Show HN: Would You Survive the Titanic?
gradio.app
gradio.app
The ship sank very slowly at first. No reason for panic, if things are just still. I would've known about watertight doors, and how ships haven't really sunk from iceberg incidents.
Plus, there's a wireless on board, and we're in a busy shipping lane. Plenty of time to get help. Plenty of lifeboats, too. We can offload people onto another ship and row back for more.
Hey! Look, lights right over there. I can almost make out portholes. They can get over here if things get ugly.
The list to port seems to be the big problem. They'll just use the bilge tanks, maybe. I barely perceive the downward list, since the ship is so long, and we're up so high. (I guess I'm in first or second class.)
The officers are so calm. Things are orderly. Capt. Smith seems calm. The band is playing. Damn it, let's get this pumped out and be on our way!
Some time passes, suddenly the ship feels like it's lurching down, perceptively. Wait!!! I run up to the front of the ship, and see water coming over the bow! Hell! The crew didn't give us any indications things were this bad.
I look behind me. Lifeboats are almost gone. Everyone else notices what's suddenly happening. It's 2:00 a.m. We're sinking. Suddenly we feel the list downward, the port list being gone.
Wait! Where'd that light go? Weren't they coming?
It's too late. The crowds are surrounding the last lifeboats. I panic and run aft.
FWIW You probably shouldn't realistically expect to offload large numbers of people to another vessel at sea, that is extremely hazardous. It was hazardous in 1912 and it's still dangerous in 2020.
If you're at sea and the master expects the vessel to reach a port, even if it's badly damaged, they will resist trying to evacuate passengers because they rightly expect some of the passengers would be injured or even killed in the attempt.
An uncontrolled fire, which is one of the worst problems you might have at sea short of actually capsizing†, is still not enough for a master to conclude the passengers must evacuate if it seems possible to reach a port instead. The intention will be - unless it becomes apparent this is impossible - to make for a port and keep passengers from inhaling smoke or burning to death meanwhile. Despite being unable to extinguish the fire your fire fighting teams may be able to buy you enough time to reach port.
Passengers often think of lifeboats first when anything goes wrong (perhaps partly because of the famous Titanic disaster in fact) but they are almost never the right choice, in fact Titanic probably harmed maritime safety to an extent by putting lifeboats top of mind for the general public.
† Titanic happened very slowly, but this is by no means the rule. And if a ship capsizes quickly lifeboats do you no good whatsoever. The Herald of Free Enterprise had about 90 seconds between the crew first realising there was a serious problem and the ship laying on its side - that wasn't even long enough to understand the situation and make an SOS radio call, let alone order passengers into lifeboats.
But yes, the offloading of passengers would have been extremely risky; and only recommended if the ship were—for sure—going down, without a doubt, and that passengers would need to get off.
Today, lifeboats are more-or-less little boats you see on cruise ships, and are used now and then for shuttling passengers to resorts. No biggie, in those cases.
https://en.wikipedia.org/wiki/SS_Eastland#The_Eastland_disas...
What would typically be the best choice? (I had no idea there was a choice)
I haven't looked into the site myself; I learned about it from David Spiegelhalter's book "The Art of Statistics: Learning from Data" [1]. The competition's problem of predicting who would have survived the disaster provides much of the motivation for the chapter "Algorithms, Analytics and Prediction".
In that chapter he applies "classification trees" to the problem, talks a bit about ROC (Receiver Operating Characteristic) curves, the problem of over-fitting, and other issues.
(Overall, I thought it was a good, well-written popular math book. A bit more technical than average, but definitely not a textbook.)
[0] https://www.kaggle.com/c/titanic [1] https://www.penguin.co.uk/books/294/294857/learning-from-dat...
According to a random page [0] on Google, £100 in 1912 is equivalent in purchasing power to about £11,497.41 in 2020.