The good news is that the sink seems to not be leaking. I was not expecting that.
Otherwise... Not much new. The cold and rain have come back in. House bound is more literal than it has been. We got a walk in and then took a trip to the post office. Pretty exciting stuff...
The stock market was pretty healthy. I expected it when Trump suggested the lock down can't last forever. Now, is that really news?
How about the reporter who asked if grocery stores and fast food restaurants ought to be closed? How much food does everyone have? A month's worth? And closing food stores will make stuff better for anyone?
Not sure what all of you think about the main street media these days, but they have not made much of a favorable impression on me. Always looking for the worst case scenario. People have enough fear without piling on with no rational reason.
I'm finishing this on Tuesday morning...
The kitchen sink is pronounced to be complete. No drips over night and so we are putting the under sink stuff back in their proper locations. To repeat, all this started to get a new faucet installed. The faucet is very nice and a good upgrade. It has a nice spraying function and an automatic rewind function as the head can be pulled from its holder to rinse sink corners and othre things.
I wanted to mention a topic that usually is bandied about here at the casa and that is models. No, not those kinds, the ones that are not correct, might be barely accurate and are used to set policy for billions of people.
I got introduced to models via golf swing physics and the climate change debate. Barb, on the other hand, is a well respected model maker for a small segment of the scientific community. We talk about them and what they are good for and what they are not.
One of the problems of science is that the researchers and users fall in love with their models (and theories too) and become quite entrenched and defensive even when it is quite clear that the models are doing much less than helping. We are seeing some of playing out in the current news. The death and ventilator and needed hospital beds are all being estimated with a model. What we all have to be careful about is relying on these, or expecting them to be correct, and assuming that the output from the model is/are data.
When you build a model you make all sorts of assumptions. You can't be right about the important parameters usually because there are a lot of them and it's typically a difficult problem. If the output of the model doesn't match the data as they show up, you have to be willing or perhaps eager to toss out the model and try again. But that's difficult. Like many things in life and government and business, vested interests show up immediately. Cure the common cold? What about the nose wipe tissue industry -- can't be tossing them out of work!
So we know the models are going to be wrong, the other item you want to try to calculate a priori is an estimate of error, also known as uncertainty. How much leeway could we expect from the prediction. And if you are looking for it, you can see some of this. They plot a curve and there are shaded areas surrounding the expected value. This is an estimate of how high/low the data actually will be. And the further out you run the model, the wider this estimate should / must be. For example in climate forecasts they are predicting that they will know the global temperature within one hundredth of a degree in the year 2100. Anyone with some exposure to this sort of thing will give it no weight. But then someone like Al Gore takes it and runs with and then media jump on it and pretty soon we don't have single use plastic bags anymore and the price of gasoline is artificially bumped to help stop climate change.
So these can be dangerous and are not commonly understood by folks not in the business and sadly that includes folks that use them.
Ok, that's enough doom and gloom for the moment. The stock market up some more. Even NY deaths seem to be dropping and maybe we are seeing the down slope of all the virus stuff.
No comments:
Post a Comment