I guess my general point is that it is easy to "cherry pick" numbers that sound good, but for many people they will not achieve those same results because some things are sub optimal.
I have seen repeatedly that solar arrays are better suited towards areas with long days of sunlight and few clouds (like New Mexico), so the power degradation due to heat seems to be a minor factor in comparison.
A nice map showing solar approx output based on location (in the USA) can be found here:
http://upload.wikimedia.org/wikipedia/en/2/2c/Us_pv_annual_may2004.jpg
http://en.wikipedia.org/wiki/Image:Us_pv_annual_may2004.jpg
(As you can see being in the southwest is much preferable to being in the northeast. A solar system in Arizona is likely to be more effective than one in Michigan)
Here is a cool site that lets you input various factors to get an output estimate:
http://www.valentin.de/onlineberechnung/pv/pv_online.html
Panel output gives some value in kW, that you then multiply by hours of sunlight to get kWh per day.
You then have to subtract off all sorts of inefficiencies from that including:
Panels not pointed in the right direction (typically southfacing is best)
Panels not angled at an optimal angle for your location.
(Most people mount panels on their roof, and mount them flush for aesthetic reasons. Most houses don't have a roof with ideal angles and directions for PV collection)
Inverter efficiency (although 90-95% can be common)
Dirty panels.
Air quality conditions (smoke, smog, haze, clouds, etc)
Shading from trees and such.
I have solar panels on my house. I can tell you that they produce far less power on a cloudy day. Also, a tree in a neighbors yard a few houses away puts a small shadow on them late in the day which hurts output as well. (Most panels are connected in series, so a shadow on only one of them disrupts electron flow through many panels in a string).
Based on looking at output for different temperature days in summer vs winter, I can tell you that I would pick a hot summer day over a cool winter day because the hours of sunlight is the really important factor.
Also, if you have just enough panels to put power into your vehicle, you may pay more for power because you switch to a different rate schedule which can cost you more for the other power you use on your household needs.
So, one problem is you need to size a system that is "just enough" for your household needs as well as enough extra to recharge your vehicle appropriately.
In PG&E area if you make extra power they will not pay you for it. You want to target "break even". If you don't get enough panels, oriented properly, then you can get stuck with larger bills due to change to "time of use metering" as mentioned in this article:
http://www.latimes.com/business/la-fi-solar8may08,0,3494064.story
http://politics.slashdot.org/article.pl?sid=07/05/09/1241243
Don't get me wrong... I am big proponent of Solar Panels and EVs, but I think we need to be careful not to overstate the benefits then have people reject the technology thinking that there is "bait and switch" or "false advertising" going on.
The numbers from Martin & Elon do seem "ballpark close", but some people are likely to find their driving habits use more than 200wh/mile, and many people will find that their home solar systems produce less than 263 kWh/m^2/year.
I recently saw a Tesla presentation showing the reducing cost of Solar panels over time. It had some mark for 2007 with something like $2/w which I would think is some kind of raw cost to manufacturer, not actual cost to consumer (which is more like $6-$8/w installed). The skeptics balk when they see costs that don't line up with what they would actually pay, or energy generation/usage that they cannot achieve. Another case in point is the note about mono-crystaline solar panels. They cost a premium, and have limited availability so many systems installed these days use less efficient poly-crystaline.