What the Contract Can Hide

The price is in large type. The conditions arrive later.

A couple choosing a household service can compare the monthly charge without much difficulty. They can see the package, the equipment and the promised start date. The agreement is harder to compare. One provision concerns renewal. Another describes circumstances in which the provider can change something. A third explains how a dispute will be handled.

They would like to make a careful choice. They also need to finish before making dinner.

A familiar moral lesson is waiting for them: if they accept without reading, they should have known better. That lesson contains a small truth and hides a much larger design problem. Reading can help. It does not automatically provide the knowledge to interpret a term, the power to change it or the ability to go without the service.

We should ask what a fair agreement requires from both sides.

What a click can and cannot tell us

A digital record can show that someone clicked a button. Depending on the system, it might show that a page was opened or how long it remained visible. Those observations are useful, but they are not a full account of understanding.

A person may open a contract and read only the price. Another may read the document carefully and misunderstand a clause. A third may rely on advice from someone who inspected the same standard terms. Some people may accept after deciding that the available alternatives are no better.

This makes research on attention valuable and limited at the same time.

Bakos, Marotta-Wurgler and Trossen observed January 2007 software-shopping sessions. Very few included opening the linked agreement. Their access measure does not establish reading or understanding; the browsing definitions overlap.¹

Attention and reading burden

Observed contract access

Measure	Value
Scope	January 2007; software-retail sessions; accesses per 1,000
Visited 2+ pages	63 / 131,729 → 0.48
Visited 5+ pages	57 / 72,282 → 0.79
Reached checkout	7 / 4,866 → 1.44
Axis	0, 1, 2
Limits	Overlapping definitions. Access ≥1 second; not proven reading.

Source: Bakos et al., Table 3

Modeled privacy reading

Measure	Value
Scope	2008 scenario; hours per person yearly
Lower	181
Central	244
Upper	304
Axis	0, 175, 350
Limits	Each site’s policy read annually. Assumption bounds, not confidence limits or observed reading.

Source: McDonald–Cranor, Table 7

Figure 2. Different methods answer different questions.

McDonald and Cranor instead modeled a hypothetical annual privacy-policy reading task. Their 2008 calculation depends on assumptions about sites visited, policy length and reading speed. The displayed range varies assumptions; it is not a confidence interval.²

These historical panels cannot be combined into a current measure of consumer understanding. Neither observed access nor modeled reading time establishes the fairness of the terms.

The agreement nobody negotiates

There is a practical reason for standard terms. A provider dealing with many customers cannot negotiate every sentence separately. Customers may also prefer a quick transaction to a long discussion. Reusable terms can reduce work and make a service possible at a lower cost.

The problem begins when the convenience of a standard document is used to settle every question about its fairness.

Suppose a provider can revise an important feature with little notice, while the customer must pay a large cost to leave. The exact legal treatment would depend on the contract and jurisdiction. As a matter of design, however, there is a clear imbalance to inspect. One party can alter the arrangement more easily than the other can respond.

Or suppose a service makes a precise promise in its advertisement, then uses much broader language in the agreement. A careful reader may discover the gap. The organization should still explain the actual commitment before inviting a purchase. The burden of aligning the sales message and the terms belongs with the party producing both.

These are questions about the substance of an exchange. Enlarging the font helps people see the terms. It does not answer whether the terms should be there in that form.

I think clear language is a duty of fair dealing. I also think it is too small a remedy when the underlying bargain gives one side powers that the other cannot reasonably manage. A plainly written unfair term is easier to criticize. It remains a problem.

That is why a good reform discussion should separate three jobs: helping people understand, giving them a usable choice and setting boundaries on what can reasonably be demanded. Different institutions may carry out those jobs. Confusing them lets a modest improvement in disclosure stand in for a much larger change.

A reader needs a question

When faced with a long document, people are often told to read the whole thing carefully. For a significant commitment, that may be necessary, with qualified help where the stakes call for it. But the reader still needs to know what they are trying to understand.

A practical first pass asks what will happen over the life of the agreement. What will you pay, and when? What must each side do? How long does the arrangement continue? What can change? What happens if something goes wrong? How can it end?

These are orientation questions, not a substitute for advice about a particular legal obligation. They make the text easier to approach because they connect it to events a person can picture.

Take renewal. The word can describe a convenient continuation or a fresh commitment with consequences the customer did not expect. The reader needs to identify the timing, the price and the action needed to stop. A notice that merely says renewal applies does less work than an explanation tied to a date and a consequence.

Take a fee. The amount matters, but so does the event that triggers it. A low-looking price can be hard to compare if necessary charges appear at different stages. A person needs the total relevant to the service they actually intend to use.

Take a promise about availability. A general assurance may not tell you what happens during an outage, who must report it or what remedy is offered. The language becomes useful when it can answer a concrete question.

A fair provider should welcome such questions. It should be able to explain its own offer in ordinary terms and keep that explanation consistent with the agreement.

Plain language has to survive the example

Imagine a clause rewritten into shorter sentences. The writer removes old-fashioned expressions and breaks up a dense paragraph. This may improve comprehension. Now ask a reader to apply the clause to a situation.

If the customer cancels on a particular date, what happens next? If the provider changes the price, what choices remain? If an item fails during the stated period, who should be contacted?

