Turorials: History of Hubzilla
The history of Hubzilla
Hubzilla is a community developed open source project based on work introduced in Friendica by the Friendica community and which previously was named Redmatrix. The core design, the project mission, and software base itself were created/written primarily by Mike Macgirvin and represent the culmination of over a decade of software design using variations of this platform and an evolving vision of the role of communication software in our lives. Many others have contributed to this work, both conceptually and in terms of actual code (far too many to list individually). Mike Macgirvin — Biography
Mike Macgirvin is an American software engineer now living in Australia. He spent his early adult years designing and repairing semiconductor fabrication equipment for a number of companies as a self-described « machine wizard ». In 1985 he became a research engineer at Stanford University for the Gravity Probe-B space mission and soon became a Unix systems administrator writing communication software and utilities; and becoming an expert in emerging internet technologies such as the now ubiquitous « World Wide Web ». He authored an email « client » called « ML » which pioneered some advanced concepts in encryption, the ability to filter message streams into different « views », and multi-protocol support; and was an active proponent of and participant in the open source software movement. In 1996 he went to Netscape Communications to become tech lead on their Messaging Server and integrate this with Collabra (groupware) into a comprehensive communications server package. He stayed on after Netscape was acquired by America Online and was tech manager of the Groups@AOL project until 2001.
During a layoff round, Mike was let go from America Online in August 2001 and purchased a music store in Mountain View, California later to be known as « Sonica Music Company ». Opening a retail store for non-essential goods at the beginning of a prolonged economic downturn was in retrospect probably not the wisest career move. Sonica eventually folded; in late 2006. Mike returned to working on software and systems support full-time and was employed briefly at Symantec before moving to Australia in early 2007. He currently lives on a farm « out in the middle of nowhere » and is employed as a Computer Systems Officer at the University of Wollongong.
Hubzilla – The Early Years
The software which went into creating Hubzilla has been through several distinct historical phases. It began in 2003 when Mike Macgirvin was looking for a content management system to power the website for his music store and found the available solutions to be lacking in various respects. The project was born as the « PurpleHaze weblog » under the nom de plume « Nerdware Communications ». It was a multi-user PHP/MySQL CMS which provided blogs, forums, photo albums, events and more. Initially it provided the basis for a social community and shopping for customers of the store, but was also linked to Mike’s personal weblog running on another domain. The distinguishing characteristic of this software was the ability for so-called « normal users » to re-assemble the components and choose different content feeds – and in essence create their own personal « multi-user CMS » as a view. Their custom view was able to communicate with anybody else that used the system, but could be partitioned so that adult sites and motorcycle enthusiast sites would not be visible to each other and not clash (or in this case Mike’s personal website and the music store website). This software was developed primarily from 2003 until 2008.
In 2006 this software was used as the prototype for Symantec’s « safeweb » reputation and community site. It was developed and enhanced until about 2008. A rewrite took place in 2008 named « Reflection » but work stagnated as the community dwindled. The need for content management systems and communications software dropped dramatically during this time as humans flocked to the new social aggregrators – Facebook and Twitter.
In early 2010, Mike left Facebook, concerned at the company’s increasing hold and control of personal information. In his words « Companies die. We watched it happen in the dot-com years. When they do, their databases are sold to the highest bidder. ». Mike used some remnants of the old CMS project to create a decentralised social communications platform. This was launched in July 2010 as « Mistpark ». The name was chosen as a tribute to his new home in the Southern Highlands of Australia. The key innovation in this project was the ability to authenticate remotely and invisibly to other decentralised instances of the software so to allow remote viewing of private photos and provide « wall-to-wall » posting across website instances. The lack of simple remote identity provenance was a serious limitation of other decentralised communication protocols.
In late 2010, the name was changed to « Friendika ». The name Friendika had some symbolic issues, since the suffix was common with « swastika » and « Amerika », both having negative connotations, however the dot-com domain was available. Friendica was in fact the first choice but the ‘friendica.com’ domain name was already registered. It became available a year later and the project was renamed to Friendica in late 2011.
Soon after version 1 was released in July 2010 – providing basic social communications, the software also took on a new role – cross-service federation; which was first introduced in August and September 2010. Federation allowed the software to « behave as » a StatusNet site and friends and messages could communicate to the other service from their own platforms. It was also hoped to provide federation with Diaspora – a project with similar scope being developed in secret in New York and first released in November of that year. Over the course of the next year, the federation ability was extended to provide integrated communications from RSS feeds, to and from email, StatusNet, Facebook, Twitter, and the emerging Diaspora project. The software provided a single « view » of your entire social space no matter what provider you or your friends used. StatusNet and Diaspora were supported natively so that one account could access any of these services. Facebook and Twitter used « API federation » which required the person to have an account on those services with which to link.
By July 2012, Twitter and Facebook had both changed their terms of service and essentially outlawed « API federation » in the way Friendica was using it. Diaspora announced they were changing their protocol and would not maintain compatibility nor provide any warning when compatibility would break (or documentation on the proposed changes). The creator of StatusNet was also leaving his project to create something new (pump.io). As the software’s primary purpose by this time was « federation of different social services into one interface », this created a bit of a crisis. The federated social web was crumbling. Also of concern was that independent and decentralised social websites shut down frequently, requiring all their members to start over again on another site. Often the effort involved to do this seemed daunting – and many people ran back to the relative safety of the large corporate providers – Facebook, Twitter, and now Google+.
Mike realised he did not want to be held hostage to the decisions that other projects and companies and independent websites make. Friendica could operate on its own without attaching to these other networks, but its vision and implementation of a federated social world depended on federation with others for its project identity – so this created an identity crisis.
