How to Research a Tech History Topic Using Primary Sources

Every dated claim on this blog’s Tech History category is checked against a source like the ones below before publication — this walks through the actual research process, useful for verifying any tech history claim you encounter elsewhere.

Step 1: identify what kind of claim you’re actually checking

A date and event (“Company X launched Product Y on this day”) needs a different kind of source than a disputed attribution (“Person X said quote Y”) or a causal claim (“Event X caused Event Y”) — knowing which kind of claim you’re checking determines which primary source will actually settle it.

Step 2: for corporate events, check SEC filings directly

sec.gov/cgi-bin/browse-edgar → search the company name

For any US-listed public company, SEC filings (10-Ks, 8-Ks, IPO prospectuses) are primary, legally-binding documents — considerably more authoritative than a news article’s paraphrase of a corporate announcement, and directly searchable by company name through EDGAR.

Step 3: for legal disputes, check the actual court opinion

Justia, CourtListener, or the U.S. Copyright Office's own
  fair-use case summaries host full text of many
  significant rulings

A court’s own written opinion is the actual primary source for what a legal case decided — considerably more precise than a news summary, which can (and, in tech history retellings, often does) compress or distort the actual legal reasoning.

For anything beyond the final published opinion — motions, exhibits, transcripts, the docket itself — United States federal court records live in PACER, which charges per page; the practical free path is RECAP, the Free Law Project’s public mirror integrated into CourtListener, which republishes millions of documents that other researchers have already paid to retrieve. Tech history litigation is unusually well covered there, precisely because these cases attract researchers. Google Scholar’s case-law search covers published opinions going back far enough to include the disputes this category cares about. The general habit: cite the court’s own words with a docket or reporter reference, and treat any journalistic summary of a ruling as a pointer to the primary document, not a substitute for it.

Step 4: for a disputed quote, look for the original recording or transcript

If a quote is widely repeated but the underlying transcript or recording isn’t cited anywhere, that’s itself a warning sign — the “Al Gore invented the internet” and “Bill Gates 640K” cases are both examples where checking for an actual primary transcript reveals the popular version doesn’t match, or has no verifiable source at all.

Step 5: for a product launch date, check contemporaneous announcements

Search the official organization's own press releases
  or archived contemporaneous news coverage from
  the actual date in question, not a much later retrospective

A contemporaneous announcement (published at or near the actual event) is a stronger source than a retrospective article written years or decades later, which can introduce errors through imprecise memory or accumulated retelling.

Step 6: for early internet-era claims, search the Usenet archives

Google Groups (holds decades of archived Usenet posts)
Internet Archive's Usenet collections (including the
  early UTZOO tape archives)

A surprising amount of computing history was announced, in the protagonist’s own words, on Usenet — the canonical example being Linus Torvalds’ August 1991 post to comp.os.minix describing his hobby operating system project. Google Groups still holds a vast (if imperfectly searchable) archive of these posts, and the Internet Archive hosts early Usenet preserved from the UTZOO tapes. A dated, first-person announcement post is as primary as sources get for this era; when one exists, it outranks every later retelling — including the author’s own later memories, which is exactly the same caveat that applies to oral history interviews conducted decades after the fact.

Step 7: cross-reference at least two independent sources

A single source, however seemingly authoritative, can contain an error — checking a date or fact against at least two independent sources (ideally including at least one primary source) catches the cases where one source’s mistake would otherwise go unnoticed.

Step 8: note explicitly when a fact is genuinely disputed or uncertain

Not every historical question has a single, clean, universally agreed-upon answer — the “who invented email” question is a case where the honest answer involves real nuance (distinguishing single-machine from networked messaging) rather than a simple one-line claim, and a careful researcher should represent that nuance rather than flattening it for a cleaner-sounding story.

Why this process matters more than any single technique

No individual step here is complicated on its own — the actual discipline is consistently preferring a primary or contemporaneous source over a secondhand retelling, every time, even when the secondhand version is more widely repeated, more dramatic, or more convenient to cite without further digging. Applied consistently, the process also compounds: every claim you verify leaves you holding the primary document for the next question that touches the same company, case, or era.

Related:

Sources: