Hacker News Reader

Comment by ChuckMcM | Hacker News Reader

I took a brief excursion on a private submarine once and the list of things that could go wrong that they asked to you explicitly release them from liability for was frankly rather frightening. It really put a damper on my desire for a 'casual' submarine from US Subs.

Given the description that the seas are quite rough I hope it is a simple matter of them losing a communications mast or something along those lines.

View on HN

Submarines are just about the worst when it comes to the duration from the second someone goofs up and the second where everyone's doomed.

And also the worst when it comes to the duration from the instant everyone's doomed to the instant everyone's dead.

TSorry, but that's a "no" from me, dawg.

I assume you mean "really short" and "really long," respectively.

An ex-submariner had a blog named after a saying from the community: "The stupid shall be punished."

Submariners are a glimpse of what humans will become in space.

Stupidity cannot be cured with money, or through education, or by legislation. Stupidity is not a sin, the victim can't help being stupid. But stupidity is the only universal capital crime: the sentence is death, there is no appeal, and execution is carried out automatically and without pity. Robert A. Heinlein, Time Enough For Love

Alternative: https://m.warhistoryonline.com/featured/u-1206-sub-sunk-dump...

It wouldn't surprise me if the sub community's saying came from a Heinlein reader somewhere in its past, similar to his popularizing TANSTAAFL ('There ain't no such thing as a free lunch') from 'The Moon is a Harsh Mistress.'

The causality arrow might go the other way. Heinlein was a naval officer before he was discharged on a medical. He never served on a sub, but he did serve on a destroyer, so he would have been familiar with their weaknesses.

If you watched Elon's BFR Mars presentation, you'll notice that a lot of it depends on fully automated docking and piloting of all spacecraft. Manual control is just too dangerous and screw ups are catastrophically expensive in those situations.

It seems like an area where automated systems should be really good. We already have automated systems that an land planes without human intervention and that's a far more difficult problem.

And yet we don't because humans are immeasurably better at dealing with edge cases than any automated system. That being said, I do expect those systems developed by SpaceX to be automated. But I'd hesitate to call it "an area where automated systems should be really good."

Would there not be an equal or a greater number of edge cases for self driving cars?

Absolutely. Which is why I'd never trust my car to drive itself -- not for years anyway. Most testing and training I've seen from companies has been done in ideal conditions with "non-ideal" conditions being considered some sort of edge case? Maybe it's because all the self driving startups are operating in sunny California. What a nightmare a self driving car would be on uneven, icy, dark roads. I'm certain it would turn itself off and ask a human to drive then. And that'll be the case for a while. Or in cities in asia where rules of the road are more like suggestions, and driving is completely reactionary.

No, driverless cars today aren't ready for reality.

I think it will be a gradual change, both to gradually improve the technology and to get users used to it.

I drove the new Volvo V90 for a couple of weeks a while ago and it of course have the adaptive cruise control and those things you are used to these days but it also can drive itself similar to a Tesla autopilot.

At first it was a pretty weird feeling that the car controlled the steering wheel (you still need to hold it otherwise it will beep at you) but gradually I got used to it and started to trust it more and more. At least when driving on the highway and after a while it became a really nice feature to have.

Truly autonomous cars are of course far away and the problems with snow storms etc. also needs to be solved but I do think more and more car manufacturers will keep adding one feature after another until we suddenly are there. And at that point it won't be that big of a leap for us mentally.

ISS hookups are still done by hand? Otherwise there aren't a lot of opportunities currently to do in-space docking. The lack of a solution could just be due to a lack of demand currently.

I wonder what calculus decided automated systems built by humans are going to be better than humans themselves.

Obviously computers have a lot of advantages, and machines can be precise. But that only applies to certain domains, and I can't imagine anything that provides certainty about which.

I'd imagine computers are more consistent. If the situation is one the automation can handle, it will handle it pretty much every time, assuming it's in good maintenance. Whereas a human's capability varies widely based on a multitude of factors from fatigue to distractions to medical issues to plain old overconfidence.