Mike had been working on this project for some time and there were a number of things which needed re-writing, including the base communication protocol which Friendica used (DFRN or the « Distributed Friends and Relations Network » protocol). These ideas were starting to emerge as a different method of communication he called « zot ». Zot began as a way to create a common language for federated websites, but there was no interest in this ability and as mentioned, the federated web was crumbling. The first version was soon scrapped and zot was re-designed and re-ignited as a streamlined communication protocol which was location-independent; e.g. not tied to any website. This would allow people to carry on unaffected if their website operator shut down temporarily or permanently. They wouldn’t have to make friends all over again, and permissions of everything on the system wouldn’t have to be changed to allow bob@site1 to see something that was private to him, even though he was now bob@site2. This was a serious problem with decentralisation. People moved and their online identities were lost and had to be re-created from scratch and existing relationships destroyed and had to be created all over again.
In July 2012, Mike left the Friendica project and began development of « zot » and a new base project called « red » in his somewhat elusivespare time. Red is Spanish for « network ». It wasn’t really a « social network » and especially not a « federated social network ». It was just Red (technically « la red »), or « the network ». Work began by removing all the « federation » components and going back to basics – communication and remote authentication. It was a major re-write and took roughly six months before even basic communication was re-established. It was also no longer compatible with Friendica – which had been given to the « Friendica community » and by this time (December 2012) was developing separately on its own track.
It became clear during this time that the single most compelling feature of the project wasn’t the social network at all, but the authentication layer and decentralised access control mechanisms. Combined with zot’s location independence it created a new model for software which had never existed previously – decentralised identity-aware web publishing and single sign-on to any compatible provider across the web. These weren’t evolutionary, they were revolutionary. One of the biggest flaws of the modern web is the reliance on different passwords for every service you use, or reliance on a single provider if you were to tie them to – say your Facebook login. Facebook can remove your account at any time. Gone. If you rely on their authentication for all your websites, your entire online identity – now gone. This is also what was missing from Friendica – a compelling software feature which could stand on its own, without requiring a social network and especially without requiring a federated social network with all the mentioned external dependencies.
An early visitor to the project noted that he had some difficulty finding the project on Google because of the choice of name – « red ». Yes, this was a poor decision in retrospect. We were buried on page 23,712 of the search results. The concept that was emerging around this identity-aware publishing was that of « a matrix of inter-connected thought streams », since we didn’t have a concept of « people » and « friends ». All were just connected « channels » with different ways to connect. So « Redmatrix » was chosen to give it a searchable name. It had nothing to do with the Matrix film and red and blue pills, though that is frequently cited (erronously); and in fact isn’t a bad analogy.
The concept of identity-aware content was alien to anything that existed previously on the web, so to make it useful we had to provide the ability to use it for content. It needed content publishing tools. This brought back concepts from the old « Content Management System » on which the software was originally based. To get it up and running quickly we created a markup language for webpages called « Comanche » which let you describe a page in high-level terms based on bbcode tags. We also added WebDAV so you could put decentralised access control on files and drag/drop from your operating system. So now you could have private photos, webpages, files, events, conversations, chatrooms – and they are visible to those you choose – no matter what site they use. All they need is zot. And your viewers could move to another site or just pop up at a different site any time they want and we don’t care. And it also had a built-in social network; with lots of additional privacy and encryption features which were added even before the Snowden revelations gave them added urgency.
Over time a few federation components re-emerged. The ability to view RSS feeds was important to many people. Diaspora never really managed to re-write their protocol, so that was re-implemented and allowed Redmatrix to connect with Diaspora and Friendica again (Friendica still had their Diaspora protocol intact, so this was the most common language now remaining on the free web – despite its faults). Diaspora communications aren’t able to make use of the advanced identity features, but they work for basic communications.
The Redmatrix project reached a point of stagnation in early 2015 as network growth leveled and active interest in the project declined. Mike met with several external high tech developers and innovators in a round of discussions that were called « Zotopia » in early 2015 to perform an independent review of the project and try to identify what had gone wrong and plan a route forward. The basic consensus is that the project suffered from bad marketing decisions which were compounded by mixed messages about the project goals and target audience. A « rival » project (Diaspora) was marketing itself as a Facebook competitor, but after some long discussions it was determined that Redmatrix wasn’t a Facebook competitor at all, and too much emphasis was being placed on the « social network » and « anti-Facebook » features. It was a novel decentralisation platform with distributed identity and permissions, and as was pointed out, the « end user » was the wrong target market. These marketing mistakes were now identified with the project name and random sampling of various « customers » showed that none of them really had a clue about the software goals or target market segment. The mixed messages were associated with the brand identity and this was a problem.
The Redmatrix community held a vote and the project was renamed « Hubzilla », with a renewed identity and focus – to provide software for creating and ultimately linking together unrelated community websites or « hubs » into a global community. This is in fact what we were building all along, but didn’t fully recognise it. The target audience for this software as it turns out is not the members or end users, but software integrators and digital community architects and builders. These in turn will be responsible for marketing their own product (their respective online communities) to end-users or members. The software solves a real world need of linking isolated and « walled garden » community sites together into a larger cooperative. The transition from Redmatrix to Hubzilla was complex and has taken several months as we consolidated the marketing and media assets to deliver a consistent message. It is still ongoing at this time, and should be completed in Q4 2015.
Mike stepped down as active coordinator for the project in early 2015 and turned management over to the community. He remains active as a Hubzilla developer. And Then…
In 2016, the project was re-architected to support multiple server « roles ». These correspond to sub-projects which can be isolated from each other in terms of supported feature sets, but all use and support the same code-base and developers are able to work together on common features and goals. The roles primarily differ in target audience, project governance and decision making structures, and this results in slightly different features and idealogy. They all share a common code repository.