Hacker News Reader

Comment by ValentinA23 | Hacker News Reader

A 1 in 36 million chance for three breaks in one day.

Why is this analysis focused on the Baltics? That's p hacking, given that it happened to happen in the baltics.

Let's instead say there are roughly 20 ocean regions we would post hoc consider "the same". Now, given a breakage, what is the probability of at least two more in the same region and day? This is a Poisson distribution with lambda=200/365/20. The probability of two more independent breakages is 0.04 % for that specific day.

But again, picking a specific day would be p-hacking. Zooming out, an event that rare is expected to happen every seven years or so.

Now, "every seven years" is a far cry from "1 in 36 million." Whenever you get crazy p values like that, there is often an error or overlooked assumption in the analysis.

----

If you like this sort of thing, have a stab at forecasting competitions! I can recommend the Metaculus Quarterly Cup. The current one is in full swing so use the remaining 1.5 months of the year to practice and then you're set for when the January edition starts.

I see, this was in fact what I had in mind. The maths I posted represent the horizon of my knowledge in probability and was surprised how well o1-preview was able to output correct numerical calculations.

Having said that how would the odds look like if we factor in the fact the Baltic Sea is one of two zones with the most geopolitical tensions (along with Taiwan).

---

Thanks for the Metaculus recommendation. I was a bit disappointed in the lack of maths in the comments in general. Can you recommend something in the vein of Leetcode with various degrees of difficulty, from very basic to advanced problems ? I'm both interested in probability and statistics

Something in thr vein of leetcode would be really useful to train people in forecasting, but given how subjective it is maybe difficult to pull off.

That's assuming independence. I'm not ruling out sabotage but the world is often not fully independent. A storm or an anchor both may affect multiple cables if they're in generally the same area which would definitely make the probability far more likely than those stated. (edit typo)

Indeed !

https://www.datacenterdynamics.com/en/news/lithuania-sweden-...

>Lithuania-Sweden subsea cable cut, was 10m from severed Finnish-German cable

I don't buy that number (no source is attributed to it), or rather, I don't believe there's a single incident causing this.

The C-Lion1 cable is predominantly North-East - South-West whereas the BCS cable is NW-SE. They do meet, but the C-Lion1 operator Cinia says their cable broke about 700 km from Helsinki, east of the southern tip of the Öland island. That's easily over 150 km south from where the cables meet.

Also, C-Lion1 was reported broken at 4m, and the BCS cable at 10am the previous day.

the contrast with independent random events is exactly the point of the comment you've replied to, isn't it?

The grandparent comment is total nonsense which sounds smart but is not. Damages from accidents are not independently random either. Or do you think it is virtually impossible that 140 people die on the same airliner? It is likely the same ship cut both either by accident or intentionally.

I am leaning towards sabotage but that two cables were cut means very little.

What are the odds 140 people would all die from an airplane crash on the same day? Wild

The point is that even without a malicious actor the odds are way lower than you'd think.

I was told in my stats class that events in the real world are almost never truly independent.

“Knowing that 200 undersea cables break every year globally, estimate the probability that 3 cables break in the baltic sea on the same day.”

I’m stealing this to use for grad-student mock-interviews—thank you!

Hint: The cables are often very close. If one breaks, the otherone also breaks :-)

Why would cables close to each other break?

because if it's an accident and someone is dragging an anchor behind them, if the cables are only meters apart then they are going to cut both

Are they?

Wouldn't they notice after the first cable? I imagine it would be noticable

Right, if it’s a case interview, then higher accuracy ought to prompt the interviewee to ask: (1) Do the 200 cuts typically occur in clusters? (2) What’s the typical density, eg are they usually collocated? (as an alternative to the above) (3) Are there pathways that avoid the sea but connect Europe and North America (getting at density in the sea in question) Etc.

That’s what makes this one so good—lots of opportunities to extend or roll-back difficulty.

I was surprised to see so many upvotes this morning and was disappointed when I realized it wasn't for another comment I made about the Anthropic Principle.

My take is that in face of coincidences supporting the emergence of intelligent life, we should expect to observe coincidences unnecessary for the emergence of life too.

An analogy: imagine you have lost the key to your mansion and try to cut one at random out of a metal sheet. If it can unlock the door, then chances are that you cut unnecessary notches (the analogy only holds for warded locks and the key you crafted is a master key).

See: https://news.ycombinator.com/item?id=42178306

I'm wondering where I'm wrong in my reasoning because the implication is weird.

What are the chances that they break in close proximity spacially, but not temporarily? (I'm assuming that it would be headline material if the lines had disconnected within minutes)

Tangent: an attacker trying hard to provoke that kind of accident would likely not have a very fast success feedback. "Let's try once more, for good measure"

Still pretty decent, given the right circumstances.

For example, the 2011 earthquake in Japan resulted in damage to 7 cables[0]. But it wasn't the quake itself which instantly broke all 7 cables - they were destroyed by underwater avalanches triggered by the earthquake. Avalanches can occur hours after a seismic event, and some underwater avalanches go on for days.

I highly doubt that's the case here, but if you're asking about chances it's not as unlikely as you'd think!

[0]: https://www.theverge.com/c/24070570/internet-cables-undersea...

A clever answer would be "it's a 50/50 chance, either it happens or it doesn't". That's statistics my simple brain can comprehend at least.

In what way is that clever? It's clearly wrong. If it were true, we'd experience three breakages at least 150 days of the year, every year.

One a day is happenstance. Twice is coincidence. Three times is enemy action.

Or they’re right next to each other and the laws of physics continued to exist for the object that struck them.

You can't even assume they follow a normal distribution. For all we know, ships drop anchor more on certain days or weather conditions. That's just the start of the rabbit hole.

I'm supporting gleenn who beat me by seconds to much the same observation.

Clusters are a thing.

And what about adding in the chances of a Russian spy ship seen relatively near by only a few days earlier:

https://www.theguardian.com/world/2024/nov/16/russian-spy-sh...

Irish sea is relatively near to the Baltic sea?

According to Google it's 854nm.

The spy ship is alleged to be able to go 15 knots which means it could make the distance in 2.5 days.

https://ukdefencejournal.org.uk/russian-spy-ship-present-off...

Russia itself is nearer

FFS: NM, not nm.

Today’s that day. Start the clock for the next one.

Reply on news.ycombinator.com