Public Web Experience Audit of 2IA.org

Full report

On this page

Scope and limits

This audit was conducted as a public-web experience review of https://2ia.org/ using ordinary browsing only. I reviewed the homepage, Current Developments, Start Here, Research Archive, Methodology, About, Contact, Corrections and Right of Reply, Lawful Contact, Privacy Policy, Newsletter, and Support pages on 2IA, and the homepage, Intelligence Desk, and a sample research page on the sister site https://internationalintelligence.org/. The review relied on public page content, visible navigation, public empty/error states, and publicly exposed labels and timestamps where available. It did not use source code, internal analytics, network logs, private credentials, repositories, CMS access, or intrusive testing.

The review has important limits. I could directly observe page structure, labels, repeated modules, empty states, and public disclosures, but I could not independently verify runtime network calls, JavaScript-disabled behavior in a lab browser, CSS focus-ring styling, exact mobile viewport breakpoints, or screen-reader announcements. Where something is inferred rather than directly observed, I label it as an inference. I do not claim formal WCAG conformance or non-conformance.

Device and task matrix

Accessibility findings

Perceivable

Observation. In sampled pages, heading structure is generally logical: a page-level # heading followed by ## and ### subsections. Skip links are exposed at the top of multiple pages. That is a good baseline for structure and bypass navigation.

Observation. Descriptive page-level text is strong, but some non-text content is not reassuringly described. Repeated image placeholders appear simply as “Image” in sampled pages, which does not demonstrate meaningful alternative text for informative images.

Observation. Dense paragraphs and long scroll depth remain a significant reading burden. The homepage runs to 798 lines in the public rendering used here, and the Research Archive to 765 lines, with many long prose blocks and repeated footer sections.

Inference. Because most pages are linear rather than multi-column, the site may reflow better than many magazine-style layouts under text enlargement. However, that same linear design likely increases fatigue for low-vision and zoom users because of page length and repeated modules. This was not directly lab-tested at enlarged zoom levels.

Unverified condition. Color contrast was not independently measurable in this review environment, and the site’s public claims about no external assets do not establish text/background contrast on their own.

Recommendation. Keep the clear heading hierarchy, but shorten leading prose, convert some long narrative stretches into tighter paragraphs, and audit all non-decorative images so that informative graphics either receive useful alt text or are marked decorative when they add no essential meaning.

Operable

Observation. Keyboard-oriented affordances are present in the public output: skip links, breadcrumbs, repeated navigation landmarks, and Back to top links all appeared on sampled pages.

Observation. Menu controls are visible as “Toggle menu,” and some pages also expose dismiss/close symbols (×) near the top.

Inference. The repeated navigation, utility bar, and footer blocks make serial keyboard traversal longer than it needs to be, especially for readers who move through headings and links linearly.

Unverified condition. I could not independently verify visible focus indicators, whether the menu can be fully operated by keyboard alone, or whether top-of-page dismissible UI behaves like a proper non-trapping disclosure rather than a modal. I also could not test whether the search field is fully operable with keyboard submission in the live browser.

Observation. No heavy motion, autoplay, or intrusive overlays were evident in the public text rendering. The site also exposes a no-script fallback on the homepage and states that Current Developments uses server-side ingestion with no browser-side feed calls.

Recommendation. Verify menu and utility-bar controls with a real keyboard pass, ensure strong focus visibility on all interactive elements, and reduce traversal burden by making repeated utility content more collapsible or less repetitive.

Understandable

Observation. Homepage comprehension is one of the site’s strengths. The homepage states what 2IA is, what it covers, and what a first-time reader can do next. The Start Here page further explains the name, purpose, and reader pathways.

Observation. Navigation is highly consistent. Across sampled pages, the same primary items recur: Home, Developments, Start Here, Intelligence Library, Research Guides, Research Briefs, Methodology, Topic Hubs, About, and Contact, with consistent footer groupings for Trust and standards, Participate, and Contact.

Observation. The archive/current boundary is handled thoughtfully. 2IA repeatedly identifies itself as the archive and labels International Intelligence as the sister publication focused on news and real-time intelligence. The Current Developments page warns that a feed records publication activity but does not by itself settle the underlying claim.

Observation. Search is visibly available across important pages, but its real usefulness was not established here. What I could observe is a recurring “Search 2IA research” control with a submit action; I could not fully execute and assess search results through this browser environment.

Observation. Correction language is unusually good: it specifies what evidence to send, what counts as a correction, what right of reply means, and how the public log should work.

Issue. Contact is less understandable at the last crucial step. The site explains boundaries and process in detail, but sampled Contact and Corrections pages do not clearly show a direct destination such as an email address, form action, or secure channel.

