nowheretogo.md
[](https://gist.github.com/)
Search Gists Search Gists
All gistsBack to GitHubSign inSign up
[](https://gist.github.com/)
You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Instantly share code, notes, and snippets.
omeid/**nowheretogo.md**
Created August 24, 2026 07:37
Show Gist options
- Star 5(5)You must be signed in to star a gist
- Fork 0(0)You must be signed in to fork a gist
* Embed# Select an option
- Embed Embed this gist in your website.
- Share Copy sharable link for this gist.
- Clone via HTTPS Clone using the web URL.
No results found
Clone this repository at <script src="https://gist.github.com/omeid/a9d6d1e3c25cb3aa577931e60e006f54.js"></script>
- Save omeid/a9d6d1e3c25cb3aa577931e60e006f54 to your computer and use it in GitHub Desktop.
Embed
Select an option
- Embed Embed this gist in your website.
- Share Copy sharable link for this gist.
- Clone via HTTPS Clone using the web URL.
No results found
Clone this repository at <script src="https://gist.github.com/omeid/a9d6d1e3c25cb3aa577931e60e006f54.js"></script>
Save omeid/a9d6d1e3c25cb3aa577931e60e006f54 to your computer and use it in GitHub Desktop.
We are not going anywhere.
[](https://gist.github.com/omeid/a9d6d1e3c25cb3aa577931e60e006f54#we-are-not-going-anywhere)
This might sound obvious, but it is worth putting it down, Software Development going forward will be largely done by AIs, you might find the quality subpar, but in terms of cost ratio, it is commercially good enough. Business will accept 99.99 at fraction of cost of 99.999. It is all about general consumer expectations, which will shift.
But most of all, not just that we are not going back, we are also not going anywhere. Software Engineering as science will be largely dedicated to AI development and outside of this discipline, it will slow down to a grinding halt.
No one is going to write new UI libraries if SOTA models know React best, no one is going to bother with new languages if SOTA models know Python, Go, JavaScript, and so on the best.
Yes, it will be easier for people to build new libraries and languages, but they won't gain traction. This might be different for large corporations who can afford to train and finetune models on their new fangled technology, but that will be the exception, and likely struggle with building a community and talent pool outside of this developing organisation as other people may not fancy using or even have access to their internal models.
We are not going back, we are not going anywhere.
[](https://gist.github.com/omeid/a9d6d1e3c25cb3aa577931e60e006f54)
**Sama-004** commented Aug 24, 2026
Copy link Copy Markdown
well i think of it like, we wont be reinventing the wheel but doing something else
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
**fwsGonzo** commented Aug 24, 2026•
edited
Loading
Uh oh!
There was an error while loading. Please reload this page.
Copy link Copy Markdown
Maybe the opposite take from me: We don't need to expose a JavaScript API anymore to attract people for eg. configuring something, plugins, etc. We can just present sandboxed Rust/$lang as a scripting layer. Why not? We just need to export "some documented scripting API", and on the other side they just need to explain what they want done. Why script in JS or Lua when it runs X times faster in a language where a lot of issues are caught at compile-time? In other words, why not configure X in whatever language that yields best/safest results with LLMs, whatever it may be.
Everything is one-shot now. "I want to configure this Webserver to handle signed tokens for my video content." => LLM understand the entire scripting/configuration layer in < 60 seconds.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
**FeepingCreature** commented Aug 24, 2026•
edited
Loading
Uh oh!
There was an error while loading. Please reload this page.
Copy link Copy Markdown
I wonder if at some point the big labs will make their own programming languages for optimal token use, validated over giant RLVR runs.
I suspect it'll probably look more like APL than Python.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
**pouyakary** commented Aug 24, 2026
Copy link Copy Markdown
For a long time I had wished to create a visual data-flow language for creating documents. Think about TeX, but instead of macros in documents; everything being programming blocks. It required a very optimized GUI and language that worked together, I needed infinite canvases, hundreds and thousands of nodes, a form of immediate GUI that had two graphic stacks (the GUI and the Skia based language in sync). Web was obviously the worse case, native kits where limiting, slow, and platform lock. ImGui and stuff are not what I want and... So in the end nothing worked.
