Slashdot Mirror
FAA Software Aims to Make Flights Easier
coondoggie writes "The Federal Aviation Administration (FAA) this week expanded a program that it says will reduce flight delays during the peak summer season. The Airspace Flow Program gives airlines the option of either accepting delays for flights scheduled to fly through storms or flying longer routes to maneuver around them. The agency said that it rolled out a new software program that ensures airports impacted by bad weather receive the maximum number of flights that can safely fly to them."
they aren't flying through storms, they fly AROUND storms- they aren't stupid. their choices increase from just straight up delays to either flying around a storm or a delay [they didn't do this already?] it is a good improvement- the delay could be a lot less and if it works well things will get better.
Sigs are too short to say anything truly profound so read the above post instead.
Is flying through storms all that good of an idea? Really?
"Flying through storms" isn't an accurate description - it's what happens when the mainstream press reports on a technical issue.
Nobody flies through thunderstorms. At least knowingly and on purpose. You fly between them (or over them if you're not stuck in a A320.)
Delays happen in the summer because the traffic trying to pass an area or line of thunderstorms enroute has to squeeze into the areas between cells. Controllers have to maintain a specified spacing between aircraft, so when you have less space for traffic, you have to accept less traffic.
In the past, the FAA would hold aircraft on the ground to keep traffic at a rate the affected area could handle. As I read the summary, it looks like they're going to give airlines the option sitting it out on the ground (rate limiting) or of rerouting well outside of the affected area - effectively a choice between a departure delay or a longer route with ahe increased fuel burn.
Choice is good.
Never shake hands with a man you meet in a fertility clinic.
Please turn in your geek card on the way out.
Anybody that has passed intro stats at any level knows (or should know) that effect size and statistical significance are two completely separate measures. You can have a 2% improvement that is highly significant or you could have a 20% improvement that isn't - without the actual data you cannot know.
Now, the question you are trying to ask is if, in medical terms, a 9% improvement is clinically meaningful. If I can show a new drug lowers blood pressure by 2mm hg every single time, the improvement will be statistically significant, but not in any way useful. Just yesterday I was running a hierarchical regression where the final predictor only improved the fit of the model by 3.3%, but the change in fit was certainly significant (p=.004)
If you want to use the word significant as a synonym for meaningful that's ok I guess (I wouldn't). But please don't add the word "statistically" to the phrase in an attempt to make yourself sound smarter. It just makes communicating statistics to the public that much harder for those of us that actually do it for a living.