Vibe coding will deliver a wonderful proliferation of personalized software
- Reference: 1765355354
- News link: https://www.theregister.co.uk/2025/12/10/vibe_coding_is_good_enough/
- Source link:
Until early November, none of the passes I've had at vibe coding prior yielded any results worth pursuing. Then I put Anthropic's Claude Code to work (courtesy of $300 in credit from Anthropic) transforming a prototyped Python data analysis project into a full-featured, professional-class tool. It took a few hours (and a fair bit of care and tending), leaving me with the impression that vibe coding had turned a corner into usefulness.
When Google dropped its latest Gemini 3 Pro model at the end of November, they left another gift under the tree: Antigravity, an integrated IDE and agentic coding tool that draws upon Google's recent acquihires from AI coding startup Windsurf. I fired up Antigravity with a thought about a bit of code I'd love to have – if I could have it with minimal effort. The answer came almost immediately: A VRML 1.0 browser.
[2]
VRML 1.0 came out in mid-1994, only to be quickly supplanted by the far superior and still reasonably-well supported standards VRML 2.0, VRML97, and X3D. That left a fair bit of early content stranded. VRML 1.0 translation tools have been lost in the mists of time. There's no way to load a VRML 1.0 world and view it.
[3]
[4]
So I fed Antigravity a copy of the [5]VRML 1.0 spec , with the request that it create a VRML 1.0 browser for macOS, written in the Swift programming language, and using Apple's Metal 3D renderer. I didn't give it much more than that - but I did ask if it had any questions. It had a few, and then set to work.
Over the course of a Sunday afternoon, and Monday and Tuesday evenings, Antigravity churned out code, which I then plugged into Xcode, compiled, and ran.
[6]
While Antigravity worked, I functioned as its quality assurance assistant, because while Google’s tool can click its way around Chrome, if you're doing pure web work, it can't (yet) operate the macOS GUI. When compiles failed, I pasted those errors into Antigravity, which corrected its mistakes. When it built, I launched it, took screenshots, and fed those back in, with comments about the accuracy of the app's rendering.
By the end of Sunday night it was all working – sort of. Antigravity seemed quite pleased with itself, having plowed through the project checklist that it created at the outset and cleverly uses to keep itself on track when working on a large project. Yet you could have driven a planet between that working version and a fully compliant VRML 1.0 browser. Antigravity had done just enough work to render very simple VRML 1.0 files but had left vast portions of the specification unimplemented.
When I pointed that out, it drafted a new plan, and we spent Monday evening methodically implementing the features it didn’t code the first time. On Tuesday evening, I asked it to generate a conformance test for the app - which the app promptly failed. I fed that output back into Antigravity as a bug list.
[7]
That's when I had my penny-drop moment. Antigravity (and Claude Code) are quite powerful, but naïve. They don't know what they don't know . Unless a human gets in the agent harness with them, keeping them focused and aware, they tend to just sputter out. It felt as though I drove one of those massive dump trucks they use in Australia's mines: incredible power, that can easily [8]tumble into a mineshaft - unless the human at the wheel remains continuously vigilant.
[9]Whatever your job, mentoring is your job – and the one that matters most
[10]AI is the flying car of the mind: An irresistible idea nobody knows how to land or manage
[11]Toys can tell us a lot about how tech will change our lives
[12]I started losing my digital privacy in 1974, aged 11
Skill at “steering” coding assistants may soon be the quality most sought after in software engineers, systems administrators and the like. It's a balance between a light touch and a firm focus.
After roughly eight hours of “steering” Antigravity, I had a VRML 1.0 browser for macOS - you can build it yourself [13]from source . It might not be 100 percent compliant with the spec - that'll take a few more hours of testing. Yet it's already “good enough” to display all the examples from my 1995 book on VRML, and other bits of content I found online. To go from a spec and a goal to an app in eight hours feels like success. It would have taken me ten times longer to code this myself - and that's if I knew how to code in either Swift or Metal.
I wrote a few VRML 1.0 browsers thirty years ago, so I came to this with deep domain knowledge.
If I'd asked Antigravity to create a fault-tolerant database that would preserve transaction history across power failures, I'd have made a muddle of things because I don't know enough in that domain to steer the machinery. That means domain experts in software engineering and operations aren't going anywhere - but they will be getting a lot more productive.
Beyond these obvious wins, it's now possible to forecast a time – before the end of the decade – when these tools have been sufficiently "softened" to allow pretty much anyone within an organization to rapidly develop an app for a specific use case. That's an interesting bit of road directly in front of us: a Cambrian explosion of weird user-specific apps, designed by users for themselves. These tools make the best better, giving everyone else the benefits of write-once, run-once disposable software.
