Actually we are not. You have altered my original statements as badly as hansen et al have altered the surface temperature database. If you were going to perform a linear regression study of your pool, which was your original statement, then as I pointed out, you would need to know how much water you began with in order to have a data set that was worth the time it took you to do it.
Then you asked me how I would do it and as I said, it is just as bad to overcomplicate a simple system as it is to oversimplify a complicated system. Your method was overcomplicated and wouldn't produce anything like actual useable information. My method was simple and would tell you whether or not you had a leak in 24 hours and
Again, you are mischaracterizing my statements and taking them out of context. I said that if you were going to use your system, and get anything like useful data, you would need to know how much water you had in your pool. As in the house example, a data set that wasn't accurate or injected a bias would be useless in determining whether or not the house was warming or cooling, a data set that doesn't involve the absolute amount of water in the pool at any given time would produce results that would be of no value in determining whether or not you had a leak. A linear regression study on useless data is going to produce nothing but useless output. Of course, you could certainly act on that data and maybe be right or maybe be wrong, but your actions would not be accurately directed by the output of your study.
You acknowledge that you couldn't rely on temperature data in the home if accurate temperatures were not taken. By the same token, you couldn't rely on the data gathered from your swimming pool unless you could be reasonably sure that you know how much water is in the pool. You chose a complicated method and as such, gathering meaningful data is going to be complicated.