An interpretation can be suggestive and still need evidence. Separating what the film shows from what we infer allows us to discuss it without turning every detail into a definitive answer.
Describing a scene and explaining its meaning are different tasks
An observation identifies something that can be located in the film: an action, a line, an edit, a relationship between images. An inference proposes how those details are connected or what they mean. Attributing intention adds another claim: that someone made a decision to produce a particular effect. These three tasks can be related, but they do not automatically draw on the same evidence.
HAL offers a useful example. In 2001, we see his communications, certain actions and the crew members’ responses. We can describe when information from him conflicts with other signals. Explaining all of his motives is a different task. The analysis of HAL must distinguish what the film allows us to observe from explanations added by other versions, texts or interpretations.
This distinction does not mean we have to limit ourselves to a literal summary. It is reasonable to infer that a behaviour affects trust, even if no character states our conclusion exactly. The point is to show the reasoning: which action changes an expectation and what consequences follow. An interpretation gains strength when we can return to the evidence supporting it, not when it piles up abstract words that could apply to any scene.
It also matters that we describe precisely what we do not see. The fact that a film does not show an action does not necessarily prove that it does not happen within the story. It may take place off-screen or be omitted through an ellipsis. However, that absence does not authorise us to invent a specific event and present it as confirmed. The scope for inference depends on the connections the rest of the story actually provides.
A document answers a specific question, not every question
A screenplay can show that an option was written down; a later version may change it. Credits identify roles and attributions, but do not recount every decision made during production. An interview records what someone says at a particular time and in a particular context. The Kubrick archive brings together materials of different kinds, whose usefulness depends on the question we ask, not on a uniform authority shared by the collection as a whole.
The relationship between a novel and a film helps demonstrate this limit. The source text may explain background or thoughts that the adaptation does not preserve in the same way. Any use of them in the analysis requires us to say where they come from. Adaptation studies compare these choices; they should not be used to silently fill the film with all the information available in another medium.
An interpretive case must also identify which version has been viewed. If a scene appears in one cut but not another, its absence has consequences for someone who has seen only one of the versions. A timecode may help, but it does not replace naming the sequence and identifying the copy. An exact figure applied to an unknown version creates an appearance of precision.
Documentaries about Kubrick add testimony while also selecting how to present it. A statement can be valuable without containing all its background. It is worth distinguishing what a participant says from the relationship created by the edit between several contributions. The guide to audiovisual portraits and documentaries helps us study this work without treating every documentary film as direct and complete access to the original process.
Hypotheses should be open to limits and counterexamples
In Eyes Wide Shut, some images are linked to what Bill imagines after hearing Alice’s account. This observation allows us to analyse how a fantasy takes visual form. It does not prove that every scene in the film belongs to the same level of reality. To support that broader claim, we would need to explain why other sequences should be read in the same way and what signals distinguish their conditions—or do not.
The page on dream, fantasy and reality explores this particular case. As an exercise in method, it is useful to look for counterexamples: scenes the hypothesis struggles to explain, information that does not depend on the identified imagination, or relationships that require additional assumptions. A proposal that can accommodate any detail by adding a new explanation risks becoming impossible to test.
The monoliths in 2001 pose another limit. We can compare their appearances, their relationship with the characters, and changes in scale or framing. These details allow us to propose functions within the film. They do not necessarily identify a single symbolic equivalent. Two hypotheses may explain some of the material without the film providing sufficient evidence to choose one exhaustive answer over the other.
In The Shining, the memory of violence at the Overlook can be analysed through the accounts, images and power relationships that actually appear. Moving from these elements to an all-encompassing explanation that assigns a fixed meaning to every number, object or coincidence requires additional evidence. Accumulating details does not fix a weak connection between them, and the selection should also make it possible to see the details that do not fit.
One practical test is to phrase the conclusion at the same level of scope as the evidence. If several scenes show an institution protecting its authority at the expense of certain people, that reading of its relationships can be supported. To claim as well that every visual decision was designed to communicate a single thesis requires a separate demonstration. An analysis does not become less interesting when it acknowledges that its conclusion is partial.
We should also distinguish compatible interpretations from contradictory explanations. A space can organise hierarchy and create disorientation at the same time. By contrast, two claims about an event may rule each other out. Pointing out this difference stops us from calling every disagreement ambiguity: sometimes functions coexist, sometimes information is missing, and sometimes an observation directly contradicts what has been proposed.
The most defensible reading identifies its scenes, distinguishes its sources and states where its explanatory power ends. It can revise a hypothesis without abandoning the analysis or searching for a key that unlocks the whole work. The aim is for someone else to be able to check what has been observed, discuss the proposed connection and recognise what information would be needed to support a stronger claim.
Sources for the examples: Stanley Kubrick, 2001: A Space Odyssey (1968), The Shining (1980) and Eyes Wide Shut (1999); film credits and scenes. Documentary reference: University of the Arts London, description of the Stanley Kubrick Archive. The analytical criteria are presented as a proposed approach to interpretation, not as statements attributed to Kubrick.
View all articles in Critical reception and debates about Kubrick
