September 21, 2009

Testing EA in Virtual Reality

In enterprise architecture, we develop IT targets and plans for the organization, but these are usually not tested in any meaningful or significant way, since they are “future tense”.

Wouldn’t it be incredible to be able to actually test EA hypotheses, targets, and plans in a virtual environment before actually setting off the organization in a specific direction that can have huge implications for its ability to conduct business and achieve results?

MIT Technology Review, in an article entitled “The Fleecing of the Avatars” (Jan/Feb 2008) addresses how virtual reality is being used to a greater extent to mimic and test reality.

One example of the booming virtual world is Second Life, run by Linden Labs. It has 10,000,000 subscribers and “about 50,000 are online at any one time.” In this virtual world, subscribers playing roles as avatars “gather to role-play reenactments of obscure digital Star Trek cartoon episodes, build and buy digital homes and furniture, and hang out on digital beaches.”

However, more and more virtual worlds, like Second Life, are being used by real world mainstream businesses. For example, many companies are developing a presence in the virtual world, such as Dell with a sales office in Second Life, Reebok a store, and IBM maintains business centers in this virtual world. Further, “the World Bank presented a report in Second Life about business development.”

“But big companies like Sun, Reebok, and IBM don’t really do business in virtual worlds; they ‘tunnel’ into them. [In other words,] To close a deal, you need to step out of the ‘sim’ and into the traditional Sun or Reebok or IBM website.”

The development of company’s virtual presence online and their connection back to the real world is potentially a precursor to planning disciplines like EA testing out hypotheses of targets and plans in virtual reality and then actually implementing these back in the real organization.

Others are actually planning to use virtual worlds to test and conduct research. So there is precedent for other disciplines such as EA. For example, Cornell’s Robert Bloomfield, an experimental economist, “conducts lab research—allowing 20 students to make simulated stock trades using real money…and seeing how regulatory changes affect their behavior. He envisions a day when he can do larger studies by setting up parallel virtual worlds. ‘I could create two virtual worlds, one with legal structure, one with another, and compare them…I might lower the capital-gains tax in one and see how business responds. There are things I can’t do with 20 people in a classroom but I can do with 2,000 or 20,000 people in a virtual world.”

Could enterprise architecture do something similar in a virtual world? For example, could we test how business processes need to change when new technology is introduced or how information sharing improves with better architectures for discovering and exchanging data? How about testing people’s reactions and behavior to new systems in a broader virtual world instead of with a more limited number of customers in user acceptance testing? Another possibility is testing the effectiveness of new IT security in a virtual world of gamers and hackers?

Modeling and simulation (M&S) can improve enterprise architecture by testing plans before deploying them. We need to to hire and train people with knowledge, skills, and experience in the M&S discipline and with tools that support this. Then we can test hypothetical return on investment for new IT investments before we open our organizational wallets.



Tateru Nino said...

Actually, those numbers aren't quite right.

Users != Subscribers. The exact number of subscribers isn't known.

* Registered users topped 16.7 million.
* Users logged in within the last 60 days: Just under 1.4 million
* Users logged in at any time: 40,000-70,000

More data here:

Adrian Grigoriu said...

Hi Andy,
I believe in what you say that an Enterprise Architecture could be ultimately modeled in a Virtual environment. To great advantage.

Long ago, I've seen a tool for telecom networks (which are standardized) attempting to calculate ROI for investments in the telecom arena. I can remember that it was provided by a consultancy.

But first the EA has to be properly modelled based on a proper framework and metamodel, unlike today's, so that one can navigate from business strategy, functions and processes to the technology resources and organizational roles performing them, in order to discover the impact of change.
Then scenarios would be played for different business models for instance.
It has to start small with a few key business processes.

I don't believe that the known EA frameworks with their scattered and weakly defined artefacts would support this goal.