Hacker News Reader

Comment by Junk_Collector | Hacker News Reader

It appears that Hurricanes are fairly clustered when broken down by decades so that the most active time period for category 5 hurricanes was the 2000's with the second most active being the 1930's followed by the 1960's. It's also difficult to determine the actual number of Cat 5 hurricanes in the past because previously they could only be measured roughly at Landfall and more modern measures can monitor any hurricane in existence meaning that there will be an increase in Cat 5's because of measurement bias. The impact of this is unfortunately, almost impossible to say with certainty. This is all covered in the first few paragraphs of the link you posted.

To claim a 3x increase would be an example of binning your data to exaggerate effect.

View on HN

Wow, this is me "binning to exaggerate effect"? So much for assuming good faith. You yourself admit this is the most active period for hurricanes, and that it's merely "difficult to say with certainty" if it's statistical variation, and you completely dodged my question of how likely you reckon such a possibility actually is. Meanwhile I haven't even tried to look at different bins, I just picked 2000 randomly. If you're going to claim I'm cherrypicking the bins here, then wouldn't it make sense to show me a more reasonable bin that that would contradict mine?

Well no, what I said was, a decade ago was the most active period for Hurricanes. This current decade is less active than the previous one, or the 1930's or the 1960's. Most decades have little to no Cat 5 hurricane activity so if you take the last 2 decades and compare them to the previous 10, then Hurricane density seems 3 times higher even though it is currently lower than it was 70 or 100 years ago depending on your measurement period. If we were to bin in 20 year chunks, then it would put this bin as the most active by a small margin. The most reasonable way to organize it would probably be year by year but category 5's are a relatively rare event so you don't see much variation if you just compare individual years. Most are 0, with high points being 2 a year, and a very rare 3 in a year. The real issue is that extreme hurricanes are clustered events. There seem to be a number of years where they hit heavily, then stop for a while before starting back up and we only have about 100 years worth of records which isn't enough to make a lot of meaningful predictions for something that seems to run on a 30 to 40 year cycle.

This is without touching on the issue of measurement bias, which is extremely difficult to determine the level of.

Reply on news.ycombinator.com