HN user

jayajay

107 karma
Posts2
Comments256
View on HN

It's nice being one of those people who wasn't using React because of its strange licensing. Facebook has its history. Now I can say it's really good to see that Facebook make the right call.

I will still not be using React. It's just too heavy, same order of magnitude as jQuery.

Yeah, there's no reason humans can't interact with the physical world through auxiliary VR worlds. We already do this with websites, VR will eventually be an interface to interact as a connected robotic device.

At the same time, the person in VR would be giving learnable information to the robot while they are controlling it (e.g. like a driver for a self-driving car). Eventually, you could wade the robot off of VR and it would be able to carry out the task. So, VR could also be a way to teach robots.

I can see this being interesting in traffic applications. Instead of the red/yellow/green traffic lights we have at intersections today, instead imagine a "laser counter" system that would "tag" various vehicles as they approached and would send a special signal to the car whose turn it was to go. The car, being self driving, would be integrated into all of this and then proceed, assuming the other cars follow the same contract. All it takes is one car to not understand the contract for this to be pretty scary though. Maybe the road could be tolled: the spike-track or barrier won't go down unless your car has proven that it can cooperate in the intersection protocol.

It would be interesting to automatically generate the training data and feed it to the ML module in real time.

You could hard-code an interactive agent that periodically peeked and coerced the input (between itself and the ML function). This way the ML module is coercible.

It would even be more cool if the ML function itself learned the interactive agent function as part of itself.

I find it slightly weird and offensive that the products contain sexual orientation information. Are the products being bought to be paired with other products, sexually? Or are they being bought to perform labor? Does the labor in any way depend on the sexual orientation of the product? Isn't prostitution illegal in the country where this company operates?

Why You Can't Say 9 years ago

The beginning of this article is interesting. If I was from the distant future, maybe I'd ask these questions followed by the answers I'm expecting from 2017:

Q: Why are people still working? A: You need to work to live.

Q: Why do men and women appear so different? A: They are biologically very different.

Q: Why are people trying to increase, rather than decrease, the amount of jobs to be done by humans? A: Idiot, clearly we need more jobs, otherwise everyone will be poor!

Q: Why are people waiting for signs and lights in metal cages with wheels? A: Those wheels in metal cages are one of our most prolific forms of transportation right now.

Q: Why are people eating other animals? A: Who doesn't, meat is tasty!

Q: Why do people have different skin color? A: You are racist.

Q: Why are some people so round? A: Are you making fun of fat people?

Q: Why are some people in a rolling chair all of the time? A: Some people are born with disabilities or break bones.

Q: Why are people still cutting other people open to fix them? A: How else do you stop internal bleeding and remove tumors?

Q: Why do 51 people decide what happens to 100 people instead of 1 person deciding what happens to 1 person? A: Ever heard of democracy, (unfortunately) the most fair system ever created to date by our founding fathers (who also owned tons of human slaves)?

Q: What is a border and how come I can't see it? A: Ever heard of the nation state? You're already subject to it just by being here. Didn't anyone tell you that?

Q: Why do people get thrown in cages for misbehaving? A: What else would we do with misfits and dangerous people?

In its current state, even cutting-edge machine learning is pretty accessible if you have a good understanding of linear algebra and calculus. If you want to do have a deeper understanding of machine learning then vector calculus, tensors, graph theory, etc. can only help you.

If every known resource acquisition task was automated, and the discovery of unknown non-automated tasks could be automated to be automated, we'd be post-scarcity and the concepts of working and income wouldn't be useful metrics anymore.

So, yeah machines are a big black hole and our jobs are doomed asteroids spiraling into the black hole. As they spiral into the singularity, humans will be displaced at an accelerating rate, and it will take more ingenuity and effort for humans to maintain "work". And, for what? In the asymptotic limit, the outcome should be no more jobs and "work" in a the way we currently define them, and humans will be truly free to creative pursuits. Never shall a beautiful human mind be wasted on labor which a machine can do.

At some point, machines will be the dominant species pushing civilization forward, not us.

Until then, we're forced to work, we're forced into employment because our world does not simply give us what we want. Food and spears don't fall out of the sky, so we will waste our time hunting and farming until we figure out how to make those things "fall out of the sky".

