And I would argue the problem is not one of complexity at all, but one of competence. At the risk of sounding condescending one only need to look at Tesla’s description of complexity:
“Roof complexity is determined by the number and severity of roof obstructions such as the pitch, number of joints, chimneys and other features on your roof that add to the difficulty of the installation.”
If we take a look at the individual criteria for Simple, Intermediate, and Complex we find:
- Number of levels (roof sections built on multiple stories of your house)
- Mounting plane (Uncrowded, More, and Heavily crowded)
- Obstructions (pipes, chimneys, and skylights
- Pitch (low, higher, and steep)
Note: Notice the adjectives. The lack of empirical measure is interesting. Looks more like alchemy than science.
To be truthful, I’m not knocking Tesla for this, estimation is as much art as it is science. However, what is required is experience (and they should prove more empirical factors, like the numbers for pitch?). So I have to think what methods did Tesla use to arrive at their original estimates, and how did they determine each factors cost? By the way, what makes each of the above factors complex? For example, why is pitch complex? Working on an 8 pitch is in some ways easier than a 5. Going back to the video links I posted notice the installer is sitting down above the install line on the lower pitch roof—this takes more time. In the second video, except for the starting course they are working off the ropes below the install line (Why a pick is not used is beyond me, especially on a long sections; (BTW-this is one reason why my rain gutters are damaged)). But I digress.
According to Musk complex roofs take more time and material to install, so let’s break a hypothetical estimation down. Aside from pitch, there are specify materials associated with each of Tesla’s reaming 3 criteria for complexity (Pitch is possibly more time than material). So the material list for a complex job will change. Associate with each piece of material is an installation cost—so as the amount of material increases the cost increases. The Art is when taken together how much extra time is needed. When McMansion started popping up, and similar roof lines became more prevalent, estimators needed to sharpen their pencils to be more competitive. Before that we would just double the cost, and in all seriousness if we had enough work, hoped we didn’t get the job. But then these crazy multiple level roof lines (even on single-level houses) became ubiquitous.
There was really not much magic here. Initially I took each hip and valley and flatten the roof out like a globe onto a 2d map. Then I and add the missing squares back in. The art of being competitive was based on past performance to fine-tune an estimate up or down. Eventually we figured out how to better associate material estimate to time. This is not to say there were not outliers, but there were fewer, but they became easy to spot. Buy the way for developers who built the same or similar house over and over this was eventually just "phoned it". For homeowners we still visited every residence and sometimes factored the owner into the equation.
So what’s my point?
First, I only provided my experience to highlight what I believe is the essential problem, lack of roofing experience (this is correctable). I would have to assume Tesla ‘practiced’ their processes for installing their solar roof, and/or used their V2 installs, to develop their estimation methods. I don’t care if Tesla practiced on a dog house, a doll house, or a yurt—this should have been ironed out sufficiently. I don’t know this for sure but knowing Tesla they might have attempted to reinvent the ‘estimation’ wheel and not used the expertise of existing roofers and estimators. If so that’s on them. This is NOT an "unforeseen circumstances at the installation location"—it’s incompetence on their part.
On the other hand, Tesla might have gotten too far out in front of themselves and is taking a mulligan again NOT an "unforeseen circumstances at the installation location". This smells of SolarCity who was so stovepipe one hand did not know what the other was doing. In my diatribe (
The Domino Effect) I tried to describe what I saw was the effect of an organization and process that is compartmentalized to the point of exclusion. If this is the case there are issues regarding organizational culture that may require Tesla to get out of the solar installation business.
Edit: The relationship between workmanship and installation cost is not a trivial one. Tesla suffers from the inability to deliver adequate local after-sales service—SolarCity had similar problems. This is a marriage made in hell.