DMOZ
| Company type | Open content web directory |
|---|---|
| Industry | Internet directory |
| Predecessor | GnuHoo, NewHoo |
| Founded | June 5, 1998 in United States |
| Founder | Rich Skrenta, Bob Truel, Chris Tolles, Bryn Dole, Jeremy Wenokur |
| Defunct | March 17, 2017 |
| Fate | Discontinued by AOL; succeeded by Curlie |
| Headquarters | United States |
Area served | Worldwide |
| Products | Web directory (Open Directory Project) |
| Owner | AOL (later part of Yahoo! Inc) |
Number of employees | Volunteer editors (97,584 as of April 2013) (2013) |
| Parent | AOL (formerly Netscape Communications Corporation) |
DMOZ—also known as DMoz and stylized as dmoz in its branding—originated from the domain directory.mozilla.org. It functioned as a multilingual open content web directory, managed under the name Open Directory Project (ODP). This directory was owned by AOL (which later became part of Yahoo! Inc) and curated by a network of volunteer editors.
The platform categorized listings through a hierarchical classification system, arranging related subjects into organized categories and subcategories.
After AOL discontinued its support, the directory was shut down on March 17, 2017. Subsequently, its homepage was converted into a static archive with links to a discussion forum dedicated to exploring possibilities for rebranding and relaunching.
By late 2017, a non-editable archived version was accessible at dmoztools.net. Meanwhile, a successor project named Curlie was launched. From 2018 onward, the terms DMOZ, ODP, and Curlie became interchangeable. Curlie preserved the original hierarchical taxonomy established by DMOZ and had become well recognized by 2022.
History
Initially created in the United States under the name GnuHoo, DMOZ was developed in 1998 by Rich Skrenta and Bob Truel during their tenure at Sun Microsystems. Chris Tolles, then marketing lead for Sun’s network security products, joined as a co-founder alongside Bryn Dole and Jeremy Wenokur. Skrenta had previously designed TASS, an early predecessor to the threaded Usenet newsreader tin. The structural design of GnuHoo’s directory loosely mirrored the organization of Usenet newsgroups.
The directory officially launched on June 5, 1998. Due to objections from Richard Stallman and the Free Software Foundation regarding the use of Gnu, the project was renamed NewHoo. Later, Yahoo! raised concerns over the term Hoo, which led to a proposed rebranding as ZURL. Prior to adopting the name ZURL, NewHoo was acquired by Netscape Communications Corporation in October 1998 and rebranded as the Open Directory Project. Netscape made the directory’s data available under the Open Directory License. Following Netscape’s acquisition by AOL, DMOZ became part of AOL’s portfolio.
Under Netscape’s management, the Open Directory Project indexed approximately 100,000 URLs with contributions from about 4,500 editors. On October 5, 1999, DMOZ achieved a milestone of one million indexed URLs. By April 2000, it reportedly surpassed the size of the Yahoo! Directory with 1.6 million URLs. Subsequent milestones included two million URLs in August 2000, three million in November 2001, and four million by December 2003. As of April 2013, DMOZ contained over 5.1 million listings spread across more than one million categories; however, by October 2015, this number had decreased to nearly four million listings within over one million categories.
Beginning in January 2006, DMOZ started publishing detailed online reports outlining project progress. The initial report covered developments from 2005 and monthly updates continued through September 2006. These reports provided more comprehensive insights compared to simplified front-page statistics that included categories such as "Test" and "Bookmarks," which were excluded from user RDF data dumps. In August 2006, approximately 7,330 editors were actively contributing; by March 31, 2007, total contributors numbered 75,151. This figure increased to 97,584 contributing editors by April 2013.
System outage and editing suspension, October to December 2006
On October 20, 2006, DMOZ experienced a critical server malfunction that suspended all editor contributions until December 18, 2006. Throughout this downtime, an outdated version of the directory remained accessible to the public. The Site Suggestion and Update Listings forms were reinstated on January 13, 2007, with the weekly RDF data dumps resuming publication on January 26. To mitigate the risk of similar disruptions in the future, the infrastructure was transitioned to a redundant configuration utilizing two Intel-based servers.
In 2016, the platform underwent an interface overhaul labeled "DMOZ 3.0," but AOL discontinued its operation the following year.
Competing directories and derivative projects
As DMOZ's influence expanded, two other web directories managed by volunteers and backed by Go.com and Zeal appeared; both have since ceased operations and did not provide licenses for open content distribution.
The collaborative editing framework of DMOZ served as a model for several open-content initiatives such as MusicMoz (a music directory), ChefMoz (focused on restaurants), and Open Site (an encyclopedia). Notably, according to Larry Sanger, DMOZ also played a role in inspiring the launch of the Nupedia project, which eventually led to the creation of Wikipedia.
