“And importantly, I predict that many will be hard to remedy.”
Our reading, the sentence that was checked:
Many of Sora's physics glitches will be hard to fix.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Sora’s Surreal Physics, paragraph 21.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-soras-surreal-physics-dff3dd-a1
We recorded 10 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-29. The source may have been edited after 2024-02-16. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Forecast
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No deadline stated
- Deadline state on 2026-10-29
- No deadline stated (the release’s reference date)
From the speaker
- I predict
- many
- hard to remedy
Our interpretation
- Many refers to several of the physics and continuity failure classes discussed in the post.
- Hard to remedy means requiring substantial remediation effort, rather than merely remaining visible at a later date.
- Assess later Sora revisions, without treating results from unrelated video models as decisive.
- The adjacent explanation about compression and data is not independently established by persistence of errors.
Counts as supported if: Development records or repeated controlled evaluations show several original failure classes persisting despite substantial targeted remediation.
Counts as contradicted if: Documented development and evaluation show that most relevant failure classes were reliably removed with straightforward fixes.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No deadline stated
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
Later evidence shows claimed improvements and continuing uneven performance. Neither establishes how hard several original glitches were to fix. The available records do not connect specific failure classes to documented substantial remediation attempts, and the benchmark audit identifies important measurement confounds. With no deadline and no decisive event meeting the rule, the forecast remains unresolved.
Limitations
- Public development disclosures provide little detail about targeted repair effort.
- Many and hard remain qualitative.
- Persistence alone cannot establish the proposed causal explanation.
Evidence (2 items)
Context only: OpenAI,
Our latest video generation model is more physically accurate, realistic, and more controllable than prior systems.
Why it matters here: The developer reports progress, a material qualification to a persistence-only argument. The announcement does not quantify effort or establish that most original failure classes were easily fixed.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: Tim Rädsch and colleagues, arXiv,
The Sora 2 performance is notably worse in April 2026 than in October 2025.
Why it matters here: Repeated measurements show that later performance was not uniformly improving. The study also identifies prompting and measurement confounds; it does not link this change to substantial attempts to fix Marcus's original failure classes.
- Publisher
- Tim Rädsch and colleagues, arXiv
- Kind of source
- Preprint, not peer reviewed
- Describes events of
- Public from
What would settle it
Dated development records and comparable before-and-after tests showing that several original failure classes resisted substantial targeted work, or that most were reliably removed through straightforward fixes.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Sora’s Surreal Physics
And importantly, I predict that many will be hard to remedy.
- Where
- paragraph 21
- Published
- 2024-02-16
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-soras-surreal-physics-dff3dd-a1: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.