Ultimately, it's probably easier to quantify the likelihood of an failure in automation than it is to quantify the likelihood of a human error.

There must be considerable unscertainty in calculating the human error rate after eight months of interplanetary travel.

The same calculus that lets me deploy thousands of servers with terraform and puppet without a mistake. Sure there are problems, but you fix them once (in dev hopefully) and never have to worry about them again. Humans otoh have the potential to make the same mistakes over and over.

Computers can integrate data from numerous sources and sensors and respond very rapidly. Humans are optimised for a limited set of sensory inputs. When those senses can be utilized a person can be ridiculously effective. But space ships do not readily conform to this.

My step father was on sub duty for 25 years. He said he signed up because it was the closest thing to being on a space ship.

There's surely an obvious answer to this, but assuming the goof doesn't cripple the submarine's ability to move, and ignoring issues of secrecy, is there a reason it can't just head for the surface and be there in a few hours? There's nothing like the bends here for the passengers, right?

No, there isn’t.

Diesel subs always operate near the surface, they are basically floating mines.

I don't know about Argentina's submarines, but the diesel submarines in WWII could get down to around 1000 feet. I wouldn't consider that to be "near" the surface.

Most diesel submarines are actually electric submarines, with a diesel generator to provide the electricity for the electric motors and to charge batteries that can be used when the submarine is too deep to run the diesel generator.

What the parent is saying is that diesel subs don't get to stay down like nuke subs. Nuke subs regularly stay underwater for the duration of their deployments, which last months. Diesel subs cannot stay underwater for very long at all. We're talking on the order of hours.

In WW2, they could do that for like 60-90 minutes. Most time was spent on the surface.

Today, a diesel sub snorkels near periscope depth.

Subs control their depth by flooding internal tanks with water (more water = sub goes deeper). They can surface very quickly (seconds to a few minutes, depending on the depth) by blowing all the water out of those tanks with compressed air.

That's assuming something hasn't gone wrong that has caused the sub to take on water somewhere other than those ballast tanks. If a portion of the sub fills with water, the empty ballast tanks may not provide enough buoyant force to lift the sub.

Correct me if I'm wrong, but I believe a neutrally buoyant, or even (a little?) negatively buoyant sub can adjust depth, perhaps even surface?, under its own power.

It can, but there are actually very few of those types of subs around and they tend to be limited to research and recreation. Keeping several hundred tons neurally buoyant is kinda hard and if the sub has positive buoyancy then it requires propulsion to stay down or else an elevator ride to the surface is a one-way trip. Needing constant propulsion is bad for a military sub, and being stranded on the surface after any sort of emergency surface action (even a test) would also be a bad thing.

Fire (and fire suppression system) is another major hazard.

It's not just the ability to move. It's the ability to control bouyancy that's vital.

Anything that compromises that, and the sub goes into an irrecoverable dive.

Same with space flight I'd imagine...

That's how I feel about helicopters. Was super excited for first ride. Suddenly as we take off, realize just how incredibly vulnerable we are.

Marvelous technology, but terrifyingly easy for everything to go wrong.

Since helicopters can autorotate, they're actually safer in the event of engine failure.

But only if it's high enough for the rotors to generate lift as it falls, there must be a death zone between 5m and 100m.

Yup: https://en.wikipedia.org/wiki/Helicopter_height–velocity_dia...

I'm not so sure it's that simple. Airplanes can glide, autorotation is hard and rotors tends to break apart violently on impact.

Helicopters can autorotate but I wonder what the statistics are for them doing so and landing safely?

Did they mention dismemberment?

No, but then again I wasn't a journalist.

Dismemberment alive and other unspeakable torture is what it looks like by now. This story is a running nightmare in my head. Not least because I've met the perp on several occasions, and he did not come across in any way as the sick monstrosity we now know him to have been for many years.

Reply on news.ycombinator.com