Content
The original structure of Gnuhoo’s ontology was derived from Usenet, with founder Rich Skrenta manually categorizing groups into a hierarchical system. The initial top-level categories—Adult, Arts, Business, Computers, Games, Health, Home, News, Recreation, Reference, Regional, Science, Shopping, Society, Sports, and "World"—have largely remained intact, while subordinate categories developed over time through editorial consensus.
In July 1998, multilingual support was introduced via the addition of the World category; by May 2005, listings spanned seventy-five languages. Since 2002, non-English entries expanded at a faster rate than English ones, with the World category comprising roughly one-third of all listings by 2005. The structure of non-English ontologies generally follows that of English, though some adaptations exist to accommodate language-specific needs.
Certain primary categories possess unique characteristics: the Adult category is omitted from the homepage but included in DMOZ’s RDF data exports; meanwhile, the Regional category is organized geographically rather than thematically, effectively creating two distinct directory trees—one topical and one regional.
On November 14, 2000, a specialized "Kids and Teens" directory was launched for users under eighteen years old. This section enforced stricter listing criteria, employed age-appropriate language, and used age tags to differentiate content suitable for children (12 and under), teens (13–15), and mature teens (16–18). By May 2005, this segment contained more than 32,000 listings and maintained separate editorial permissions as well as dedicated RDF dumps.
Beginning in early 2004, UTF-8 encoding was adopted site-wide; prior to this change, English categories utilized ISO 8859-1 encoding while other languages employed their respective encodings. RDF data dumps have been UTF-8 encoded since early 2000.
Maintenance
Editors were responsible for maintaining directory listings by adding new sites, updating current entries, or performing both tasks. Their duties included fixing spelling and grammar errors, monitoring linked websites’ status, and eliminating spam or duplicate submissions.
Robozilla, a Web crawler, regularly scanned all sites listed in DMOZ to verify their availability. It flagged sites that appeared relocated or missing, alerting editors who then reviewed and took appropriate action. This procedure was crucial in minimizing link rot within the directory. Sites identified with errors were automatically placed into an unreviewed queue awaiting editorial evaluation.
Because DMOZ influenced search engine rankings (see PageRank), expired domains listed in its directory became targets for domain hijacking. This issue was managed by routinely purging expired domains from listings.
As corporate involvement waned, volunteers created various tools—including link checkers, category crawlers, spellcheck utilities, RDF dump search tools, bookmarklets, and Mozilla-based extensions—to assist with ongoing maintenance tasks and processing unreviewed queues.
Licensing and prerequisites
Initially, DMOZ data was distributed under the Open Directory License, which required explicit attribution on every page utilizing the data. This license also obligated users to frequently verify updates and discontinue use if changes occurred. Due to its restrictive redistribution conditions, the Free Software Foundation categorized this license as a non-free documentation license.
In 2011, DMOZ transitioned to a Creative Commons Attribution license, which is both free and compatible with the GPL.
RDF data dumps
DMOZ information was made available through weekly RDF-like data dumps hosted on a download server, with archives of previous versions maintained. Although these are referred to as RDF dumps, the files conform to valid XML standards but do not meet RDF validity, necessitating specialized tools for processing the data.
Users of content
The core directory services of prominent search engines and portals—including Netscape Search, AOL Search, and Alexa—are powered by DMOZ data. Google Directory utilized DMOZ content until its shutdown in July 2011.
In 2004, Overture introduced a search platform that integrated Yahoo! Directory results with metadata from DMOZ. The search engine Gigablast announced in 2005 that it offered a searchable version of DMOZ, allowing users to perform searches within specific categories and effectively creating more than 500,000 vertical search engines.
By 2006, DMOZ reported that 313 English-language websites and 238 sites in other languages were using its data. These numbers do not include sites employing DMOZ content without complying with its licensing terms.
Editing policies and procedures
Applicants seeking to become DMOZ editors must submit an application demonstrating their editing capabilities and disclose any potential conflicts of interest. The majority of applications are declined; however, reapplication is permitted. Initially, editing rights are granted for a limited category, with experienced editors able to request broader editing privileges. New editors receive support through mentorship programs and internal discussion forums.
Communication among editors occurs via private internal forums dedicated to editing topics. Senior editors may be granted extended privileges: edit all allows editing across all categories; meta includes responsibilities such as reviewing new editor applications and handling abuse reports; cateditall and catmod offer similar powers confined to single categories; catmv authorizes category restructuring. These privileges are assigned by administrators following consultation with meta editors.
The admin role was introduced in August 2004, enabling holders to grant editall+ rights and approve policies affecting the entire directory—functions previously exclusive to staff editors.
All editors are required to adhere to DMOZ's Editing Guidelines, which cover criteria for site inclusion, listing standards, category naming conventions, conflict of interest policies, and community behavior standards. Breaches of these guidelines can result in suspension or revocation of editing rights. The guidelines are periodically revised based on discussions within editor forums.
