I have written piles of essays on why this "build to learn" approach is incredibly wasteful:
- it locks you in to local optimization
- it burns trust
- it can still only function at a human pace
- "fix the problems" relies on a problem-finding feedback loop no one sets up
- there is never a v2
Without the synthesis process of critique, everyone who vibe coded a thing now has their own mental model, which means they are no longer actually a team.
This kills the project.
When you treat low fidelity artifacts (regardless of the amount of detail) as disposable sketches, this doesn't happen.
So teams are paralyzed! Everyone is freaking out that their turf is being stepped on.
Because they are obsessing over the artifact. They can only see the purpose of the artifact as being shipped to production, rather than as a way to explore the problem/solution spaces.
The other cool thing about critique is that it resolves a problem vibe coding has created.
Product teams used to function sort of OK when one person (the PM) owned having ideas and everyone else executed. But now everyone can "come with an idea."
You can't ship EVERY idea down the fidelity cascade
As an aside, this mindset is often described as "lean" because you skip the "waste" of low fidelity work.
Buddy, no. Low fidelity is not the waste. Work in low fidelity is what prevents the high detail work from being waste. Lean is when cheap-as-free sketches are all you need to avoid that waste.
But that's not how people who are unfamiliar with critique think. They see high detail and think high fidelity. Then they think, "what comes next after high fidelity? Oh, it's production."
"Therefore we must ship this vibe-coded prototype."
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAH!
And once you have those answers, you THROW THE THING AWAY. Its purpose has been served. You no longer need it.
You do not, under any circumstances, get sign-off from people who "liked it" and then roll it unchanged into the next level of fidelity.
A real critique is not a pass/fail gate. It structures the conversation to answer questions about the artifact that the designer has. Is it fit for purpose? What is working about it? What isn't?
Scaffolding the conversation to lead towards those answers are the purpose of the artifact.
Professional UX designers (and to extent, all creatives) are less wowed by all this because being able to mock something up quickly is a skill we learned very early on.
And the thing you do with that mock after you make it is that you critique it. Separate yourself from the object.
The mistake is confusing detail vs fidelity. Vibe prototypes let you add a lot of detail quickly, just like how component libraries do.
But all of that detail is essentially distraction and waste. Showing off your tasteful drop shadow on the card view is not stress-testing the merit of the solution
If your understanding of the process is limited to the fidelity cascade, this LOOKS like a huge breakthrough.
We've been having the same conversation with design systems for the past 10-odd years. "Wow this lets us skip to high fidelity, we're so productive now!"
WRONG!
Vibe coding broke people's brains because they had a bad understanding of the software design process.
The pop culture model goes something like this: start out with a sketch and then render that same idea in progressively finer detail.
So when tools could "skip" to high detail, people went WOW.
If LLM boosters were so confident in their tools. they wouldn't be ashamed to put "made with AI" on their stuff. They would consider it a badge of excellence.
The simple fact that they try to hide it tells you everything you need to know.
If you like Civ, Landnama is the closest of these. It also has my favorite art style of the bunch.
Important to note also that any problem framing by necessity contains a grain of solution, so while the users will be aware that there is a problem, they are not always going to know how to articulate it. The example of the underpowered shotgun comes to mind.
I have a couple of games on my phone that could be played with one hand, so long as you can prop it up on something. Loop Hero, Dicey Dungeons, Landnama, Little Wars all rely on tapping stuff rather than try to emulate a controller.
So Arc Raiders is malware, basically.
If you use Discord either uninstall the game or stop using Discord because that "game" is accessing way too much data it shouldn't, even reading your DMs and sending them to the devs.
Super fun thing to find when Discord wants access to everyone's IDs.
PC World retracted the original article because it was nonsense www.windowscentral.com/microsoft/wi...
Their unjust war vs our limited scope special military operation
The government probably thinks that this won't be a problem because none of the truck routes pass through the strait of Hormuz either
Less mulling, more pulling!
"Users don't know what they want" is a great way to burn all of your money and goodwill and in the end STILL have to do the thing people were asking you to do from the beginning.
Microsoft joins Mozilla in backing down on its AI push, after being yelled at for 2 years straight
So much money wasted on ignoring something they already knew
The key to unlocking a much more low-stress period of my writing life was learning to ask the question: "What do you want to feel that you're not feeling?" It focuses the origin of the note and takes it away from whatever weird solve they're pitching.
RTO is one of the classic method traps; everyone is managing how much they have Done The Thing and not whether or not doing the thing has helped improve anything.
Scrum and AI are two other notable members of that set.
This is a good principle for design in general:
When users tell you there's a problem, they are usually right.
When users tell you how to fix it, they are usually wrong.
(bonus: if you replace "user" with "stakeholder" this still applies)
Inevitably they will blame psychosis. And we've seen this before with companies and academics claiming lung cancer is caused by stress not smoking!
Remember Hans Eysenck? www.theguardian.com/science/2019...
> This research programme has led to one of the worst scientific scandals of all time
1/n
Any gov design folk got experience of managing mental models stakeholders in how a service 'should' be found?
stores have no idea how much shoplifting takes place, they only know when stock doesn't match records. how much is due to human error, incorrect shipments, breakage, and theft is a guess. literally cannot know if shoplifting is on the rise.
this was a radicalizing discovery for me.