Vibe coding looks to be the gift that keeps on giving. Happy holidays! ®
Get our [14]Tech Resources
[1] https://www.theregister.com/2025/11/24/opinion_column_vibe_coding/
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aTlS2HTX7jwD_MtPnvbGPQAAAIw&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aTlS2HTX7jwD_MtPnvbGPQAAAIw&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aTlS2HTX7jwD_MtPnvbGPQAAAIw&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://paulbourke.net/dataformats/vrml1/
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aTlS2HTX7jwD_MtPnvbGPQAAAIw&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aTlS2HTX7jwD_MtPnvbGPQAAAIw&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2025/12/01/google_antigravity_wipes_d_drive/
[9] https://www.theregister.com/2025/11/19/every_job_is_mentoring/
[10] https://www.theregister.com/2025/10/15/ai_vs_flying_cars/
[11] https://www.theregister.com/2025/09/18/toys_and_tech_futures/
[12] https://www.theregister.com/2025/08/13/digital_privacy_senseless_data_preservation/
[13] https://github.com/mpesce/vrml-vibes
[14] https://whitepapers.theregister.com/
100% this - just because a bunch of clowns can hammer together a red bull soapbox to race down a hill doesn't mean they can design a real car, and that is proved by the destruction most of them leave in their wake.
Vibe coding is the IT equivalent - if you want to have a bit of a muck around, then go for it and have a laugh - but don't expect to design anything serious or dependable in any way because it will end in bumps, bruises and very likely complete disaster at some point down the road...
"I don't call it "vibe coding" because "vibes" don't come into it."
I've taken to calling it Jive Coding, for somewhat obvious reasons, some of which you touched on.
Strangely, the folks who have taken to this new scam instantly understand what I mean, and hate me for it. C'est la vie.
Do you smell Tulips !!!???
This ^^^^^^^ 10^10^10^10^10^10^10 times at least.
If I was a great and talented artist I could produce great art with an A3 piece of paper and a selection of coloured Crayons.
I would NOT expect Average Joe/Joanna to be able to do this !!!
The pitch for 'AI' is always 'Would' do this ... never 'Could' if you have the right skills to start with.
'AI' is too random and lacking in control to be unleashed on the world ... You need tools to work 100% of the time.
Warnings and listing of exceptions does not work in the 'Real world' ... people will use the 'AI' expect 'right answers' and impose the randomness on the rest of the world.
I do not want 'AI' generated ANYTHING if it is not checked for completeness/validity/overall safety by SKILLED people not another 'AI'.
The continuing push to find some arena where 'AI' is accepted & used to generate profits and start the general acceptance into our everyday lives is beyond desperation.
It is not fit for use ... period !!!
Unless it improves immensely not just in quality but repeatability across all possible arenas, then 'AI' is just a clever trick !!!
If 'AI' is involved in the production of things that impact people & lives in general then this scam cannot be allowed to continue.
Real harm WILL happen as a consequence of ignoring the limitations of the current 'AI' ... real harm that will cost real money (as money is the only thing that talks when referencing 'AI') !!!
The bubble cannot burst soon enough !!!
:)
Job losses
What you're saying is that an experienced, senior developer can indeed make use of an LLM to steer it to some code that might actually do what you want it to.
I think we're probably going to have to learn to walk before we can run. The example in the article was too complicated. I don't think we should be trying to have it flesh out complete apps yet. Instead, that senior developer is going to feed the LLM the various stories in the application. Maybe down to button click level. What if you could feed it a unit test and have it build an implementation? It won't have inexperienced coders (or business analysts) building apps right now. But it will allow corporations the ability to gut project teams. You won't get rid of the highest paid developer yet, but you'll dump 5-10 junior developers for each one of them. That's a significant savings already.
And of course, as we use the tools the AI designers will learn the deficiencies and the senior developers will work themselves out of a job. Kind of like training your off-shore replacement.
Mixed opinions
Make your mind up Register - it's not that long since you published an opinion article decrying Vibe; https://www.theregister.com/2025/11/24/opinion_column_vibe_coding/ :)
Re: Mixed opinions
The journalists here actually have different opinions so articles can contradict previous ones. That's a useful attribute that prompts debate and discussion.
Read both the articles and make a decision for yourself.
(also, use an anchor tag and you'll get an actual link in your post: [1]https://www.theregister.com/2025/11/24/opinion_column_vibe_coding/ )
[1] https://www.theregister.com/2025/11/24/opinion_column_vibe_coding/
Re: Mixed opinions
Thanks for the tip about tagging.
My comment was slightly tongue in cheek but when I started consuming news it was through print. When newspapers/magazines put 'opinion' on something then it generally represented the views of the paper (or at least the Editor) - this has stuck with me.
As for Vibe coding, I think it's fine for hobbyists but like all AI not worth killing the planet for.
Re: Mixed opinions
I get the reference to Opinion pieces in Newspapers.
I noticed that in the US of A papers it is common to have opinion pieces by multiple people in the same paper.
The cynic in me thinks that this is a way to NOT have an opinion for the Newspaper and that readers will favour the piece they agree with without assigning the view to the Newspaper as a whole. The opinion pieces are always 'Left vs Right' to cover the political landscape of the US of A.
