very little was needed to fully optimize on page content in the eyes of Google.[48] With an inverse relationship between query position (in the HTML tag hierarchy) and weight, it simply boiled down to putting one’s important keywords higher on the HTML tag hierarchy.[49] With this relatively simple algorithm in place, webmasters quickly discovered tricks to vastly boost their SERP (Search Engine Ranking Position). One of the first instances of deceitful SEO came in the form of link keyword stuffing under on page content. As a response, Google removed these sites from its index, a practice it occasionally reverts to as a way of punishing disingenuous webmasters looking to cheat the system. The seemingly arbitrary changes to Googles PageRank algorithm however, led to both community outrage (“many declaring the “death of PageRank”) and a noticeable decline in precision of search results.[50] Google was also constructing its index via a large monthly crawl. Not only did this lock search results to this one month window, it also meant that results would show stale content. An update dubbed “Everflux” introduced fresh crawling (daily crawls) to supplement the larger, primary crawl.[51] Daily crawling added another layer of relevancy (based upon date and time), to content ranking.[51] The inconsistencies of index versions across data centers during the early implementation of Everflux panicked webmasters, who saw their SERPs fluctuate wildly from day to day.
Boston/Cassandra/Dominic (2003) Edit
Google's “Boston” update in February 2003 saw major algorithmic changes and the promise of frequent index updates.[52] “Cassandra” marked a much more aggressive attack on shady SEO techniques like hidden and disguised keyword links, by emphasizing link quality[52] This was taken a step further in “Dominic”, which sought to analyze the quality of all backlinks to prevent the then emerging practice of splogging (the practice of creating nonsensical offsite content to boost SERP of another site).[53] To combat practices like “Googlebombing” (putting irrelevant, often negative anchor text linking to popular websites) Dominic tinkered with the weighting of anchor text while stringently scrutinizing back links and internal linking.[54] An exploit where webmasters would link to the same site using different anchor text (thereby allowing both links to unfairly contribute to sites PageRank) was addressed by allowing only one site (given duplicate site links with differing anchor text) link to flow to PageRank.[55]
Fritz/Supplemental Index/Florida (2004) Edit
Fritz finalized the “Everflux” implementation, meaning Google's index would receive some degree of updating every day.[52] Daily crawling added another layer of relevancy, (based upon date and time), to content ranking.[51] The creation of a supplemental index was designed to house content Google felt didn’t quite fit in its main index (due either to a low PageRank or shady linking practices).[56] Storing some sites in a separate index that was to be searched only when no good match was found in the primary index, Google hoped, would seamlessly filter out duplicate and untrustworthy content.[57] Questions about the efficacy of a multi-index system (particularly on improving recall) arose, and it remains unclear as to whether Google has retained this system.
Google's “Florida” update looked to be the proverbial “nail in the coffin” for SEO abusers. With Florida, context and relevancy were determined not just by the appearance of keywords, but of synonyms and supporting vocabulary throughout the page.[58] This sought to eliminate keyword and inbound anchor text dumping.[59] Features that had long been experimented on, like phrase proximity and keyword stemming, became official and users were now being penalized for overly using commercial/popular keywords.[58] With Florida coming down hard on SEO, many webmasters with commercial presences on the web took a drasti