Recommendation. Preserve the strong editorial clarity, but add explicit page metadata, simple search help, and one plainly visible correction/contact destination so the reader’s next step is never ambiguous.

Robust

Observation. The site repeatedly describes itself as a plain PHP release with no database and no browser-side third-party dependencies, and the Current Developments page describes same-origin cached snapshots rather than browser-side source fetching. The homepage also includes a no-script fallback. These are positive public robustness signals.

Observation. The Current Developments page is robust in one important editorial sense: on failure it prefers an empty, labeled seed over simulated “live” content, and it says failed refreshes should preserve the last successful snapshot rather than replacing it with something misleading.

Unverified condition. Those robustness claims were disclosed, not independently verified through source inspection or runtime monitoring. I could not test markup validity, ARIA use, actual feed generation, or browser compatibility across assistive technologies. The linked Atom feed also could not be fetched in this review tool, so feed robustness remains unconfirmed.

Recommendation. Treat the public transparency as a good foundation, then validate it with a visible feed-health check, feed-reader testing, and a public “last generated / last validated” signal for both the HTML developments page and the Atom feed.

User experience and current developments

User experience review

The site is unusually explicit about its identity, methods, and reader purpose. For a first-time reader, that is a strength. 2IA states its editorial stance, distinguishes archive work from operational or activist identity, and provides multiple reader entry points such as Start Here, Methodology, Research Archive, and Request Records.

Navigation consistency is also strong. The same core menu and footer taxonomy appear on sampled pages, which supports orientation and backtracking. Breadcrumbs help as well.

The main user-experience weakness is burden, not ambiguity. Pages are long, often densely written, and repeatedly end with large sitewide blocks. For readers on mobile, at higher zoom, or using keyboard traversal, that creates fatigue even when the content itself is clear.

Related-content quality is generally good when present. The dossier page I sampled links onward in a structured way through “Keep Reading,” “Related route,” and source references, which supports orientation. At the same time, generic link labels such as “Open this page →” on the sister publication are less descriptive out of context than the stronger action labels used on 2IA.

Download and print usability are present at least in some interactive reader paths: the homepage’s pressure-map area exposes “Copy pressure map,” “Download text,” and “Print-friendly version.” That is a strong pattern, but it was only directly observed in one area rather than as a sitewide standard.

Empty-state honesty is a notable positive. On 2IA, the developments page openly reports “Awaiting first refresh.” On the sister publication, the Intelligence Desk exposes a plain “No items match those filters” result. Both are better than silent blanks.

Current-developments assessment

The public developments implementation is editorially careful, but operationally incomplete at the time of review. The page states that every entry should name source, evidence class, publication date, and freshness state; it explains source boundaries; it names configured sources and publishers; it discloses a 30-minute cache window; and it says failed refreshes preserve the last successful snapshot. But the page simultaneously shows 0 current entries, 0 healthy sources, and no last successful refresh, which means readers cannot yet benefit from those promises in practice.

Observable answers to the user’s timestamp and status questions

Readers can currently see the source publisher, cache window, and overall freshness state on the developments page. Readers cannot yet see an item’s original publication time, retrieval time, or last successful update because no live item cards are present and the page explicitly says no successful refresh has occurred. The page does clearly state that source-published material is not automatically independently verified, and it says a failed refresh should retain the last successful snapshot, though that retention behavior is described rather than visibly demonstrated with a real retained item in this empty state.

The site does a strong job separating live/current material from evergreen/archive work in principle. The homepage says 2IA remains the archive, while International Intelligence is the sister publication for news and real-time intelligence. But because the developments page is still empty, the practical usefulness of that separation for current readers is limited right now.

The first-party feed is present as a public promise and navigation item, but this review could not independently confirm feed-reader usability because the linked Atom URL failed to fetch in the review browser. That should be treated as unverified, not as a confirmed broken feed.

Privacy and trust assessment

Visibly verified

Sensitive-submission warnings are prominent and repeated. Contact-related pages plainly tell readers not to send secrets, classified material, stolen data, credentials, exploit code, or unnecessary personal information, and they explicitly state that ordinary contact is not a secure drop.

Reading public material does not visibly require an account. The privacy policy states there is no visitor login or application cookie in this release, and the sampled pages themselves were publicly readable without any account gate.

Newsletter signup and donation processing were not active on the sampled public pages. The newsletter page says signup is not active in this source-controlled site, and the support page says donation processing is not enabled in this release.

