Original URL: https://www.theregister.com/2008/11/19/decho_analysis_and_commentary/

EMC remixes Decho and the Funnymen

Connecting a Hoover to a personal information bank

By Chris Mellor

Posted in On-Prem, 19th November 2008 08:02 GMT

What is Decho about?

Decho is Pi and Mozy combined, headquartered in Pi Corporation's offices in Seattle, and with a Pi research centre in Montreal, the Mozy base in Pleasant Grove, Utah and coding development in Bangalore. There are twenty people in Seattle and eighty or so in the other three centres.

The few staff we know of include Arathi Kundapur, the director of engineering in Bangalore, and Charles Fitzgerald, a former Microsoft platform strategist who is Decho's VP product manager. A CEO is said to be close to appointment.

Mozy is the shipping product and is just a hoover. A very nice hoover to be sure, and one with a variety of front-end nozzles, but they all do the same thing: suck up data and drop it off in EMC's Fortress data centres around the globe. Pi does something with the data. Let's see if we can work out what that something might be.

Pi Corp

Pi (Personal Information) Corporation was started up in 2004 by Paul Maritz, using funding from private equity Warburg Pincus of New York. At the time Maritz said Pi desktop software would be open source and Linux-based.

Maritz set Pi up to "create, repurpose, store, share and access personal information in novel ways" and assumed a world where broadband networked consumers and knowledge workers have intelligent devices to create, store and access information. As far as the PC is concerned, "information that is stored is hard to find, share with others and keep up-to-date. New metaphors based on search and selective views that are automatically generated and updated are needed."

Okay, so Mozy sucks the stuff up to the cloud, and Pi searches it and creates selective views helping people find information and share it more easily and more securely.

The information that Pi creates won't be Word documents or Quark projects or digital photos or whatever. It seems that PI will create information from original content, Pi being a metadata play and the metadata helping us to manage and control our digital lives, the digital echoes of our real lives.

Smart Desktop

In July 2006 Pi Corp, then focused on consumers, bought Smart Desktop Inc, another startup aimed at improving knowledge worker productivity by commercialising a TaskTracer project from the Intelligent Information Systems Laboratory at Oregon State University (OSU). The research group there, directed by Professors Tom Dietterich and Jon Herlocker, developed technology that uses machine intelligence (AI or Artificial Intelligence) to automatically classify, sort and organize information for people by observing and learning from their interactions with their personal computers. Smart Desktop has an exclusive worldwide license to the TaskTracer technology, patents, and intellectual property from the University.

Its CEO was John Forbes, a 20-year vet of the software industry and the two professors above. These were VP of Engineering Dr Jon Herlocker, co-inventor of technology commercialized by Net Perceptions, a pioneering company in collaborative filtering technology and Chief Scientist Dr Tom Dietterich, then the president of the International Machine Learning Association. The Smart Desktop team is based in Seattle too, and is run as a separate operation within Pi, now Decho, with Forbes as its president.

At the time Herlocker said: "Everyone knows how frustrating it can be when you’re trying to locate files on your computer or network. Our software automatically organizes information in a way that is more natural for a person rather than a computer – learning what files, email messages, contacts and other information go with which projects.

“So when you need to switch to a different project, you can instantly bring up all that information associated with the project. This is going to revolutionize the way people and computers interact. Pi Corporation felt that this technology was so valuable they wanted not only to invest in it, but to own it."

Pi has said that it aims to create "innovative online services that make it easy for individuals to gather information, organize it, share it instantly and securely, and access it from any web-based device. Pi users are not required to set up servers, explicitly copy or upload information, or surrender control of their personal information to an online hosting company. Pi is currently (2006) in limited beta test with its first product and will expand its offerings over the coming months."

John Forbes said at the time: "Combining forces with Pi Corporation will accelerate the development of our products for information workers, and we can take advantage of Pi's innovative technology for securely sharing information across networks. With Pi's offerings for consumers and Smart Desktop's solutions for information workers, we plan to develop a consistent information management platform for computer users worldwide."

Maritz offered this thought: "With Smart Desktop's innovative technology, the computer can start to think like a person and organize things in a way that makes sense to human beings. This capability will help Pi fulfil its mission of helping people organize their information in better ways, and make their information available, any time, anywhere without compromising security or privacy, and without being bound forever to a single mega-online provider."

From this report we see that Smart Desktop runs on a PC and automatically groups content - either local or accessed web pages - on it into projects, based on existing organisational structures set up by the user in Outlook or a browser (bookmarks and bookmark folders maybe). Content is then assigned to projects using the topic of the content, metadata about email or an activity's context. One aim of this is to have a single view of all the content pertaining to a project wherever that content came from.

This seems to be a kind of folder analogous to but distinct from a file system folder. Users could create new projects in Outlook and Smart Desktop will add (tag) all relevant emails and documents and web pages etc to that project as they come in or are created. Project content will be capable of being viewed with a historical timeline.

Personal information bank

Okay, so we have a picture of Pi software running in the front-end consumer/knowledge worker device, mining content for metadata and using the Mozy Hoover to backup information and metadata to the cloud for safe-keeping and sharing.

EMC chief technology officer (CTO) Jeffery Nick, said more in this video when he talked of a next-generation information infrastructure platform for cloud computing. It needs (SmartDesktop) technology to index and classify information in our personal workspaces and derive metadata associated with the tasks we are performing across disparate applications. It federates the information into views that are relevant to the projects we are working on.

There has to be the ability to ingest this into the cloud (Mozy) for personal data storage, protection, isolation, privacy and secure sharing. You will be able to access it on your PDA when you are away from home or work. The vision is of a personal information bank which is not locked or tethered to a home PC or office desktop or server and which is based on a multi-petabyte, enterprise-class data centre with enterprise quality of service, protection, security, disaster recovery, indexing, classification and metadata capture, all made accessible to millions of people through the cloud.

Pi, then, will combine consumer and knowlege worker (Smart Desktop) original multi-sourced content - photos, documents, financial records, contacts, correspondence, professional portfolios, mail, music etc - indexed and classified using AI techniques to provide metadata to deliver timelined task or project-oriented views of the content from the cloud with that content and metadata stored (Mozy) in the cloud as our own individual personal information bank.

Simple really. The Mozy hoover will become a Pi Dyson with a transparent dust (data) container into which you can look through a variety of different views for data automatically grouped into different projects. This software may appear next year. ®