Everything in the US of A is politics and taking a side on ANYTHING is always translated to a political standpoint, to avoid losing readers you publish multiple opinion pieces to 'balance the debate' [Roll eyes] and to take NO actual standpoint as a whole.
There are rare exceptions, usually because the Newpaper owner is actually running the Newpaper as a mouthpiece for a political side they support.
:)
once but faster and more chaos
Why does this just sound like the mixed horror bag that was cottage industry of ms access dB to solve everything.... Badly. VB apps that commit all sorts of crimes against coding and logic. Yes users can be empowered, but we've been trying to undo those house of cards messes for decades. Now there techbroa are trying to make it happen after with less controls.
I see a future full of failing apps leaking data and causing disruption at the worst times as the skills to deal with the fallout wither and die.
It's not vibe coding when you know what your doing
Vibe coding is a non-developer with no development experience driving an LLM to make code.
"They're now good enough to do things well"
They're really good at deleting hard-drive partitions, for example: https://www.theregister.com/2025/12/01/google_antigravity_wipes_d_drive/
Re: "They're now good enough to do things well"
Just not neccessarily the 'thing' that you intended !!!
More and more I am reading promo for 'AI' which seems to be at odds with reality !!!
'AI' has not suddenly got better ... the promotion has just stepped up a factor or two.
If 'Vibe Coding' is so good, why are we no seeing any large corp or industry praising the value of it to them and the real 'positive' impact on their bottom line.
P.S.
Praise from people with a 'stake' in 'AI' selling more does not count !!!
So no 'MS' and their recent '30% of code is produced by 'AI'' !!!
:)
I think I need to find my exploit kit, generate some interesting VRML 1.0 files and send them to you for your both our pleasure.
If this was slashdot...
...I'd tag this article as flamebait :)
4GL
Rapid Application Development
Low Code
Vibe Coding
All wet dreams of MBA-types wanting to use cheap labour to replace expensive skilled workers.
Just like most things "AI", they look good at first sight ("Look at this amazing app little Johnny has built"), but are likely to cause so many issues with security, data integrity, reliability, stability, scalability, maintainability, yada yada... that the lifetime cost will be much more than it would have been if a skilled developer/team was employed to create it in the first place. Of course the manager of the initial development will have trousered his bonus for delivering something on time, and has long-since moved on.
I like reading sci-fi novels, and one thing that's always true of futuristic computers is that they are apparently very easy to hack. Whether it's an astromech droid or a teen in a hoodie, if the heroes need to get through a computer system, there's someone in the party that can do it in seconds, or minutes if there's some need for drama.
Each time, I wonder: this is set centuries if not millennia in the future, the story imagines amazing advances in all fields of human knowledge... except computer security, which appears to have gone backwards? Why?
Well, now I know.
Ask and you shall be answered !!!
Answers:
1. Because Security is hard to do well and computers are also hard to 'do well'.
2. Security is ALWAYS added on at the end because building it in is hard (See 1.) and it slows down producing/testing the code.
3. If it is hard to do and costs money to do as well someone will always take the shortcut and HOPE they get away with it !!!
4. Bacause no one looks back in history to learn any lessons ... apparently !!!
5. Because Sci-Fi is based on real-life, at some level, therefore see (1. & 2. & 3. & 4.)
6. Because without some flaw somewhere the Sci-Fi story would be very short and sad as the main actors are caught and killed.
:)
I read a Google "vibe coding" gush-agrada thing last night...
In the end some "associate" bloke had "vibe coded" an html front end using Google's latest and greatest, that looked like it had crawled out of a Contoso example from 1999. No back end, thank God although Google reckons that's soon. The whole thing looked like a non functioning joke nailed into GitHub and lurking on Google Cloud.
I think we're safe.
"rewrite expectations about how IT will operate before the end of this decade."
You mean it can get worse ?
This article on El Reg is the equivalent of some ancient Christian walking into the arena, knowing the lions are already waiting and blowing a raspberry to the Emperor's Seat.
But I'll bite:
What you're saying is that an experienced, senior developer can indeed make use of an LLM to steer it to some code that might actually do what you want it to. Simply assuming that you are one. Though making the bold claim that your app is "professional level" had me go "orly?"
I'll leave aside for now how you checked the code wasn't a dismal mess, whether you had tests (unit and integration) generated and verified they weren't just a giant set of methods mocking everything and returning "assert True"
How will this aid an inexperienced coder (or worse, a Dunning Kruger manager)? Is the success criteria "well it seems to do what I want and if I squint the results look correct"?
I've done what you have and frankly, I don't call it "vibe coding" because "vibes" don't come into it. I have an idea, some very precise prompt instructions that can on go on like a Tolstoy novel, and then expect to be doing a lot of hand-holding. To the point where I wonder I shouldn't have just written the whole mess from scratch without boiling the planet.