Digital Accessibility Must Preserve the State of Our Work

Author: Joshua W.J. Brown
Published: 28 Aug 2026
Publication Type: Submitted Article

Table of Contents:
Synopsis - Definition - Overview - FAQs - Insights, Updates - Related Content

Synopsis

This report argues that digital accessibility must extend beyond interface conformance to include what the author calls procedural memory - the reliable preservation of the state of a person's work as it moves through portals, forms, complaints, automated support and AI assistants. Key facts center on the difference between events such as drafted, attempted, sent, delivered, received, accepted for intake, escalated, superseded and closed, and how many systems collapse these distinct states into a single vague confirmation. This affects people with executive-function, memory, attention, fatigue or communication barriers, along with seniors and others who must otherwise reconstruct administrative history themselves through screenshots, old emails and repeated explanations. It matters because when a system fails to preserve state, it transfers hidden compensatory labour back onto the user, turning the person into the system's memory. The main finding is that conversational continuity in AI is not the same as preserving procedural state, and that a visible, inspectable state ledger showing what happened, when, its current status and who owns the next action would benefit everyone while removing real disability barriers - a perspective made authoritative by the author's lived experience and three and a half years building an AI-assisted continuity system of his own.*

At a Glance

• 1 - A service is a sequence of actions over time, not a single screen.
• 2 - A website can pass an accessibility test yet still force a disabled person to redo the same work three times.
• 3 - A state ledger should preserve history rather than silently rewrite it, so attempted never quietly becomes sent.
• 4 - When two cases are duplicates, a humane system should show which record survives and why, instead of leaving the user to guess which reference number still controls.

Topic Definition

Procedural Memory

Procedural Memory, in the context of digital accessibility and administration, refers to a system's ability to record and preserve the true state of a person's work as it progresses through a process rather than only capturing the words exchanged along the way. It means keeping an accurate, inspectable history of what action occurred, when it happened, what its current status is, whether an earlier submission was superseded, whether a delivery failed, and who or what currently owns the next step. Unlike conversational continuity, which can summarize a discussion and sound attentive, procedural memory tracks what actually happened to a form, message, document or case, so a user is never forced to reconstruct their own administrative history from screenshots, timestamps and repeated explanations. Treated as an accessibility requirement, it reduces the hidden compensatory labour that badly designed systems push onto people with memory, attention, fatigue or executive-function barriers, as well as seniors and anyone navigating complex institutional processes.

Overview

Accessibility Does Not End at the Interface: Why Digital Systems Must Preserve the State of Our Work

A website can pass an accessibility test and still make a disabled person do the same work three times.

That is the gap I keep running into as more of life moves through portals, automated support systems, AI assistants, email workflows and digital forms. We have become much better at asking whether someone can perceive and operate an interface. We are much worse at asking what happens to the person's work after they use it.

The distinction matters because digital administration is made of states.

A form can be drafted. A message can be attempted but fail to send. A complaint can be sent, delivered, received, accepted for intake, escalated, superseded by another case, or closed. These words describe different events. Yet many digital systems flatten them into one vague experience: a confirmation screen, a reference number, or a conversational history.

When that happens, the system has not eliminated labour. It has transferred the labour back to the user.

The Hidden Accessibility Burden is Reconstruction

For a person dealing with executive-function, memory, attention, fatigue, communication or other access barriers, reconstructing administrative history can be expensive work.

Imagine sending supporting documents and receiving an automated acknowledgement. Later, a staff member says the file is incomplete. Was the document never delivered? Was it delivered but not attached to the case? Was it attached but not accepted? Is the staff member looking at a duplicate case? Did an earlier escalation create a new reference number without preserving the old one?

Those questions are not semantic trivia. They determine what the person must do next.

If the system does not preserve the answer, the user becomes its memory.

The person searches old email, screenshots portals, compares timestamps, copies case numbers, explains the same chronology to another employee and resends evidence that may already be sitting somewhere inside the organisation. What looks from the outside like persistence is often compensatory labour created by poor information architecture.

This is why I have become convinced that accessibility must include procedural memory.

Conversational memory is not procedural memory

The arrival of generative AI makes this problem easier to hide.

An AI support system can produce a fluent summary of a conversation. It can remember that a customer mentioned an earlier complaint. It can sound continuous and attentive. But conversational continuity is not the same thing as preserving procedural state.

A useful administrative system must know more than what was said. It must know what happened.

Was the attachment received? Which case is controlling? Did the earlier case close because the issue was resolved, or because it was marked duplicate? Was the matter escalated to a person with authority to decide it? Is another response required from the user, or is the next action now the organisation's responsibility?

If those distinctions are not represented explicitly, an AI layer can make the system feel smarter while leaving the underlying accessibility problem untouched.

Build a Visible State Ledger

A better design is not especially mysterious. I think every consequential digital service should expose a simple state ledger that both the organisation and the user can inspect.

