HNHacker News
TopNewBestAskShowJobs

wnkrshm

1,173 karma · joined August 28, 2014

optics
submissionscomments
wnkrshm··on AI is coming for your children
Why sneak, the article already shows that anyone with an agenda can now easily produce their own books, without much effort.
wnkrshm··on Missing Titanic sub faced lawsuit over depths it could safely travel to
Thethers and ships that can hold their position for thethered ROVs are very expensive. OceanGate didn't even have their own ship, they chartered one and that ship released their base platform from which the sub takes off.
wnkrshm··on Missing Titanic sub faced lawsuit over depths it could safely travel to
The sad part is, people who couldn't afford it also went: like a woman who spent all of her savings for the trip (she's on the BBC show).

Luckily she made it back up alive (3 times, since 2 times they couldn't find the titanic, they get told where to go by the mother ship).

wnkrshm··on Missing Titanic sub faced lawsuit over depths it could safely travel to
I think they couldn't modify the sub to be safe - the big worry is the integral part of the sub (and the one with the highest cost in this one), the composite pressure hull.

You cannot ensure its safety, since you cannot evaluate it for growing defects. As is best practice for all other DSVs in the world that use metal pressure hulls.

I suspect a metal hull and all it entails (way more complex design and careful selection of materials, equipment, passenger space etc.) would have cost too much.

wnkrshm··on Missing Titanic sub faced lawsuit over depths it could safely travel to
I think there was simply no economically feasible alternative to the composite hull, so Lochridge was fired.

I also suspect that if you go with a metal hull, you don't have positive buoancy anymore - you need syntactic foam. You need larger thrusters, those are heavier too, so more foam etc.

The engineering becomes harder and also costlier, in addition to the material and equipment costs.

What you gain is the ability to evaluate your pressure vessel for growing defects - which you didn't do in the first place (and can't do) with your composite hull.

wnkrshm··on Submarine missing near Titanic used a $30 Logitech gamepad for steering
They did the automated dive test, then kept diving without any idea how long the hull will last. They didn't automatically dive until failure.
wnkrshm··on Titanic tourist submersible goes missing with search under way
Then let me amend what I wrote: It's as difficult as designing a spacecraft, because you reach the limits of what materials that can be used can offer.
wnkrshm··on Titanic tourist submersible goes missing with search under way
Even then it's horrible, since the thing is sealed from the outside (as basically all crewed DSVs are)
wnkrshm··on Titanic tourist submersible goes missing with search under way
Don Walsh has been on the Deep Sea Podcast a few times, I can recommend it for anyone interested.
wnkrshm··on Titanic tourist submersible goes missing with search under way
The dome shape is a structural necessity to withstand the pressure, like in a soda can. Else you would need more material, more material means more weight, more weight means more bouyancy required which again means more weight etc.

It's like designing a spacecraft but with different constraints.

wnkrshm··on Titanic tourist submersible goes missing with search under way
Carbon fiber hulls are used extensively in remotely-operated vehicles, Edit: (I was mistaken there, those are metal hulls - but there is a plethora of research RVs with glass fiber hulls) e.g. the ones that the Navy uses (torpedo shaped) but they fail after a while. [0]

The glass fiber design is based on the Deepflight Challenger design which Virgin wanted to use but deemed only safe enough for one dive:

"Based on testing at high pressure, the DeepFlight Challenger was determined to be suitable only for a single dive, not the repeated uses that had been planned as part of Virgin Oceanic service." [1]

[0] https://www.mdpi.com/2077-1312/10/10/1456

[1] https://en.wikipedia.org/wiki/DeepFlight_Challenger

wnkrshm··on Titanic tourist submersible goes missing with search under way
They copied the glass fiber + titanium rings idea from the Deepflight Challenger that Virgin wanted to dive with. [0]

Virgin deemed the design not safe enough for more than a single dive and quietly scrapped the diving events: "Based on testing at high pressure, the DeepFlight Challenger was determined to be suitable only for a single dive, not the repeated uses that had been planned as part of Virgin Oceanic service." [1]

[0] https://www.compositesworld.com/articles/composite-submersib... [1] https://en.wikipedia.org/wiki/DeepFlight_Challenger

wnkrshm··on Reddit is removing moderators that protest by taking their communities private
Your analogy isn't complete.

The analogy is more like: 12 year olds playing a soccer tournament that draws a crowd and the people owning the yard are selling drinks and showing ads.

wnkrshm··on Reddit is removing moderators that protest by taking their communities private
If you pay, your data set is just worth more (Edit: to the company), since it proves you have money and can be made to spend it.
wnkrshm··on MrBeast has become a viral sensation for his acts of altruism
MrBeast is exactly the media phenomenon a broken system would come up with in some 80s cyberpunk movie, as a bandaid to give people the illusion of hope: the 'happy hour' show in a boring dystopia.

That is not MrBeasts fault, it's the environment's. Say you see MrBeast give thousands of people help, and you have the same problem as these people - sure, you're happy for the other people but why didn't you get any help? The show is over, you're left alone, your problem is currently not interesting.

Edit: There is also the power gradient in all of this performance that this video[0] highlights well. Nobody can complain about how they're being treated or portrayed, because they are in need. That is taking away something from the people who receive something, this kind of charity treats them as less - they're extras on stage.

