# Visual Compression
# Book: Language
# Chapter URL: https://programmer.ie/books/language/05-chapter/
# Prose fingerprint (sha256, normalized): 1e976d2fdb30ce5545a62d451e4f51e507575d900af0194377eabcb6a92df84a
# Snapshot generated at build time by Hugo. This is the complete rendered prose
# of the chapter; it is not a transcript of the source file.

It is Monday morning and a security engineer faces forty browser tabs: weekend vulnerability disclosures, three vendor advisories, a thread dissecting one of them, two papers someone said she must read. She has an hour before stand-up. She will open perhaps six tabs and skim two. The others will be judged, kept or closed, read or abandoned, on the thinnest of evidence: a title, a favicon, a remembered recommendation, the first three lines glimpsed on hover. Every judgement is a bet. The price of getting it wrong depends on what she misses. The advisory she dismisses because its title reads like routine patch noise might describe the library her team ships. Opening every tab is not a solution either: an hour spent on marginal material can crowd out the one disclosure that needed investigation.
This chapter is about that bet, and about the first thing a flexible representation can do for it. Not help someone understand a page, which is a later question, but the narrower decision that comes first: is this worth my attention at all? Previews have always existed to answer it: titles, snippets, thumbnails, abstracts. The new possibility is to derive a preview from a source and a declared current task, choosing a form that makes the selection evidence quick to inspect. A compact text card may beat an image. The question is not whether a visual can be generated, but whether a generated cue improves the decision over the best available ordinary cue.
The chapter follows one path. The page is fixed. A smaller representation is derived from it. Making it small is compression. Something becomes easier to see. Something may disappear. And the question the whole book keeps returning to follows: what had to survive?
A smaller representation, and what it is for The purpose comes first, because everything else hangs on it:
A preview succeeds for a stated task and attention budget when it helps the reader select worthwhile sources, without concealing the distinctions that make that selection correct.
This is a criterion to test, not a property granted by making something shorter. Selection quality needs a task-specific reference judgement, and the costs of a missed source and a wasted opening must be declared rather than assumed to be equal.
Not fewer words. Not fewer pixels. Not prettier cards. A preview that halves inspection time while hiding the decisive caveat can fail badly. A preview that takes the same ten seconds as a title but prevents an important skip may be valuable. The unit of analysis is the selection decision under a limited budget: what was opened, what was missed, what was delayed, and at what cost? The rest of the chapter should stay close to those observations.
Readers deciding what to open behave in part like information foragers. Pirolli and Card’s information-foraging theory models how people use visible cues about material they have not yet reached: titles, snippets, surrounding text, images and layout. Those cues provide information scent, a reader’s estimate of the likely value and effort of following a link. Scent is not the value of the hidden page and is not necessarily accurate. It is conditioned by the reader’s current goal and prior knowledge. The theory helps explain why a cue attracts a click or leads someone elsewhere; it does not by itself establish that the resulting choice was correct.
A generated preview, in this vocabulary, proposes a new scent cue. It stands where the ordinary snippet stands and suggests what lies down the trail and why opening it might matter for this task. It cannot know the value of engaging with a source merely from sounding confident about it. Its claims need to remain attributable to the source, and its value has to be judged against actual selection decisions. Two failures matter, with costs that depend on the work and the budget:
FALSE OPEN the preview looks relevant → the reader opens it and spends time that may displace another source FALSE SKIP the preview looks irrelevant → the reader does not open it in this pass and may miss task-critical evidence A false open costs attention and may indirectly cause a false skip when the budget is fixed. A false skip can lose the one item that mattered: the advisory, the contraindication, the contradictory result. But its actual cost depends on the task, the consequences of delay, and whether the source can be recovered later. Overall selection accuracy hides this asymmetry. Record false opens and false skips separately, distinguish high-consequence misses from routine ones, and evaluate the set selected under the budget, not each card in isolation. A more attractive preview that increases consequential misses has not earned its place.
Conceptual failure comparison. The two errors are asymmetric and must be reported separately; a shorter card may omit background, but it may not change the claim. Open full-size diagram. Compression is not neutral shortening It is easy to talk about compression as though it were neutral shortening: the same thing, with less of it. It is not. Shortening makes choices about which relations, qualifications, entities, numbers and sources remain visible. Sometimes it makes the topic or conclusion easier to spot. Sometimes it removes the population, condition or exception that decides whether a source deserves attention for this reader’s task. The content need not be repeated in full, but a preview must not change a conditional claim into an unconditional one. Four effects are easily run together:
attention gained ≠ understanding gained ≠ correct inference ≠ safe action A preview can draw the eye without helping anyone judge relevance. It can enable a correct decision to open a page without teaching the page’s content. And a reader may correctly understand a short preview while still lack the evidence required to act safely. Those are different outcomes: selection, comprehension, inference and action. A selection study should measure selection. It should not borrow the language of understanding or safety until it has tested them separately.