Recently with GPT 5.6, I had seen Taylor Holliday (@wtholliday) making an experimental GUI kit in rust exactly for this purpose. It was amazing and so I made a book about its architecture and organization, then used it to create a GUI system for my own. And implemented shader based scrollbars, infinite canvas, nodes, the runtime, all with 20$ worth of subscription:
Now this as you see has so many problems and is so half baked, but this has to be illegal. Impossible. I had the model make me a rope based text editor with all the intricacies of caching I-beam positions when jumping up and down, selection, scrolling, ... and I even managed to build parentheses that render right:
I think the corporate is always headed for the lowest hanging fruit, but if you wanted to do radically hard things, now you can do, and its possible. You don't need to fine tune a model, this is already a software with its own GUI and compiler that works and works way better than I could write before.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
**jsawyerdev** commented Aug 24, 2026
Copy link Copy Markdown
The “we are not going anywhere” argument mistakes today’s autocomplete bias for tomorrow’s engineering reality.
Yes, AI will write more code. Obviously. But “AI writes code” is not the same as “software engineering freezes around whatever models currently know best.” That is the classic Vibe Coding delusion: treating code generation as the whole job, when the hard parts of software are deciding what should exist, what must never happen, what failure costs, what contracts matter, and what survives contact with users, attackers, data, latency, scale, regulation, and time.
Vibe Coding is great at producing plausible surface area. It is terrible at knowing whether that surface area should exist. It can spray React components, API handlers, migrations, tests, and documentation into a repo faster than a human can review them. That is not progress by itself. That is debt creation with a nicer interface.
The claim that businesses will accept “good enough” quality because it is cheaper is only partly true. They will accept cheap code for cheap problems. Landing pages, prototypes, internal dashboards, disposable glue: sure. But the moment software touches money, identity, infrastructure, health, contracts, safety, compliance, reputation, or customer trust, “99.99 at a fraction of the cost” becomes a very expensive sentence. Reliability is not a decimal decoration. The last fraction is often where the actual business lives.
And the idea that no one will build new languages, libraries, or UI frameworks because models know React best is backwards. Models lower the cost of experimentation. They make it easier to port examples, generate docs, produce adapters, write migration scripts, and teach unfamiliar APIs. The bottleneck for new technology is not “can someone type enough boilerplate?” It is whether the new thing solves a real problem strongly enough to justify switching. That has always been true.
If anything, AI makes bad dominant stacks more vulnerable. When the cost of trying alternatives drops, incumbency becomes less comfortable, not more. React will not win forever because models have memorized yesterday’s internet. Python will not absorb every future workload because it has more Stack Overflow residue. Tools win when they make real systems easier to build, reason about, operate, and change.
The future is not “everyone vibe codes and engineering stops.” The future is that shallow coding becomes cheap, and therefore less valuable. The premium moves upward: architecture, taste, correctness, product judgment, debugging, security, data modeling, performance, observability, and knowing when generated code is nonsense wearing a confident hat.
AI will replace a lot of typing. Good. Typing was never the discipline.
What it will not replace is accountability. Vibe Coding has no accountability. It has vibes. It produces code without ownership, tests without conviction, abstractions without taste, and confidence without comprehension. That is fine for a weekend demo. It is poison as an engineering culture.
We are not going back. Agreed.
But “we are not going anywhere” is surrender dressed up as realism. Software has always evolved around constraints: hardware, networks, teams, users, economics, security, deployment, and failure. AI is a new constraint and a new lever. It will reshape engineering. It will not end movement.
The useless crap is not AI-assisted development. The useless crap is pretending that prompting a model into emitting familiar code is the same thing as building durable software.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
**omeid** commented Aug 24, 2026
Copy link Copy Markdown
Author
> @FeepingCreature > > I wonder if at some point the big labs will make their own programming languages for optimal token use, validated over giant RLVR runs. > > I suspect it'll probably look more like APL than Python.
In my estimate, it is entirely possible.
However, I think it won't be a new language but rather evolution of some existing language, maybe Python, possibly Go, perhaps Zig or TypeScript; the specifics are hard to weigh. But for an entirely new language, and the ecosystem it requires around, the economics and incentives are no longer there.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
**Gogolian** commented Aug 25, 2026
Copy link Copy Markdown
In past old jobs came to a halt, while new jobs and opportunities arose.
Will it be the same now?
Who knows?
I guess we will find out soon.
Let's enjoy the ride.
Sorry, something went wrong.
Uh oh!
There was an error while loading. Please reload this page.
Sign up for free**to join this conversation on GitHub**. Already have an account? Sign in to comment
Footer
[](https://github.com/) © 2026 GitHub,Inc.
Footer navigation
- Terms
- Privacy
- Security
- Status
- Docs
- Contact
- Manage cookies
- Do not share my personal information
You can’t perform that action at this time.