[0] https://www.youtube.com/watch?v=1LfJ-VxuqIQ

wnkrshm··on Is AI killing the stock photo industry? A data perspective
Imo it's only right to have something that is made of public goods (i.e. everyone's data and human historic record) to become a public good in itself.
wnkrshm··on Is AI killing the stock photo industry? A data perspective
Real life is what good illustrators actually train from: figure studies, value studies etc. - from life. (Edit: you study like the old masters, not necessarily the paintings of them - though it is also done ofc, but not the most important part)
wnkrshm··on Is AI killing the stock photo industry? A data perspective
The difference is that now only the owner of a machine profits where before lots of individuals profited. More power to the owners, I guess.
wnkrshm··on It’s infuriatingly hard to understand how closed models train on their input
That alone would not be sufficient though, you could have learned about it by chance - it would be more useful as a meta-encoding, like a watermark, with multiple 'trap concepts' that are connected enough with everything the model knows as well as 'white spots' at certain points, e.g. about a fictional book or movie. Edit: so the existence/absence of ctrap concepts encodes the 'version'.
wnkrshm··on Tell HN: It's not possible to pay $5 toward planting a tree
Planting trees is the least-cost and least-impact part of actually cultivating a healthy forest that is not another monoculture patch of land (e.g. firs).

To get your trees to survive their early life and not get eaten by animals (tasty green branches and roots without nasty bark) or die from deer massage (they scratch themselves on the bark and can damage trees to the point of decades old trees dieing), you need people actually maintaining the cultivated area until it becomes a quality mixed forest - otherwise, only the fast growing trees will survive like spruce and beech.

Spruce acidify the ground and cover it with needles, nothing else grows there - so not good for biodiversity. Beech are low-density wood and don't bind much carbon.

Setting up the development of an old-growth forest is lots of work.

(Edit: if someone plants a tree, odds are it'll get eaten - you will not get a tree for 5USD, you'll likely feed the local fauna)

wnkrshm··on It’s infuriatingly hard to understand how closed models train on their input
I feel rather than people trying to poison or fingerprint the data of a third company, the AI-operating companies themselves will add such markers to their models, to find out who used their model output for training.
wnkrshm··on Unproductivity
Disclaimer: I'm not a therapist.

Try to find out whether your own emotions are blocking you - when I was going through a major depression without treatment, I felt like the worst, an absolute impostor even though I luckily wasn't perceived as such.

Personally, getting treatment made me feel much better and relieved the pressure I was putting myself under - my own perception of the quality of my work improved and it was easier to get into a flow state at work.

wnkrshm··on US military AI drone simulation kills operator, then takes out control tower
I mean, it's not hard to imagine that scenario in a simulated training environment (though as others reported, this story is misreported, there was no sim, only a thought experiment):

You have a training environment that is symmetrical, you have enemy drone operators, that use an enemy communications network to give commands to units and to control their drones. You could maybe model the ability of each component to function/to contribute to the operation as a bayesian graph.

You equally have the same infrastructure for your own drone.

Now you try to get the drone to pick the right targets to impact the enemy operation maximally - if the operator could possibly minimize the drone's contribution, the operator not vetoing will maximize the success of each target engagement.

You could see similar behavior with all kinds of "reach the goal in this simulated environment" training, where the model ended up using glitches or trivial solutions, if those weren't sufficiently penalized.

If the story as reported were true, the fault would be in the simulation environment - the model for the drone just picks the path of least resistance.

wnkrshm··on Japan’s government will not enforce copyrights on data used in AI training
Large IP holders could eventually just copyright every image that can exist if AI output is copyrightable - large IP holders can divide claims to the entirety of human illustrative output within a few years, without any significant white spots left. They sample the space, the company automatically publishes to their own platform and done.

Which claim is then violated is up to models again that classify and where it's a toss-up, companies will find a negotiated licensing price. A license for the models has to be paid by the competitor anyways, because they'll need to use it to classify their own creation against it to tell whether they're infringing.

That goes well with GPT6 running the legal show automatically.

wnkrshm··on HDF5eis: Storage IO solution for big multidimensional time series sensor data
You cannot load or write (may only have been one of those operations, strangely) a file containing non-ASCII characters with the C/C++ libraries if you target Win.

If you target Linux, those work out of the box due to the file system doing the lifting.

With windows we had to use the short file name (e.g. FILEN~1.EXT) as a workaround.

Also, you have to watch out for what the library does when it writes to (Edit: Windows) network shares - we've seen it write NaNs where there were values in the data it should save, maybe a latency issue, maybe a configuration issue - but not something you get told about.

I would like to move away from it.

wnkrshm··on Sam Altman Says OpenAI Will Leave the EU If There's Any Real AI Regulation
Good, space for better gardening of the technology.
wnkrshm··on A guidance language for controlling LLMs
Does it strike you that this conversation seems to consist of futurology forum clichés ("What progress will AI allow?" "AI's impact on physics") that probably inform this from the training data?

What also strikes me is the shorthands of communication here sounds like cliché from 2nd rate scifi novels, the likely source of the format.

Since what is cliché? It's what's present in the majority of a genre and what the training sees as structure.

wnkrshm··on AI can’t do hiring because it lacks the data
That's why you need a Bachelor's degree to be eligible for training, at least in the US Air Force. Edit: They need people who have proven they can study and understand material.
wnkrshm··on Ask HN: What are the most amazing algorithms?
Monte Carlo integration and its many many offshoot techniques (biased and unbiased) are still fascinating to me.

It's less impressive if you simply see the result, e.g. in ray traced images but what is done mathematically, is evaluate an expression of infinitely-nested luminance integrals.

wnkrshm··on The EU AI act is coming, this time for real probably
The EU is more in line with China on this one, I'm all for letting other societies try this cultural nuclear weapon on themselves first. See what sticks.

I feel that's the way things have been going culturally/economically in France/Germany: Look at the world (usually the US but now also China more), see what is useful/competitive, adopt it in moderation. So this stance is extremely in line with the usual approach, in my opinion.

← PreviousPage 2 of 21Next →