPNext are the mothership(s) of low-code.
They are useful for a very wide range from indie / small businesses and high tech capital market discount brokerages (#1 ranking).
I have seen my employers waste so much money on sub-standard products. Few (one) of which I hope to get to market using Frappe / ERPNext as an indie venture.
Many thanks @rushabh!
I would add another example of a use case. Access already solved most of the no-code issues. In our SME I wrote a key business application using standard Access forms and a database. There's a few lines of VBA code but that's it. I could of wrote a bespoke app in a modern language which would of taken days and weeks. But access allowed me to get a system working in a few days. We're still using that system years later.
Developers will complain this system isn't scalable and lacks testing, but for a small business or small team that's all that's needed.
Another key problem is that with simple tools you empower non-technical users to do their job faster. If you obfuscate the data behind an application you become dependent on software developers for maintenance which increases cost and reduces resiliency. For large corporates, this can mean removing an army of low-cost administrators and replacing them with an army of high-cost software developers.
When teaching the rudiments of programming to children -- like Scratch -- maybe.
In all cases, once again, "no code" doesn't make the programming go away, it just puts a visual skin on it that makes it prettier to look at but more difficult to manage over a system's lifecycle.
In the 1970s, secretaries at MIT used a version of Emacs -- Multics Emacs, which predated GNU Emacs -- to type up documents; and what's more, they automated common tasks with Emacs Lisp. They were only ever told they were _customizing_ the editor, though, not programming it, so they could approach Lisp without fear.
The representation could be anything -- literally anything at all -- and as long as the end user doesn't believe they are doing "programmers' work" they will use it if it makes life less tedious.
Scratch is a good one. My daughter started using it and I was very excited to see her embrace it. It is a good tool to start thinking logically, understanding flow, and planning out a program...and debugging!
The same people who hate on no-code also often hate when they have to develop some simple CRUD app (which is where no-code excels). I personally think that there will be a huge no-code revolution, because of the simple fact that there are still a lot of manual paper-based poor efficiency processes in businesses.
Without no-code software, the cost to automate a simple business process which takes a low paid admin worker 3 hours per week might be:
So at $500 per day and this incredibly conservative timeline, let's say that solving that simple business process costs $4,000 capex with $1,000 annual costs including infra. This saves 3 hours of work for someone on minimum wage... $2,340 per year.
A few problems:
Ok, so that's fine from a company perspective, but look at it from the minimum wage workers perspective - They might be doing something that they KNOW they themselves could automate, but their IT department won't do it because they deem their time too valuable, and they aren't allowed to do it themselves because they aren't deemed knowledgeable enough.
No-Code shouldn't aim to solve all problems, it should aim to solve the problems that are below IT's threshold but still deliver user value, otherwise there is value left on the table and you have people doing work that could be better automated or put on a better platform.
It's not to replace everything, its to build something that's better than excel and paper, when your IT department can't/won't build it because of capacity/hurdle rates/investment.
Great article!
I have a different point of view.
There are two assumptions I disagree:
Assumption 1: No-code will not benefit from code learnings (or do some equivalent progress)
Probably, because the nature of domain (code vs GUI), the solutions may be different, but it's possible to attack the same problems.
many no-code tools have some type of versioning plus real time sync
many no-code tools have some type of abstraction level extensibility with plugins
...
Assumption 2: Software development is mainly solving complex problems
The other force no-code in the no-code side is, as software industry gets more mature, more patterns are discovered.
Just look how much boilerplate is in each company or unnecessary rework with bad abstractions.
Most of companies doing software today are just rearranging things on screen, or creating some simple pipelines.
Even though RAD, low-code, model driven, etc.. is not a new thing, time has a important role here:
Also as software is getting more popular in last decade, we saw a lot of space for "almost the same" solutions to coexist, allowing more of the production process to be "productized"
This can be different to silicon valley style startups, but most of the market is solving the same old problems everyday.
Full disclosure, I actually have a no-code/low-code startup (
) hence I'm very biased!
Does Mathworks Simulink count as "no-code"? Because if so, then "no-code" seems to be incredibly useful across a diverse set of critical use cases.
If it's anything like Labview, incredibly useful but also incredibly awful.
Haven't use Labview but that characterizes Simulink quite well. It's really useful for simulating and prototyping your control system, but when you need to create high quality code and good libraries it gets frustrating very fast.
I've not used LabVIEW, but a cursory look indicates that it's not substitutable for Simulink.
IMHO, yes. So does Microsoft Excel, Airtable, n8n and AppSmith (
https://github.com/appsmithorg/appsmith
)
Every app that you ever built for the end user is a no-code tool. Reframe your question and you get your answer.
With that said, no-code is jumping the gun. Start with _less_ code.
Related, it really bugs me that some startups are throwing the āno codeā tag into their pitch decks or Product Hunt launches just to hop on a trend.
As you noted, every app built for the end user is āno codeā in some respect.
> Related, it really bugs me that some startups are throwing the āno codeā tag into their pitch decks or Product Hunt launches just to hop on a trend.
It's the blockchain hype for 2020.
> Related, it really bugs me that some startups are throwing the āno codeā tag into their pitch decks or Product Hunt launches just to hop on a trend.
Astute observation - guilty! lol.
Three forces are in mostly at play in any no-code solution.
