Hacker News Reader

Comment by dwheeler | Hacker News Reader

What I especially appreciate are the recommendations at the end. The Titanic was a disaster, but the good news is that it was carefully analyzed and changes were made to greatly reduce the impacts of problems in the future. For example, laws were changed changed so that there had to be enough lifeboats for everybody on the ship.

Currently software developers are often not learning from previous security disasters. I hope that will change in the future. There are already efforts to start writing up about important problems, and that will be a start.

View on HN

The regulation to have life boats for all was the trigger for the capsizing of the Eastland in 1915. The addition of lifeboats cased the ship to be too top heavy and it flipped over while boarding. This killed 844 passengers, more than the passengers killed on the Titanic [1].

While it’s important to have backups in failure cases. It’s important to also consider the impact backups have in contributing to other failure cases.

Another example, people often wonder why airlines do not equip passengers with parachutes in the case of a crash. The answer is that people do not know how to use them, and would likely injure or kill themselves. Furthermore, they wouldn’t help in the period of flight where accidents are most likely to happen, which is take off and landing [2].

Disasters are a systems problem. And when looking to mitigate, we need to consider the efficacy of the mitigation and the total influence of mitigation’s on contributing to the likelihood of other failure cases.

[1] https://www.smithsonianmag.com/history/eastland-disaster-kil...

[2] https://simpleflying.com/commercial-planes-parachutes/

Blaming the new regulation for The Eastland capsizing is verging on dishonesty. The boat was poorly designed and had already nearly cap sized in 1904 and 1906. It was bought for a bargain in 1914 because it was known to be unsafe. The addition of the new six lifeboats was only one modification amongst many which were done without due care and merely the straw that broke the camel back.

Lifeboats in particular were probably the wrong fix, not because of Eastland, which was just an unsafe ship and would doubtless have found some other excuse to kill everybody, but because lifeboat maintenance and testing causes injuries and occasionally deaths, while in the modern era the need to evacuate passengers to lifeboats grows smaller.

I think MAIB (the Marine [edited: this said Maritime but that's wrong] Accident Investigation Branch, a UK government agency) wrote a report about this at one point. In coastal waters particularly the ship's master is realistically never going to encounter a situation where abandoning to lifeboats is the right choice. Fire is the most likely reason to consider it, but modern vessels should be able to contain plausible fires well enough to reach safety, even if as a result the ship is damaged beyond economic repair. Just loading the passengers into lifeboats and putting the lifeboats into the sea is a pretty fraught endeavour, you will probably injure either a passenger or crew member doing it - and that's before any risk from being in a tiny boat out at sea.

More than 1500 people died when Titanic sank.

> Currently software developers are often not learning from previous security disasters.

Perhaps some developers aren't, but many are. It may also be that their recommendations or efforts are ignored or overruled ("that'll make things too slow/costly/etc").

Most importantly, Titanic-like disasters involving numerous losses of life generally prompt regulation or lawsuits at the very least. There is a significant financial incentive not to let accidents happen.

Data breaches or misuse of personal data is not punished (at least not in the US), neither by the government nor by the market. There is no financial incentive there.

There is a significant financial incentive not to let [cost-ineffective] accidents happen.

> Currently software developers are often not learning from previous security disasters. I hope that will change in the future.

Current security disasters are typically exclusively financial or social and don't involve the death of those impacted. So, in a certain sense, I hope this doesn't change in the future.

Reply on news.ycombinator.com