The answers reveal more than a readability score. A person may recognize every word and still misunderstand how the conditions fit together. A document can be easy to read one sentence at a time and hard to use when two provisions interact.

Eric Martínez, Francis Mollica and Edward Gibson tested paired excerpts expressing the same intended contract content in plainer language and legal language. Participants understood and recalled the plainer versions better. The experiments concerned short texts, with limits including participant attrition; they did not test whether the underlying bargain was fair or what happened in actual purchases.³

This suggests a better test for organizations. Do not ask only whether people like the new wording. Ask whether they can accurately predict important outcomes from it. Include situations in which the answer is inconvenient to the provider. A summary that makes customers feel confident while leaving them mistaken is a poor summary.

The test should also include people who are tired, unfamiliar with the topic or using assistive tools. A document written for a relaxed expert may fail in the setting where it will actually be used. The provider has more opportunity to revise the document than a customer has to redesign the circumstances in which they must read it.

There is a limit to simplification. Some distinctions are real, and removing them can make the explanation misleading. A short summary can point to the full terms and make the most consequential parts visible. It should never quietly replace an important condition with a reassuring generality.

Good plain language keeps the difficulty that belongs to the subject and removes the difficulty created by the writing.

Someone can read on behalf of many people

Individual attention is not the only possible protection.

A purchasing team can inspect terms for an organization. A consumer group can compare a class of agreements. A union or professional association can help members understand recurring conditions. A specialist can notice a problem that a casual reader would miss.

The value comes partly from doing a difficult task once and sharing the result. It also comes from concentrating enough demand or public attention to make a provider care about the answer.

We should not romanticize this either. An intermediary may have conflicts, limited resources or a narrow view of the people it represents. Its advice can become outdated when the terms change. A comparison that overlooks important conditions may simply move the misunderstanding to another place.

Credible scrutiny needs a visible method. Which version was reviewed? What was compared? Who paid for the work? What remains uncertain? If the terms change, how will the conclusion be updated?

Those questions apply to automated summaries too. A tool may help a reader locate a topic, but a fluent summary can omit a condition or misstate how two clauses interact. The user should be able to trace an important claim to the actual text. High-stakes decisions call for appropriate expertise, not confidence produced by tidy prose.

The broader institutional point is that every household should not have to build a legal department before using ordinary services. Shared scrutiny can make routine transactions safer, provided the scrutinizer is accountable as well.

The missing option is a different provider

Competition can improve terms when customers can compare them and move. Both conditions deserve inspection.

Suppose two providers advertise similar prices but use different renewal dates, fees and cancellation steps. A comparison requires more than placing the prices side by side. It requires describing the same likely course of events under each agreement.

Now suppose the customer identifies a better offer. Moving may require new equipment, a gap in service or the transfer of records. A technically available alternative can carry a practical switching cost.

These possibilities do not prove that competition is absent or ineffective in every market. They explain why counting providers is only part of evaluating a person’s room to choose.

An honest comparison should include the cost of leaving the old service and beginning the new one. An institution that wants competitive pressure to work should care about understandable prices, usable transfers and terms that can be compared without private detective work.

We can also distinguish different kinds of choice. A person may care deeply about the product and have little view about the dispute procedure. A provider may compete intensely on the visible feature while keeping other terms difficult to compare. Making a hidden condition prominent can help, but it should be tested to see whether it changes understanding and behavior in the intended way.

There is no single cure here. Some terms need clearer explanation. Some need collective bargaining. Some may warrant legal limits, decided through the relevant public process. The choice of remedy should follow the specific problem rather than a general preference for either markets or regulation.

When the terms change after you settle in

The most revealing moment may come after the customer has built a routine around a service.

A new agreement arrives. It is longer than the old one, and the message says that continuing to use the service will have consequences under the new terms. The customer now needs to know what changed, when it takes effect and what choices they have.

A full replacement document is a weak way to explain a change if it forces every user to compare two long texts by hand. A clear account of material changes, linked to the relevant passages, would let people spend attention where something actually happened.

The provider also needs to preserve a record. If a later disagreement concerns what was promised at a particular time, both sides should be able to identify the applicable version and the notices given. This is a design proposal about traceability; it does not decide how any court would interpret a specific sequence.

A changing agreement should not be allowed to hide behind the familiarity of an unchanged service. The more a customer depends on the relationship, the more seriously the provider should take notice, transition and usable exit.

That duty is compatible with legitimate change. Costs rise. Features evolve. Old arrangements become impractical. The demand is for a fair way to handle the change, with enough information and time for the other side to respond.

A bargain worth accepting

The couple choosing their household service finish dinner later than planned. A better process would not necessarily have removed the agreement. It would have made the important commitments visible before the purchase, given them a reliable way to compare the offer and preserved a clear account of what each side accepted.

They might still choose a service with a limitation. Adults can accept trade-offs. The goal is to make the trade-off real enough to judge, rather than bury it under a ritual of consent.

I do not want a world in which every small purchase requires a negotiation. I want a world in which routine agreements deserve to be routine. Their safety should come from clear terms, reasonable boundaries and credible ways to correct failures, as well as from the customer’s care.

The organization drafting the agreement has already decided that the words matter. It should not act surprised when someone asks them to mean something fair.

Sources and limits