A correction route exists in policy terms. The corrections page explains what a requester should send, how review is supposed to work, and what a public correction log should contain, while also stating that no public corrections are logged yet in the source-controlled site.

Disclosed but not independently verified

The privacy policy discloses that this release does not add tracking pixels, external fonts, CDN assets, heatmaps, third-party analytics, or unnecessary cookies, and the footer repeatedly states “No browser-side third-party dependencies.” Those are important claims, but this review did not independently verify network requests or runtime dependencies. They should therefore be treated as disclosed, not confirmed.

The site also discloses that hosting may keep ordinary server logs and that local storage is used only for lightweight interface state such as dismissing a notice. Again, those are disclosures rather than independently verified behaviors in this audit.

The developments system discloses that visitor requests do not contact source publishers and that the public site reads local snapshots instead, but this too was not independently network-verified in the review environment.

Ambiguous

Contact and correction routing are substantively strong but operationally ambiguous. The Contact page speaks of “Contact Paths,” the Lawful Contact page says Contact is the “actual path,” and the corrections page explains the request process, yet the sampled public pages did not expose a visible email address, mailto: link, or published encrypted route.

Search is consistently present, but public usefulness and accessibility of actual search results were ambiguous in this review because I could observe the control, not meaningfully exercise the full search journey.

Potentially misleading

The clearest “potentially misleading” pattern is the phrase that Contact provides the practical route while the sampled Contact page reads more like a boundary-and-policy page than an actionable submission endpoint. That mismatch is not deceptive in intent, but it can frustrate a reader who has already followed the correction/contact flow and expects a concrete destination.

The “Local Atom Feed” link is publicly exposed in navigation and on the developments page, but feed usability could not be confirmed because the review browser failed to fetch the Atom URL. I am not asserting the feed is broken for all readers; I am asserting that compatibility was not established and the public promise therefore remains incomplete from a reviewer’s perspective.

Not testable without internal access

Actual server-log retention, spam-protection implementation, cache invalidation behavior, underlying update schedules, third-party request behavior at runtime, security headers, processor arrangements, and any unpublished correction-workflow tooling are not testable without internal access or runtime inspection beyond this review’s scope. The privacy and developments pages discuss these areas, but discussion is not verification.

External relationship assessment

2IA does a good job explaining what the sister publication is. On the homepage, in page utility bars, and in footers, it labels International Intelligence as a sister publication and explicitly says it focuses on news and real-time intelligence. It also marks the link with an external arrow and says the relationship does not make every sister-site claim a finding of the archive. That is excellent boundary-setting and strengthens trust.

This relationship generally strengthens the product family rather than undermining it. 2IA remains strongly branded as an archive, methodology-forward reference space, while the sister site presents itself as a research desk/intelligence desk with a current cache, search filters, and live-ish desk orientation. Readers can distinguish the two editorial roles from the labels alone.

The weak point is the return journey. In the sampled International Intelligence homepage, desk page, and research page, I did not find a visible mention of 2IA, and searches within the sampled sister pages returned no match for “2IA.” That means the forward link from archive to sister site is clear, but the reverse link from sister site back to archive is not comparably legible.

The result is not identity collapse, but asymmetric family discoverability. Readers are unlikely to confuse 2IA with International Intelligence once they are on 2IA. They are more likely to lose the family relationship after they leave 2IA for the sister site. That is a fixable UX gap rather than a branding failure.

Prioritized defect list and thirty-day improvement program

Prioritized defect list

1. Direct contact/correction destination is not visibly actionable

URL: https://2ia.org/contact/ and https://2ia.org/corrections-and-right-of-reply/ Screen or task: Contact and correction submission Observed behavior: The pages explain what to send and what the process should be, but sampled public pages did not expose a visible email address, mailto: route, or encrypted contact channel. The Contact page promises “Contact Paths,” and Lawful Contact says Contact is the actual path, yet the actionable endpoint is unclear. Affected readers: People requesting corrections, right of reply, media contact, collaboration, or support; journalists checking an error; vulnerable submitters looking for a safe boundary. Severity: High Recommended outcome: Publish one direct ordinary-contact route, one direct correction route, and—only if genuinely available—one secure route, each with clear scope and response expectations.

1. Current Developments is transparent but not yet operationally useful

URL: https://2ia.org/developments/ Screen or task: Reader seeking current developments Observed behavior: The page shows 0 current entries, 0 healthy sources, and no last successful refresh, while presenting detailed metadata architecture for a feed that has not yet visibly run. Affected readers: Readers looking for current developments, beat reporters, researchers seeking freshness signals. Severity: High Recommended outcome: Make at least one successful refresh visible, expose item-level timestamps and freshness labels, and keep the honest empty-state language only when the pipeline is truly empty.

