Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
Rhapso
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
21 ms
·
211.
▲
by
Rhapso
3y ago
I'm a engineer with memory problems. If you can build coping mechanisms you get very used to extrapolating and re-formulating things on the fly. But you can also forget edge cases and technicalities more easily. Ultimately this isn
212.
▲
by
Rhapso
3y ago
Corporate communication can't be isolated from being fundimentally coercive. Your employer can't politely ask you to reconsider anything without it being an implicit threat.
213.
▲
by
Rhapso
3y ago
Not just subsidized, but build and managed by the government!
214.
▲
by
Rhapso
3y ago
How do solar panels interact with condensation?
215.
▲
by
Rhapso
3y ago
As far as places to start with human genome editing, adding better debugging output seems a good one.
216.
▲
by
Rhapso
3y ago
Yeah and it is largely ignored. It's developmental problems in high c02 atmosphere that really scares me. Adults handle temporary or even extended exposure well, but we don't have studies about extended c02 for kids, and there are
217.
▲
by
Rhapso
3y ago
But hell itself is run exclusively off of cheap child labor with no safety equipment!
218.
▲
by
Rhapso
3y ago
If you have an angle grinder, quite a few premium/discount brand pairs become suddenly compatible with each other. Same batteries with different plastic bits to keep them incompatible. Remove the plastic bits and suddenly your black an
219.
▲
by
Rhapso
4y ago
It really depends on how fast both of you are going relatively to each other and who is asking the questions (and how fast they are going...)
220.
▲
by
Rhapso
4y ago
> No biological animal can evolve faster than this. Sure one does. Us. When one human learns something, they communicate about it and often, they all do. As always, robotics and automation doesn't enable any fundimental new capacity
221.
▲
by
Rhapso
4y ago
I have very bad news about what most of the electronics in the average home are doing right now. Especially the Chinese ones...
222.
▲
by
Rhapso
4y ago
Everything already was personal politics, it's just becoming politics you don't like seeing there.
223.
▲
by
Rhapso
4y ago
Simply put, no. Ultimately all Machine Learning methods are extensions of gradient descent. The "gradient" of the space of inverse hash functions is intentionally a fractal hellscape that is basically impossible to traverse. AI fu
224.
▲
by
Rhapso
4y ago
It's in compound, conductive liquids like plasma probably see stronger forces.
225.
▲
by
Rhapso
4y ago
It's interesting, when I was a teenager gtp would have been amazing (I cut my programming teeth on Markov model chatbots). As an adult who knows how the stack works, gtp is a letdown. Like chatting with a flatline construct in neuroma
226.
▲
by
Rhapso
4y ago
Time to find out if all of it did, or just the outer layers! Even if what remains isn't enough to notice conventional magnets an MRI has the Gauss to find out for sure....
227.
▲
by
Rhapso
4y ago
It may get ripped out by the superconducting magnet. Perhaps not in a desirable direction of "out" either (your fingernail won't stop it) In practice most people notice something is wrong before there is bloodshed.
228.
▲
by
Rhapso
9y ago
Even viewing it more generically, the same scaling issues apply, and the same inevitable re-decentralization awaits it. Centralized systems trade robustness for efficiency. The drive to lower costs will always result in re-centralization.
229.
▲
by
Rhapso
9y ago
Lightning networks to handle scaling limitations are the first step on the slow re-centralization of bitcoin. Eventually an overwhelming majority mining will be controlled by a centralized/federated oligopoly. Be prepared to rebel and
230.
▲
by
Rhapso
9y ago
It is actually really frustrating. The Tangle idea is pretty good, but the rest of it is crap. The product is getting such a bad reputation that using the good ideas for a better product later will be hard.
231.
▲
by
Rhapso
9y ago
When I call a system a black box I am not saying we cannot understand it, but rather we cannot easily extract the rationale used by the system to discern the classification. Layered neural networks are good examples of this where the signal
232.
▲
by
Rhapso
9y ago
https://en.wikipedia.org/wiki/Granger_causality
233.
▲
by
Rhapso
9y ago
I have issues with js as a language but those are not security issues. My security-related issues are with the nodejs package infrastructure. https://www.csoonline.com/article/3214624/security/malicious... Th
234.
▲
by
Rhapso
9y ago
No, just raised poor in Appalachia on spring water (no floride) and a cultural expectation that false teeth and dieing in your 60s are normal.
235.
▲
by
Rhapso
9y ago
I'm increasingly of the opinion that teeth just need to go. They don't heal, so all your efforts are only to postpone their decay. The pulp and nerves only serve to help them grow and then only provide a liability. Modern implants
236.
▲
by
Rhapso
9y ago
I can't tell what this product actually does within 30s of opening the page. Maybe I am not your target demographic but no indications of how the software works without a login is a turnoff.
237.
▲
by
Rhapso
9y ago
Thanks, I really like the core ideas of ssb but the nodejs implementation and lack of a security review give me a lot of professional reservations. I'm inclined to fork and use a more off the shelf method of securing p2p connections. T
238.
▲
by
Rhapso
9y ago
Are there plans to implement the ssb protocol in a language other than JS?
239.
▲
by
Rhapso
9y ago
+100 awesome for the annotated source. A lot of these demos get released as minified JS which makes it kinda useless to use them as a guide or hack on them.
240.
▲
by
Rhapso
9y ago
I think a way to look at it is this: The singularity happened, likely sometime in the 90s. It was, what any singularity is, the point where what came before it is not useful to predict what came after. It is no surprise that the event defin
More ›