[cross-posted to Urban Sprawlette]

We’ve made a lot of progress on the Nudge project recently, and it’s time to think seriously about potential applications. Infrastructure is in place for automatic discovery of structures, algorithms, patterns, models, equations… but when you’re building a tool it’s all just play-talk until you consider some interesting challenges and see if you can frame them readily as something your tool seems useful for.

At breakfast yesterday Barbara and I spent some time riffing on various “hard” design and optimization problems that it would be nice to solve. And since we’re thinking of our ubiquitous and time-stealing house, her thoughts went towards HVAC.

In particular, the esoteric (but economically important) calculation of how to size air cooling equipment for a building. The standard — that is, the Standard — is spelled out in Manual J, now in its eighth edition from the ACCA.

Here’s how it works, in principle: You want to build a house? You’ve got it designed, with floorplans and siting and what most folks think of as “design” done? Well, how large does your A/C system need to be?

A good salesman or contractor, especially one who doesn’t really care how much money you spend after you leave his care, he can pick something he’s familiar with that’s “big enough” to manage temperature control and ventilation and such, regardless of whether your house is super-insulated or what fancy-dancy windows you’ve got. Heck, that’s easy: the biggest you can afford.

If you press a professional contractor that this kind of approximation isn’t exactly what you had in mind when you set out to save energy costs and create a “smartish” house and save in both short and long-term, he can turn to Manual J to do a “proper” calculation.

Manual J is big. I haven’t seen the copy I’ve ordered from the library yet—and even that’s the abridged version—but I know that there’s a complex algorithmic calculation. The required cooling load calculation of a house depends on the size and position and material of windows, the overall envelope, the insulation, position, geographical location, foliage cover, exposed foundation, ceiling heights… loads of stuff you might consider “design variables” if you weren’t already holding a finished house plan in your hands. As far as I understand it, the contractor enters this information into an ACCA-designed spreadsheet, Excel stuff happens, and out pops a slightly less salesmanlike estimate of the HVAC needs of your house. And then you can refer to Manual S to pick out equipment.

Now looking at the ACCA description of the work, I’m seeing things like this:

MJ8 also accommodates homes that have exceptional architectural features and life style accessories such as:

• Dwellings that have limited exposure or no exposure diversity
• Homes with large south-facing glass area or rooms with unusually large glass area
• A thermally isolated solarium
• Customized internal load estimates
• …

And so on, for 561 pages, nominally.

Now you might be able to see where I’m heading by now, and you’d probably be right: That’s sounds like a nice place to slap a pattern discovery system.

And so I think we will.

But what I’m sitting here thinking about is the ACCA itself, and the social process that goes into eight consecutive editions of this sprawling empirical model. There must be reams of data… somewhere, and there must be reports and whitepapers and supporting evidence that makes clear the design process underlying the Manual J model (let alone everything up the Manual S (which is the highest-lettered I’ve seen so far).

As an indirect customer of the ACCA, I have to say it would be nice to have access to that data. To try to determine whether a simpler, clearer model might be more accurate and robust than this spreadsheet. A history of the models, a public record of how things are done. Oh, hell, maybe a conversation about what might actually be going on.

But I’m a dreamer, surely. Somebody has to pay for all that data collection. Not everybody is trained well enough to manage the complex calculations underlying the first-principles models or the empirical analyses. What would happen to standards of quality if anybody could chime in and criticize or amend something as important as these calculations?

After all, the goal of the ACCA is to make “the industry more successful.”

Nonetheless, I’d like to be considered a part of that industry, speaking as a technically astute consumer who pays their bills. I’m more successful whenever expertise is not masked by obfuscatory gravitas, when decisions can be clearly justified, when data can be re-used and expanded at will. When people can see what’s going on inside, and participate.

So I’m increasingly tempted to reach into the building trades, specifically through their multitudinous standards organizations, and start chipping away at some silo walls.