sooheon

37 ms·

421.
by sooheon 6y ago
> But if the liability is unlimited & unknowable, your new chemical might turn a profit & do some good - or you might be sued into bankruptcy, and you can't know for sure which it will be, no matter how much study you do. It
422.
by sooheon 6y ago
Yeah, it seems like keeping the cond inside the dotfiles, rather than in a dotfiles manager is the way to go.
423.
by sooheon 6y ago
It's only unwinnable if your definition of winning is to cause more harm to society than the value you create. Your arms race argument is valid, but I think there's a lot of progress to be made before that negative effect dominate
424.
by sooheon 6y ago
I want the use of chemicals priced correctly for negative externalities so that I don’t need to prefer one or the other.
425.
by sooheon 6y ago
Protection from damages should come from due diligence. Potential liabilities are unlimited only because potential damage to humanity is unlimited, and its worth pricing that in. This would mean successful developments free from negative ex
426.
by sooheon 6y ago
The problem statement is simple: there needs to be lawful accounting for negative externalities, even when discovered after the fact. And the price to pay should be roughly equal to the damage caused by the externality. Solve these two prob
427.
by sooheon 6y ago
The latter. It does not infer graphs from unstructured data, it gives a set of backend agnostic APIs for constructing featurized graphs, message passing and pooling/aggregating (with gradients being handled by your framework of choice)
428.
by sooheon 6y ago
The transformer is a fully connected graph neural network. Read https://arxiv.org/abs/2004.08483 for further ideas.
429.
by sooheon 6y ago
That the internet has affected human attention span is not an uncommon idea[1]. The submission literally has the internet-age phrase "attention economy," and we all understand exactly what it's saying. [1]: https:/&#x2F
430.
by sooheon 6y ago
Thanks for that image
431.
by sooheon 6y ago
They are notable for being good (or lucky) stock pickers bullish on trendy futuristic tech. They also have mounted an unprecedented marketing campaign that breaks away from typical Wall st old boy's club or stodgy retirement advisor st
432.
by sooheon 6y ago
> Isn't there a mental model that sooner or later they wouldn't be able to have returns higher than market returns Reversion to the mean. Also survivorship bias: at any given point there is a hotshot fund that outperforms the m
433.
by sooheon 6y ago
See also Parkinson's law: "Work expands to fill the time available for it's completion"
434.
by sooheon 6y ago
There is a more direct link from surveillance to physical harm (loss of freedom, imprisonment, death) than there is from lab meat. I agree, physical harm != harm, but harm != harm we should restrict others' freedom in order to prevent.
435.
by sooheon 6y ago
Obviously not. But the set of harms that should reasonably be given as an answer to "why not?" is not equal to the set of all imaginable harms.
436.
by sooheon 6y ago
Anecdotally, yes. Until fintech apps started actually functioning in the country where I live, the activation energy to set everything up to gain exposure to equity markets was too high for myself and everyone my age that I knew. I'm v
437.
by sooheon 6y ago
> I do think 20 year olds will be less prudent with money than someone who is 40 or 60. And? If you can trust them to get behind the wheel of a car with lives at their fingertips, to drink and smoke and do with their bodies as they wish,
438.
by sooheon 6y ago
So it offends your sensibilities, but GP is correct, it doesn't physically harm anyone. It probably will help reduce conscious suffering in the long run.
439.
by sooheon 6y ago
The only thing this is missing is OTP generation. 1Password does this for me, I don't know of any open source toolchain that makes this simple.
440.
by sooheon 6y ago
> By many measures retail investors have outperformed institutional investors since March Source please?
441.
by sooheon 6y ago
Debatable, but for personal computing there is no contest.
442.
by sooheon 6y ago
I'm surprised they're getting rid of long held keyboard shortcut conventions though. Previously, modal popup options (i.e. for saving docs) could be selected via Option-letter, i.e. "Save _a_ll" could be selected with Op
443.
by sooheon 6y ago
Is the idea is to avoid using local resources to back up to two places, but only once to send to cloud and then handle cloud->NAS elsewhere?
444.
by sooheon 6y ago
If it was your company, would you rather have 99% market share and 1% of profits, or 1% market share and 99% of profits?
445.
by sooheon 6y ago
Could this mean Jax support is coming via XLA?
446.
by sooheon 6y ago
> Marginal cost of developing App Store is nearly zero per app. And the marginal cost of copies of software is nearly zero, and yet developers still want to get paid, no? > Apple doesn't increase your reach. They merely let you t
447.
by sooheon 6y ago
And with the right marketing, sometimes they just want to be the kind of person who owns that drill, whether they end up using it or not.
448.
by sooheon 6y ago
You're right, that does help, alot. Sounds like in principle, Skija is an attempt to take the benefits of Skia and bring the shortcomings (for this use case) up to par. What's taken for granted is 1. Skia is better/more/
449.
by sooheon 6y ago
The point about information density cannot be stressed enough for a "pro" tool. Its keyboard friendliness is also highly underrated due to being a GUI -- I found it's rivaled only by emacs/vim in that I can go an entire
450.
by sooheon 6y ago
The 12 inch is still my favorite MacBook experience, having owned pretty much every form factor since pre-unibody white plastic. Can't wait to see what they can do in that hyper minimal portable niche with Apple Silicon.

More ›