1. First-party Atom feed compatibility was not established

URL: https://2ia.org/developments.atom Screen or task: Feed subscription Observed behavior: The site links to a Local Atom Feed, but the review browser failed to fetch the feed URL. This does not prove a universally broken feed, but it does mean ordinary-public compatibility was not independently confirmed. Affected readers: Feed-reader users, newsroom monitors, low-distraction current-awareness readers. Severity: Medium-High Recommended outcome: Validate the Atom feed publicly, expose a “last validated” signal, and provide a human-readable feed help note on the developments page.

1. Long, repetitive page structures increase reading and keyboard burden

URL: https://2ia.org/ and https://2ia.org/research-archive/ Screen or task: First-time reading, mobile reading, keyboard traversal, zoomed reading Observed behavior: Sampled pages are very long and end with repeated utility/footer modules; the homepage and archive pages are especially heavy. Affected readers: Mobile readers, low-vision readers, keyboard-only readers, readers with attention or fatigue constraints. Severity: Medium Recommended outcome: Shorten the top third of major pages, suppress or collapse repeated boilerplate, and move lower-priority modules behind clearer “More” breaks.

1. Meaningful image alternative text is not established

URL: Sampled across https://2ia.org/ and https://2ia.org/corrections-and-right-of-reply/ Screen or task: Reading pages with images using assistive technology Observed behavior: Repeated image placeholders appear as “Image” in the public rendering, which does not demonstrate useful alt text for informative images. Affected readers: Screen-reader users, low-vision readers, text-browser users. Severity: Medium Recommended outcome: Review every image and ensure the alt text is either meaningfully descriptive or empty when the image is purely decorative.

1. Article-level dating is not consistently visible on sampled 2IA archive content

URL: Example sample https://2ia.org/false-positives/trace-the-error-chain/ and https://2ia.org/about/ Screen or task: Journalist or researcher citing a page Observed behavior: In sampled 2IA pages, visible “Published” or “Updated” metadata was not readily apparent near the title, even though the archive concept itself values evidence and updates. Affected readers: Journalists, researchers, citation checkers, returning readers judging staleness. Severity: Medium Recommended outcome: Show publication and update dates prominently near every page title and in archive cards, not just in search snippets or conceptual descriptions.

1. Sister-site return path is weak

URL: https://internationalintelligence.org/ and https://internationalintelligence.org/desk.php?lang=en Screen or task: Moving from 2IA to its sister publication and back Observed behavior: 2IA clearly labels the sister link, but sampled sister-site pages did not visibly mention 2IA or provide an obvious reciprocal archive link. Affected readers: Readers trying to distinguish archive work from current-desk work and return to source-preserved archive material. Severity: Medium Recommended outcome: Add a persistent reciprocal family label such as “Back to 2IA Archive” with a one-line explanation of the editorial distinction.

1. Search usefulness is present in concept but not established in practice

URL: Sampled across major pages, including https://2ia.org/developments/ and https://2ia.org/research-archive/ Screen or task: On-site search Observed behavior: “Search 2IA research” appears consistently, but public usefulness, query help, accessibility, and result quality were not established through this review. Affected readers: First-time readers, keyboard users, readers seeking a specific topic quickly. Severity: Medium Recommended outcome: Add visible search guidance, example queries, and a public no-results/help state that can be reviewed without guesswork.

Immediate trust and accessibility repairs

Within the first ten days, the highest-value repairs are the ones that convert policy clarity into usable pathways. Publish a direct correction/contact destination, confirm the first-party feed publicly, and add conspicuous page-level publication/update metadata. At the same time, run a focused accessibility pass on skip links, focus visibility, image alt text, and the top utility/menu controls because those are the areas where the site already shows promising structure but lacks enough public proof of execution.

Navigation and content-presentation improvements

Over the next ten days, reduce burden rather than rewriting the editorial voice. Shorten page intros, compress repetitive modules, keep the first action choices close to the top, and make external link treatment more consistent. Search should receive visible help text and a reviewed empty-state. Any page that expects citation or return visits should surface “Published,” “Updated,” and “Last reviewed” metadata prominently.

Current-developments and sister-site refinements

In the final ten days, close the gap between the archive and current products. Make the developments pipeline visibly live or keep it clearly pre-launch, but do not leave it in a half-usable state. Expose feed status, last successful refresh, and item-level retrieval/publication times once entries exist. On International Intelligence, add a reciprocal family cue back to 2IA so the archive/realtime distinction works in both directions, not just from 2IA outward.