Criticism and disputes
There have been ongoing accusations that volunteer editors at DMOZ prioritize their own websites while obstructing competitors. To address these issues, ODP staff and meta editors have enforcement powers to discipline editors who engage in abusive behavior. In 2003, DMOZ implemented a Public Abuse Report System allowing users to report misconduct by editors online. Discussions about challenges within DMOZ have become more frequent on webmaster forums. Although policy advises submitting a website to only one category, as of October 2007 Topix.com—a news aggregation site owned by DMOZ founder Rich Skrenta—was listed in over 17,000 categories.
In its early stages, DMOZ granted editing permissions to representatives from certain companies such as Rolling Stone and CNN for listing individual pages. Links to CNN articles were added until 2004 but were completely removed in January 2008 due to outdated content and difficulties maintaining them. Since then, no comparable editorial policy experiments have been conducted.
Ownership and Management
The ownership and leadership of DMOZ have been subjects of debate, with some of the initial GnuHoo volunteers feeling deceived into participating in a commercial venture. In its early stages, DMOZ lacked formal management structures—there were no official forums, guidelines, or FAQs in place. Eventually, the ODP editor forums emerged as the informal governing authority, where staff viewpoints were regarded as definitive decisions. Senior editors who had earned trust were granted additional responsibilities, such as approving applications from new editors, establishing a tiered system where paid personnel held ultimate control over policies and operational procedures.
Robert Keating, a principal at Touchstone Consulting Group, acted as AOL’s program manager for DMOZ beginning in the early 2000s. His roles evolved to include senior editor for AOL Search and managing editor for AOL’s DMOZ initiative before he transitioned to overseeing media ecosystem management within AOL Product Marketing.
Editor Removal Procedures
The removal of editors is managed by DMOZ staff and meta-editors, primarily addressing issues related to abusive editing or inappropriate conduct. These disciplinary matters are discussed within private forums inaccessible to the editors involved, a method that has been criticized for its lack of transparency compared to more judicial approaches.
Meta editor Arlarson pointed out that misunderstandings surrounding removals often arise due to misinformation spread by former editors. Current editors are bound by confidentiality rules that prevent them from publicly discussing specific removal cases; nonetheless, general reasons for removal are outlined in official documentation. Consequently, some editors only become aware they have been removed when they lose access to their accounts.
Former contributors such as David F. Prenatt Jr. (netesq) and The Cunctator have claimed that removals were linked to disputes with staff over policy amendments, especially concerning copyright regulations. They argue that allegations of incivility were sometimes employed as a pretext to dismiss editors who opposed certain changes.
Blacklisting Allegations
Senior editors at DMOZ possess the ability to add "warning" or "do not list" annotations to specific domains; however, no single editor has the authority to entirely block a site from being listed. Domains marked with such notes may still appear in listings, and these annotations can be removed following community discussions.
Hierarchical Structure
By 2005, there was growing criticism regarding DMOZ’s hierarchical organization, which many considered overly complicated. With the advent of Web 2.0, alternative classification methods such as folksonomies gained popularity. Some editors advocated that folksonomies, network structures, and directed graphs provide more intuitive and manageable solutions compared to rigid hierarchical models.
Search
ODPSearch, the search engine utilized by DMOZ, is based on the open-source Isearch software and is distributed under the Mozilla Public License.
Editor Forums
Initially hosted on proprietary Ultimate Bulletin Board software, the ODP Editor Forums transitioned to the open-source phpBB platform in June 2003. By 2007, these forums operated on a customized version of phpBB.
Bug Tracking
For tracking bugs, ODP employed the open-source Bugzilla system alongside Apache as its web server software. Squid proxy was also used until August 2007 when server storage was restructured and Squid was discontinued.
Interface
The software and database behind DMOZ were not open source, although in 1998 Richard Skrenta contemplated releasing it under the GNU General Public License. This consideration sparked objections from the GNU project, mainly concerning the licensing of DMOZ's content. Eventually, the directory's content was made available under a Creative Commons license that aligned with the GNU license.
As a result, various alternatives to DMOZ appeared, allowing editor communities to develop and manage their own Web directories featuring open source software and open content.
ChefMoz
Chef Moz, a derivative of DMOZ, served as an open content directory focused on restaurant listings, curated by a volunteer group of editors and owned by Netscape. Similar to DMOZ, it employed a hierarchical structure to categorize listings into main categories and subcategories.
On February 17, 2011, the DMOZ administrator known as "lisagirl" announced that Chef Moz had ceased operations.
Growth
Between its inception in 2000 and the cessation of editor contributions in November 2009, ChefMoz expanded to become the largest online restaurant directory worldwide. The count of restaurants indexed over the years was as follows:
| Year | Number of restaurants |
|---|---|
| 2000 | 48,000 |
| 2001 | 75,000 |
| 2002 | 179,000 |
| 2003 | 208,000 |
| 2004 | 258,000 |
| 2005 | 266,000 |
| 2006 | 275,000 |
| 2007 | 315,000 |
| 2008 | 325,000 |