- Abstraction
- Flexibility
- Cost of labor to produce functionality
Higher the abstraction lower the flexibility. However, for enterprises the ability to produce functionality without requiring a skilled work force matters a tonne. The success of Airtable & likes speaks volumes. That's because the world is not as complex as the article likes you to believe.
As developers, we underestimate nocode solutions. And enterprises probably over estimate.
I know Fortune 500 companies using online databases like QuickBase to operate their businesses. QuickBase is a sort of web-based relational database meets excel formulas. Business owners love tools like this. I know of a handful that run $50M+ ARR online businesses using no-low-code platforms where the only technical person on payroll is the IT guy that reboots the phone system each day.
This is a good take. To add to it a bit, there's another perspective that makes no-code solutions useful: team decoupling. You know Conway's Law about how the software is structured like the organization that produces it? No-code is that, but it's giving "coding power" to parts of the team that didn't have them before. E.g. a design team can accomplish a reasonable number of funnel tweaks using an A/B testing tool, and it means they get their job done faster (= more effectively), and the engineering team can focus on other problems. It's not a panacea and there's definitely risks, but I've seen it be helpful in this way in many contexts.
A ton of businesses have been running on no-code for decades. Excel, MS Access, Squarespace, Gmail and many others. Not every business is heavy on writing code. They just want to get things done quickly.
reifying workflows
In my experience as a software contractor, non-technical and even _semi_-technical people always underestimate the complexity of even the most mundane task.
I wonder if the author has ever written a sales order processing system. It _seems_ easy on paper, and you can probably crank out a naive implementation in a single morning ...
meh.
Its deeply patronizing to hear gate-keeping developers discuss when is 'no-code' acceptable...
Talk to a first line employee or manufacturing factory worker end user who wants to make minimal data apps on fly without the extra time or baggage.
Google Appsheets or Microsoft Power Apps are fantastic tools of this nature.
"Its deeply patronizing to hear gate-keeping developers discuss when is 'no-code' acceptable..."
I think we're mostly emotionally responding to the patronizing implication in these sorts of articles that no-code is the future and what we do is so easy that it's going to be replaced by a pretty GUI and some colored bubbles.
Does anyone deny these tools have some uses? They sure do. But they also manifestly have some serious weaknesses, as discussed in the article, and I'd add others beside. They may get incrementally more useful in the near future, maybe even build a few billion-dollar companies, but the odds of them displacing _any_ conventional code in the near future is basically zero; no matter how successful they may become, the net amount of "yes-code" programming work in the world is still going to go up.
I don't think a factory line worker should get a Real Language... I think it's that credulous journalists or companies with a marketing department ought to stop writing these "oh man, software engineers ought to _watch out_!" articles and instead focus on the actual capabilities and deliverables... which, frankly, end up obscured, oversold, and outright ignored when you try to pitch them as replacements for conventional software. Any startup in this space, I'd advise against positioning yourself that way... you're in for some angry customers.
UML is great but I think the key area where Google has had a whole lot more impact is with its text-to-speech and speech-to-text technologies for search. There's still a double digit percentage of people on this planet who aren't able to read or write. They haven't been able to benefit from advanced technology at all, until recently, due to the patronizing literacy requirements held by most software. In fact, I seem to recall, at one point or another, Google even made their search engine available by phone in a language that had no written form. Talk about conquering the digital divide.
It really is. High level programming languages were the original "no-code" compared to assembly of course.
That said: no-code is not "no engineering"
I guess it's kind of the reverse of the "perfect being the enemy of the good" adage. "Just barely works on a good day being the enemy of the good enough" maybe?
As a developer, I almost have no problem with companies choosing the no code tools. As cynical as this will sound, those apps will age like milk as per usual, and they will always want custom uses cases that will require platform knowledge. And, those platforms become old, and stop being cool. Itās almost like āno-codeā debt, itās gotta eventually get paid.
As a developer, I have the opposite opinion. I wish companies would stop buying this snake oil, because when the platforms _do_ get old, the companies will take their time migrating off them. The platform will be sold to some tiny support shop in Bumfuck, Nowhere, who will gladly send representatives to HQ to renegotiate the support contract. So despite it being a pain in the ass to literally everyone who touches it, it somehow stays in the company because millions or even billions' worth of business goes through it, and replicating the exact logic in $SENSIBLE_LANGUAGE is not an expense the VP or CTO will approve because "what we have works fine". So supporting it, integrating with it, and coding around it falls to the schlubs on the dev teams. Schlubs... like me.
Going back a few threads, I just wanted to tell you that I think you're right that we have to fight hard against severe illness/injury. At least, I've decided to fight it instead of allowing it to kill me early. We'll see how things go. It may still get me, but I'm going to fight it.
I'm sorry to have insulted you, I honestly figured you were another naive 20-something year old giving advice on something they know nothhing about (and I work in tech, so I can't just not hang out around with them).
Just as higher level language abstractions have taken many years and many iterations to get right, so, too, have/will low-code platforms.
Yes, some are prone to āspoilage over timeā. But to fully discard/dismiss the entirety of low code products is either foolish, or a conclusion that is based on old/incomplete information.
I use no-code shader graphs and they work great. I'm not trying to look at shader files all day.
In a practical sense but not intentionally so most JavaScript work is no code. Original work is not expected and developers are both terrified and bitterly hostile to writing original code. The result is frustration for everyone involved, including the end user.