It's not so much that humans are intelligent as it as that machines are just really dumb, which is really just because humans are still too dumb.

You are entitled to be skeptical based on your personal experiences. That one smart guy has had terrible experiences with 100% automated pre-2020 systems is not a good enough reason to shy away from the idea of pure automation.

Our life force diffuses over the options we perceive as viable and available. Many years from now, when we are all gone, kids will look back and wonder how humans could have ever been doing such jobs. Much in the same way we look back and find it hard to imagine how humans could spend the entire day chasing down prey and running from lions. It's that much of a leap that is required.

That's just bad design. The self-checkout could easily prompt a few questions before you check out, based on what you scanned.

"Hey, you got milk. Why not add a 4 pack of Oreos for 5% off? They are right behind you!"

You can't just word associate and chant to make things related

Oh but you can. It works on most humans... all humans. Everyone is subject to priming, bias, aggressive marketing and advertising.

Like MS Paint, but for VR. Interesting, the "websites" of VR will be worlds... I can totally see people throwing in this kinda stuff "just because wow new technology" in the same way that websites back in the 90s threw in snowflakes falling on the screen, custom pointers, epileptic lighting, audio, flash, etc.

This is a great share. The Fourier example is a great way teach about Fourier series and decomposition. Very cool!

Lately my Pixel has been achieving sub KBps speeds on very good Wifi connection (laptop in same room 100MBps), and it reminded me of the old days with dial-up on win 98 -- but worse. The estimated download-time for the LinkedIn app (70MB... gg) was a whopping 6 months! What a great way to get me to guzzle up my mobile data.

Even as a non-Facebook user, if you were to only post on HackerNews all of the time, a lot could be inferred about you. Apparently, the value of sharing exceeds the cost in privacy. There are many types of thoughts of that people would share on the internet, some more revealing than others (e.g. "Today is Friday" vs. "I am 190lbs"). Some people must be very skilled at posting inert content which does not in any way tie to their personal lives, others may not be -- or simply don't care, or believe the sharing value to be very high, etc.

In terms of psychology, UI/UX design and engineered content could try to maximize the likelihood of getting a user to post emotionally-charged or revealing content.

People aren't paranoid enough. Being paranoid carries so much negative stigma, it feels that we are selecting against it. Try to a remember the last time you revealed your paranoia and weren't a complete buzzkill. But hey, you can fear what you know. You can even fear what you know you don't know. But you can't fear what you don't know you don't know!

That difference approaches zero as programming more symbolic and as implementations become more available.

Libraries like TensorFlow, Theano, etc. are great because then I can focus on what I want (math) instead of spending my time telling a computer how it should multiply things together in an imperative sense.

He was talking about burners being used to be seemingly undetectable, which got me thinking about people with nearly imperceptible tells, and how both of these "nearly invisible" things are seen by big brother. You are right though... it's a very questionable segue.

To slightly bastardize the term "burner", people might think they don't have any burners, but everyone's got burners. I started to notice that one of my professors, around every ~25 sentences, he would capitalize the first two letters of the sentence. I mean it's a dead giveaway, because most people do this a lot less frequently.

But these figurative "burners" are everywhere. Little, tiny, seemingly untraceable ticks, invisible to the naked eye. Computers are telescopes in the bit world, eyes on us at all times. People don't seem very concerned. You can't just pour water and sand over your digital footprints to erase them... I would not be surprised if FB held on to my data 7 years after I deleted my account.

We've built this new world where we are accustomed to leaving tiny traces of our minds in nooks and crannies, where they are waiting to be scraped up and analyzed. As those traces get larger and those nooks and crannies get unearthed, the default level of privacy will shrink. One day it might be that privacy itself becomes an antiquated concept.

I deleted my Facebook in high school in 2010 also, out of paranoia of this exact future. The problem is that if someone sums over the comments I make on other media, they could deduce my identity anyways. In the big picture Facebook, Reddit, Twitter, HackerNews, etc. are all in the same boat. I imagine there exists software which sums over many common sources (FB, TW, HN, IG, YT, RD, etc.) to find correlations in prose structure, post time patterns, topic similarity, sentiment similarity, etc. to identify people, or at least reduce the search space. The NSA would likely use such software, if it existed.