At minimum, it should show:

• The action that occurred
• The date and time
• The current status
• Whether an earlier case or submission was superseded
• Whether delivery failed
• Who or what currently owns the next action
• What the user must do next, if anything

The ledger should preserve history rather than silently rewriting it. If an email fails, "attempted" should not become "sent." If a complaint receives a reference number, "received" should not become "accepted for intake" unless someone or some validated process actually accepted it. If two cases are duplicates, the system should show which one survives and why.

This is good administration for everyone. For disabled users, it can be the difference between completing a process and being trapped inside it.

Accessibility standards are a beginning, not an endpoint

The Web Content Accessibility Guidelines (WCAG) provide an essential foundation for making digital content perceivable, operable, understandable and robust. They remain one of the most important accessibility frameworks in the world. Source: World Wide Web Consortium, WCAG 2.2: https://www.w3.org/TR/WCAG22/

But an accessible button cannot compensate for an institution that forgets what happened after the button was pressed.

That is not a criticism of WCAG. It is a reminder that accessibility is larger than interface conformance. A service is a sequence of actions over time. If the user can access every screen but cannot reliably determine the state of the process, the service can still produce disability barriers.

My Second Brain Taught Me What Institutions Forget

For roughly three and a half years I have been building what I call a second brain: an AI-assisted system made from my own writing, research, notes, terminology, source records, project history and decisions.

I do not use it because I want a machine to become me. I use it because I need continuity. It helps me retrieve work I have already completed, externalise executive load, compare current information with older records and stop complex projects from resetting every time I move between systems.

The most useful lesson has not been that AI can write. It has been that state is evidence.

Knowing what changed, what did not change, what was superseded, what remains unresolved and who must act next can matter as much as the content of any single message.

Institutions should learn the same lesson.

Stop Making the Person Carry the Database

Disabled people are often praised for resilience when what we are actually demonstrating is repeated recovery from badly designed systems.

There will always be legitimate reasons to verify information. There will always be mistakes, security checks, appeals, revisions and human judgment. The answer is not to eliminate verification.

The answer is to stop making verification destroy the record it is supposed to protect.

A humane digital system should be able to say: We have this. We do not have that. This earlier record was replaced by this one. Your submission was received but has not yet been accepted. Nothing further is required from you right now. The next action belongs to us.

That is not a luxury feature.

It is accessibility.

Author Bio

Joshua W.J. Brown is a Toronto-based Canadian writer, filmmaker and independent researcher. His work spans disability, technology, institutional language, evidence and public systems. His writing has appeared in Canadian Dimension, The Galway Review and A Thin Slice of Anxiety, with additional work forthcoming.

Frequently Asked Questions

NOTE: Researched FAQs by Disabled World (DW)

What is the difference between conversational memory and procedural memory

Conversational memory recalls what was said in a discussion and can summarize it fluently, while procedural memory records what actually happened to a form, document or case, including its status and who owns the next action.

Does passing a WCAG audit mean a service is fully accessible

No, because meeting the Web Content Accessibility Guidelines confirms that screens are perceivable and operable but does not guarantee that the system preserves the state of a user's work over time, which can still create disability barriers.

What is a state ledger in a digital service

A state ledger is a simple, inspectable record that shows each action taken, its date and time, its current status, whether an earlier submission was superseded, whether delivery failed, and who is responsible for the next step.

How does poor information architecture affect disabled users specifically

When a system flattens distinct events into one vague confirmation, users with memory, attention, fatigue or executive-function barriers must reconstruct their own history, which shifts hidden labour back onto the people least able to absorb it.

Can generative AI solve the accessibility problems described here

Not on its own, because an AI layer can make a system feel smarter and more continuous while leaving the underlying procedural state unrepresented, which means the accessibility gap can remain hidden rather than resolved.

Why should organizations preserve failed or superseded records instead of overwriting them

Preserving history lets both the organization and the user see what truly occurred, so an attempted send is not mistaken for a completed one and duplicate cases can be reconciled without forcing the user to guess.

How can seniors benefit from procedural memory in digital systems

Seniors navigating portals, benefits applications or support systems gain a clear, reliable trail of what was submitted and what remains outstanding, reducing confusion, repeated effort and the need to re-explain their situation.

Insights, Analysis, and Developments

Editorial Note: The piece reframes a familiar praise narrative, noting that what is often celebrated as disabled resilience is really repeated recovery from poorly built systems, and it points squarely at institutions rather than individuals as the party that should change. Its central proposal is modest and concrete: verification, security checks and human judgment are legitimate, but they should never destroy the record they exist to protect, and a well-designed system should be able to state plainly what it holds, what it does not, which earlier record was replaced, and where the next action sits. Written by Toronto-based writer, filmmaker and independent researcher Joshua W.J. Brown, whose work on disability, technology and public systems has appeared in Canadian Dimension, The Galway Review and A Thin Slice of Anxiety, the argument draws authority from lived experience and grounds its accessibility framing in the Web Content Accessibility Guidelines while making the case that conformance is a starting point rather than a finish line.*

* Editorial additions by Ian C. Langtree.