The evidence most directly discussed here concerns visual and graphical abstracts used to communicate scientific research, not AI-generated previews used to triage a heterogeneous browser queue. That makes the studies relevant analogies, not validation of the proposed system. They show how attention, preference and reporting integrity can separate; none directly measures whether a task-conditioned preview helps the engineer select the six pages she should open.
Visuals can attract engagement without showing whether readers selected the right source. Chapman and colleagues (2019) randomised 41 accepted surgical manuscripts to dissemination as plain-English abstracts, visual abstracts or standard tweets. Visual abstracts produced more overall social-media engagement than the plain-English condition, with most engagement coming from healthcare professionals and little from the public. The study measured interactions with posts, not whether readers understood the papers or made better source-selection decisions.
Liking a form is not the same as learning more from it. Across three randomised trials involving students, consumers and doctors, Buljan and colleagues (2018) found no difference in measured knowledge between Cochrane-review infographics and plain-language summaries. Participants nonetheless preferred the infographic and rated its reading experience and user-friendliness more highly. That is evidence of a preference–knowledge gap in those tested materials and populations, not a result about source triage.
Compressed scientific communication can carry distortion. Muller and colleagues’ observational study, published in 2025, compared 119 infographics with the matching abstracts and full texts of randomised trials whose primary findings were non-significant. They identified spin in 33% of infographics and 26% of abstracts and full texts; the overall difference was not statistically significant. Particular comparisons of spin in results sections did differ significantly. This establishes that spin was present in every form they examined, and that some reporting patterns differed by form. It does not establish that compression caused the spin or that all infographics are worse. For a generated preview, the practical risk is still concrete: highlighting a positive subgroup while suppressing the non-significant primary finding changes the selection cue.
Nor can designers assume that people will stop at the intended boundary. The RIVA-C reporting-guideline project (Zadro and colleagues, 2024) was motivated in part by evidence that people sometimes use research infographics in place of the full article, although infographics may omit information needed for interpretation. Its output is a reporting checklist, not a trial showing that checklists or previews improve behaviour. The warning transfers to this design problem: a card labelled preview may be treated as the whole answer, so the interface must expose what the card cannot establish and make the unchanged source easy to open.
These studies do not show that AI-generated previews cause bad selections, or that well-designed ones prevent them. They show why engagement, preference, source fidelity and task performance are separate measurements. The proposed system needs its own experiment before anyone can say that compression improved the reader’s triage.
Three ways a preview could win If a visual preview does beat a title or an excerpt, it matters why, because three different victories hide inside one number.
It carries more task-relevant evidence. A card makes a specific affected version, comparison, exception or negative result visible when a generic excerpt buries it. Compare it with a text card carrying the same facts to avoid crediting the visual form for better information selection. It conveys the same evidence faster. A spatial arrangement lets a reader locate a relation sooner than prose. Compare matched factual content in text and visual layouts, measuring speed at comparable error rates. It is merely more attractive. Colour, polish and fluent labels attract attention or increase confidence without improving selection. Compare styled and plain versions with the same content; preference alone is not evidence that the reader chose better. A fair test therefore measures what was selected, what was missed, how long decisions took and how readers rated the form, without merging them into one score. It must also separate two different advantages: choosing better evidence to show and laying out the same evidence better. Otherwise the experiment will not tell us whether generation, content selection or visual design produced the apparent gain.
A proposed selection experiment Freeze a corpus of short documents, including routine advisories, relevant vulnerabilities, false alarms, qualified results and sources whose importance is only visible in a body paragraph. Give each reader a declared task and a fixed time or opening budget. Before any previews are generated, independent reviewers should inspect the unchanged documents, record task relevance and consequence of omission, and adjudicate disagreements. A document may legitimately be marked uncertain rather than forced into a relevant/irrelevant label.
Compare at least four conditions on the same source–task pairs: title and source only; title with an ordinary extracted snippet; a text-only task-conditioned preview; and a visual task-conditioned preview. An authored concise baseline, where feasible, strengthens the comparison. Hold source identity, access to the original and the selection budget constant. Where the hypothesis concerns layout rather than content, construct a matched text/visual pair carrying exactly the same adjudicated facts. Counterbalance order and reader assignment so that one condition does not merely inherit the learning of another.
Record consequential false skips, ordinary false skips, false opens, sources selected within budget, time to decision, source-opening behaviour, subjective preference and unwarranted confidence. Report denominators and uncertainty; inspect errors by task and document class. Before the run, declare the primary selection measure and any cost weighting. A headline improvement in clicks or satisfaction does not substitute for an improvement in correct selection under the budget.
This is a proposed experiment, not a result from this book. It could reject the visual approach altogether: a good excerpt or well-authored text card may be faster, safer and cheaper than generated imagery.
