next*

Recently, we've been hosting the NASA CoLab Luna Philosophie talk series at Brickhouse in San Francisco. It's a public conversation between NASA thought leaders and top scientists and the community: "Anyone who is interested in an open, creative dialogue on human, space related topics is encouraged to attend!"

The second NASA CoLab@Brickhouse session took place earlier this month. Tom Cochrane of NASA Ames discussed his most recent project, Virtual Reality System Engineering Environments for the Space Program. He reflected on his lengthy career -- 25 years of using ever more powerful technology to engineer complex simulation models ranging from ground water movement to space flight.

Building on the tradition of SimCity and SimEarth, Tom and his team at NASA Ames have built SimStation, SpaceStation:SIM, and SimCEV. These are virtual environments for browsing and understanding real human space flight systems environments. The team is currently developing SimConstellation, a viewer that will support NASA's plans to return to the moon.

Unlike Sim games from Maxis, NASA simulations are mission critical: if the engineers get it wrong, the astronauts don’t come back. The simulations monitor oxygen levels, rocket fuel, and a range of other factors to accurately replicate what’s supposed to happen. You can drill into each phase of the mission and rotate the views to see exactly what’s going on inside the craft.

This degree of complexity is extraordinary. For instance, in order to land on a particular spot on the moon, the spacecraft has to orbit the moon in a particular orbit. To achieve that orbit, it must get onto a precise space trajectory out of earth’s atmosphere, and for that the craft must launch from a particular spot, at a particular speed. And that’s just for modeling orientation...

One memorable quote from Tom: “I remember being a brat on the Apollo program – they did all their documentation using typewriters and carbon paper.”

Decades later, Tom and his team have built an entire environment that generates simulations. An environment that's light-years beyond what you could visualize with a typewriter.

Thanks to Ricky Montalvo for his high-quality video of the talk. Watch it here:

Video by Ricky Montalvo, Yahoo! Developer Network, Total runtime 59min:33sec

download (m4v)

Popularity: 8% [?]

We’ve had some great speakers at Brickhouse and last week’s talk by Jeff Hammerbacher of Facebook, part of the 2008 HackHouse talk series, was no exception. Jeff is head of data and analytics at Facebook. He spoke about his role, what his team has accomplished, and some of the challenges they face.

It was standing room only, highly technical, and totally engaging. Jeff described what it’s like to manage unprecedented growth of highly sensitive user data in a challenging business environment ("we’re collecting tens of terabytes of data every day”). The team first tried a standard MySQL database, but quickly outgrew it, and had to devise unique ways of doing weblog analysis.

I was personally interested in how Facebook deals with constantly changing data elements – it’s very hard to track historical trends when the data is always changing. At Facebook, they’ve built custom analytics packages to handle this problem and intend to release some of it open-source. Go Jeff!

Special thanks to Nikhil Bobb, from the BravoNation team, for inviting Jeff to come in and present; and to Ricky Montalvo, from the Yahoo! Developer Network, for the video.

Total runtime: 1hr:11min
download (m4v)

Editor's note: Jeff Hammerbacher's presentation, The Facebook Data team: Infrastructure and insight is also available on Slideshare. Thanks Jeff!

Popularity: 11% [?]

Think of Brickhouse as a distribution channel for nimble ideas. As a large operating company (and Internet giant), Yahoo! is geared to manage monster products like Mail, Messenger, and the yahoo.com front page. Managing hundreds of millions of registered users poses a unique set of operating criteria, it's a very different challenge to focus on smaller ideas.

And yet, there is an extraordinary amount of innovation being generated across Yahoo!. In less than two years, Yahoo!'s Hack program alone has generated over 1,400 ideas in various stages of completion. In addition, Yahoo has an internal suggestion board with over 1,600 concepts, and 250 or so ideas have been submitted directly to Brickhouse over the last six months. This doesn’t take into account new technologies and products constantly being churned out in the business units. So, there's no shortage of good ideas.

To manage this abundance, Brickhouse operates like a venture fund. There’s an investment thesis, deal flow to manage, and investment criteria to apply -- all within a portfolio framework. People pitch us ideas; we pick the exciting ones and flesh them out with the "founder." We brainstorm to overcome issues and mitigate risks, and our output is a summary startup business plan. Then, on a quarterly basis, we gather the best ideas and present them to our Board of Advisors, which includes leaders from Yahoo!'s executive management team.

We convene the Board (getting their calendars lined up is a whole other challenge), present the ideas, and get guidance on which are most interesting. Then we go execute (the easy part for me).

In a traditional venture model, a newly backed venture goes and hires a great team. At Brickhouse, we have something better: a resident team of fabulous, experienced developers. Team members sign on for the ideas they find most exciting: once an idea gets the green light, we can step on the gas almost instantly. Today we handle two or three projects in parallel.

The Brickhouse venture model evolved as a way to tackle Yahoo's idea backlog. We wanted to find the shiniest needles in that haystack. Our team divvied up the ideas into lots and graded each against the following filtering criteria:

• how big is it
• how close is it to Yahoo!'s strategy
• how easy is it to build
• how easy would it be to gain market traction

This quick and dirty approach yielded the top 50-60 ideas fairly rapidly, and we then went into analysis mode, to craft and define the hybrid venture model outlined above. Simultaneously, we've spent a torrid half-year assembling the team, launching the facility, drawing on whiteboards, playing with nerf guns, and getting started on our first products. Stay tuned, because our first shiny new things will launch in the coming weeks, as 2007 winds down.

(Photo from Scott Beale / Laughing Squid)

Popularity: 9% [?]