Excite, Lycos and HotBot are all online right now. All three load, two of them still have a working search box, and none of them has operated a web crawler since the early 2000s. They are not defunct and they are not search engines. They are domain names with commercial value, running somebody else's technology behind a familiar logo — which is a distinct and under-documented way for a search engine to end.
The pattern
A 1990s search brand is worth something long after its engine is worthless. The domain still receives type-in traffic and links accumulated over three decades; the name still carries recognition with people who were online in 1998; and search traffic, however modest, can be monetised through a syndication contract with almost no staff. Meanwhile the asset that made the brand famous — a crawler, an index, and the engineers who maintained them — is the expensive part, and it is the first thing to go when the company is sold, broken up or bankrupted. So the brand and the technology separate, and the brand is the half that survives, because it costs nothing to keep alive.
The result is a specific kind of misinformation. Lists of search engines still name these three, sometimes with descriptions of index sizes and crawling behaviour that were accurate in 1997 and have been wrong ever since. What a reader needs is not the history, flattering as it is, but a plain statement of what happens when they type a query today.
Excite
Excite started as Architext, founded in June 1993 by six Stanford students, and launched as Excite in October 1995. Its technology was genuinely distinctive: statistical analysis of how words co-occurred within documents, used to infer conceptual relationships between terms, so a search could retrieve documents about a concept rather than only documents containing the literal query string. This was marketed as concept-based search and it is why Excite was taken seriously alongside AltaVista rather than dismissed as another keyword matcher.
It then consolidated aggressively — buying the Magellan search engine in June 1996 and WebCrawler from AOL in November 1996, so that for a period one company ran three search brands — and merged with the cable broadband provider @Home Network in January 1999 in a deal valued in the billions. Excite@Home filed for Chapter 11 bankruptcy in October 2001, and the technology did not survive the asset sale that followed; what changed hands was the brand and the audience, not a working engine.
Today Excite is a portal shell. Its own privacy policy states that "Excite search services are provided through a third party, System1 LLC," and that a user performing a search is taken to a results page owned and operated by System1. There is no Excite crawler, no Excite user-agent and no Excite index. Which upstream engine sits behind System1's feed is not disclosed by either party. A small but telling detail: the results host disallows crawling in its own robots.txt, which is what a syndication front end does and what an index owner does not.
Lycos
Lycos has the strongest technical claim of the three. Michael Mauldin's spider, developed at Carnegie Mellon and running from 1994, did not attempt an exhaustive sweep of each server; it made a probabilistic walk over discovered references with a built-in preference for documents carrying many inbound links — an early, crude use of link popularity as a crawl-priority signal, years before link analysis was formalised as a ranking method. It stored a compact summary of each document rather than the full text: the title, the headings, the hundred most heavily weighted words, the first twenty lines, the byte size and the word count. It honoured the robots exclusion standard and identified itself to webmasters. Its index went from tens of thousands of documents at launch in July 1994 to hundreds of thousands within a month and tens of millions by late 1996.
Lycos retired that spider in the autumn of 2001 and began serving results from FAST, the engine behind AllTheWeb. It has taken partner results ever since, through an ownership chain that runs from a Carnegie Mellon spin-off to a NASDAQ listing, to Terra Networks in 2000, to the South Korean company Daum in 2004, to Indian ownership in 2010. No current, dated statement from Lycos naming its present results supplier could be found, and the site carries no "powered by" attribution anywhere — the load-bearing fact is simply that there has been no Lycos crawler since 2001. What is visible on the search property today is a promoted browser extension that changes the user's search settings, which places modern Lycos closer to the search-default distribution business than to search.
HotBot
HotBot is the clearest case of a brand that was never the technology in the first place. It launched in May 1996, built by HotWired — the online arm of Wired magazine — as the shop window for an engine belonging to Inktomi, a startup founded in January 1996 by a Berkeley professor and a graduate student. Inktomi's contribution was architectural rather than algorithmic: it ran search across a coupled cluster of ordinary commodity workstations in parallel, where AltaVista depended on a single very expensive high-end machine. That bet — scale out on cheap hardware rather than scale up on one big box — is the architecture the entire industry subsequently adopted, and HotBot's marketing pitch of crawl freshness, claiming to index the whole web weekly, was the visible consequence of it. A click-popularity signal from Direct Hit was layered on in February 1999, an early use of behavioural data in ranking.
By the end of 2002 HotBot had been relaunched as a chooser that let the user pick which of four other engines to search — at which point it was a switchboard, not an engine. Yahoo bought Inktomi in March 2003, removing the original technology supplier from the independent market. Lycos, which had owned HotBot since acquiring Wired Digital in 1998, sold the domain in October 2016 for a documented $155,000, severing the brand from search infrastructure entirely.
What hotbot.com is today has nothing to do with any of that. It is operated by an unrelated company as an AI chat aggregator, offering access to a number of third-party large language models through their official APIs. Its terms mention search engine services, but no index, crawler or search partner is disclosed anywhere on the site. Someone arriving because they remember HotBot from 1997 will find a product that shares only the name.