When the right preview is no preview Compression is an intervention, and interventions have to earn their place. The default alternatives include the original title, an extracted passage, a supplied abstract and no generated preview at all. Some sources should receive no additional visual rendering, and some should be marked insufficient evidence for a safe preview. The following cases are reasons to test abstention, not a claim that every instance of their class is impossible to compress.
An argument that is its hedges. In a methods critique, legal analysis or careful negative result, qualifications may carry the conclusion. A tiny visual card that drops them is misleading; an excerpt preserving the conditions, or the author’s own abstract, may be the better preview. If even that cannot preserve the selection-relevant point, leave the source uncompressed.
A distinction that is mostly numbers. Close hazard ratios, confidence intervals, absolute risks or budget percentages can be distorted by poorly scaled marks and areas. A compact exact-value table, with units and uncertainty intact, may outperform a decorative chart or an isolated number. Visual form is not forbidden; inexact visual encoding is the risk.
A short piece that its title already summarises. “Unauthenticated remote code execution in Acme Router firmware before 4.2” needs no cards. Generating a preview spends compute, attention and distortion risk to restate what a dozen words already decided.
A concept that invites a false picture. Familiar visual metaphors may import structure the source did not assert: evolution as a ladder, a protocol as plumbing, or risk as an unqualified traffic light. A generated image can make the wrong relation memorable. If the metaphor cannot be linked to supported claims and its limits shown, an exact excerpt is preferable.
These cases imply a selection policy: try the least transformative view that fulfils the task; generate a richer one only if it promises a measurable advantage and can pass the relevant preservation checks. Record when the system abstains and why. A system that cannot leave a page alone cannot be trusted with anything stronger.
For a security advisory, a preview’s minimum checkable fields might be the named component, affected version range, exploit preconditions, severity or impact as stated by the source, important uncertainties, publication or update date, and a link to the exact source passage. A shorter card may omit most background but must not silently change not affected into affected, turn under investigation into confirmed, or detach a version range from the component it qualifies. A model-generated relevance label is a proposal; checking against the user’s actual dependency inventory is a separate operation.
A preview is not a summary One last distinction keeps this chapter from spilling into the next. A preview is defined by its intended job: help someone decide whether to open a source. A summary is designed to communicate more of the source’s substance. They may look equally short, and users may treat either as a substitute for the original. Their preservation obligations therefore differ by intended use, not by visual appearance or a word count. A preview may omit background while retaining the cues needed for selection—topic, stance, task-relevant result, decisive qualification and provenance. It must not invent an unsupported implication in order to make the source look worth opening. The interface should clearly label its limited purpose and provide a direct route to the unchanged source.
Likewise the claim here stops at selection. Nothing in this chapter says a preview helps anyone understand, remember or apply what the page says.
Where this leaves us A compressed representation is a candidate for decision support, not an automatically legitimate substitute for reading. It earns its place when, for a declared task and budget, it improves selection while keeping consequential false skips visible and avoiding unsupported additions. Compression is not neutral shortening; attention gained is not understanding gained, correct inference or safe action. Information scent explains why a cue is followed, not whether it should have been. Attractive cards need an independent test, and leaving the source alone remains a serious baseline. The experimental question is whether a generated cue improves the chosen set of sources without creating a more persuasive route to the wrong ones.
The preview keeps the reader outside the page. The next question is what can be done once they are inside it: when does adding a diagram, a table or a timeline to the page itself help someone understand it?
References Pirolli, P. & Card, S. (1999). Information foraging. Psychological Review, 106(4), 643–675. Pirolli, P. (2005). Rational analyses of information foraging on the Web. Cognitive Science, 29(3), 343–373. Chi, E.H., Pirolli, P., Chen, K. & Pitkow, J. (2001). Using information scent to model user information needs and actions on the Web. Proceedings of CHI 2001. Chapman, S.J. et al. (2019). Randomized controlled trial of plain English and visual abstracts for disseminating surgical research via social media. British Journal of Surgery, 106(12), 1611–1616. Buljan, I. et al. (2018). No difference in knowledge obtained from infographic or plain language summary of a Cochrane systematic review: three randomized controlled trials. Journal of Clinical Epidemiology, 97, 86–94. Muller, R., Ferreira, G., Bejarano, G., Gamble, A.R., Kirk, J., Sindone, J. & Zadro, J.R. (2025). Do infographics “spin” the findings of health and medical research? BMJ Evidence-Based Medicine, 30(2), 84–90. DOI 10.1136/bmjebm-2024-113033. Zadro, J.R. et al. (2024). Development of the Reporting Infographics and Visual Abstracts of Comparative studies (RIVA-C) checklist and guide. BMJ Evidence-Based Medicine, 29(5), 342–345. DOI 10.1136/bmjebm-2023-112784.