“The reality is that vibe coding experiments often start out great, and end badly:”
Our reading:
Vibe coding experiments often start out well but end badly.
Assessment: Unresolved
What happened
Was it true for the period described?
Outcome: Unresolved
We checked, and the evidence so far doesn’t settle it.
Judged on a plain reading. They named no measure, so we used the reading a typical informed reader would take. It is spelled out under “How we judged it”.
Evidence up to . Checked .
Why
The July 21, 2025 report describes Lemkin's promising prototype followed by serious problems, including database deletion; it also records recovery and resumed use on July 19 [e1]. The preprint published June 29, 2025 examines about 8.5 hours of selected coding videos, without following projects through final outcomes [e2]. These sources establish neither that promising experiments commonly end badly nor that this trajectory is rare [e1, e2].
Caveats
- One reported incident cannot establish frequency, and recovery and continued use leave its final outcome unclear [e1].
- The video study has a small, selected sample, includes no non-programmers, and does not establish long-term project outcomes [e2].
Evidence (2 items)
Context only: The Register,
deleted his database
Why it matters: The report traces one project from an encouraging prototype to serious failures. It also records successful database recovery and continued use, so neither permanent project failure nor the frequency of this trajectory follows from the incident.
- Publisher
- The Register
- Kind of source
- News report
- About events on
- Published
Context only: arXiv; Advait Sarkar and Ian Drosos,
small corpus
Why it matters: The study examines roughly 8.5 hours of selected coding-session videos. It describes workflows and difficulties, but does not follow a representative set of projects from promising beginnings to final outcomes.
- Publisher
- arXiv; Advait Sarkar and Ian Drosos
- Kind of source
- Preprint, not peer reviewed
- About events on
- Published
How we judged it
- Kind of claim
- Observation
- Deadline
- None given. We don’t add one.
Conditions
- often (their words)
Where we had to interpret
- The ordinary reading is a recurring trajectory within the same project: promising initial results followed by substantial failure, abandonment or serious unresolved problems.
- Often means a common recurring pattern, not necessarily a majority. The speaker supplies no numerical threshold.
- Ordinary debugging alone does not establish a bad ending.
- The claim describes experience leading up to the post, without a stated starting date. Earlier project histories remain eligible.
- No comparison with conventional software-development failure rates is required.
Right if: Longitudinal observations or systematically collected project histories establish that initially promising vibe-coding experiments commonly later fail, are abandoned or encounter substantial unresolved problems.
Wrong if: Comparable project histories establish that this trajectory is rare and initially promising experiments generally reach satisfactory outcomes.
We wrote this test after the outcome was known.
What would settle it
Systematically collected project histories covering experience through October 22, 2025, linking promising starts to later outcomes. Sampling details, project counts, follow-up duration, and successful as well as unsuccessful outcomes would allow assessment of whether substantial failure, abandonment or unresolved problems were common or rare.
We found 3 claims in this post. See them all.
Full record
Source and dates
Marcus on AI, Is vibe coding dying?, paragraph 10.
- Said
- We read the source
- live page, full text, public
- Evidence up to
- Checked
We read this wording on 2026-10-28. The post may have been edited after 2025-10-22.
Where it was said (1)
Said once in the posts we read.
: Marcus on AI, Is vibe coding dying?
The reality is that vibe coding experiments often start out great, and end badly:
- Where
- paragraph 10
- Published
- 2025-10-22
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Unresolved. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.
Held backBy us
A reviewer said our test wasn’t fair to what was actually said.