throwawaykeno · HackerTrans

throwawaykeno

no profile record

comments

10 anni fadiscuss

FWIW the answer to all of the questions you ask in the first paragraph is "yes, but come on, that level of skepticism is a waste of time and money". It's like demanding proof that the reviewer wasn't hallucinating when reviewing the paper.

> but having a proof that humans can reason through from beginning to end makes people feel more comfortable.

Saying that humans can't reason about computer-checked proofs is like saying humans can't reason about source code; which is to say, it's obviously false in the general case.

10 anni fadiscuss

> What recourse does Oracle have now

Your parent already answered that question (although I don't blame you for not noticing):

https://en.wikipedia.org/wiki/Judgment_notwithstanding_verdi...

10 anni fadiscuss

I don't think it's fair to call this law suit frivolous.

From a developer perspective, the ideal and just end result has always seemed rather obvious. However, the multi-year history of this case demonstrates the correct legal answer not so obvious, at least from a legal perspective.

For example, settling the issue on fair use was definitely not how most people thought/hoped this would shake out way back in the beginning of the law suit(s).

10 anni fadiscuss

> (or don't goto any college)

I don't remember that being part of the conclusions.

Also, it's hard for me to take that study seriously. Good state schools can be as good as or better than Harvard. Basically I took the conclusion to be "there's a lot of greatness outside of the ivys"

10 anni fadiscuss

Sorry, I wasn't implying that anyone cares. I even state a few times in this thread that people specifically don't care in aggregate. And even if people did care, it wouldn't have a major impact on the extent to which barista jobs are automated.

But it is just interesting, from a technical/hacker perspective, that good barista work is a pretty hard robotics problem!

10 anni fadiscuss

I'm not sure why you're being down-voted. People driving their own truck for a living are probably not swimming in capital, and outfitting a truck with self-driving capabilities (or buying a new truck) is likely to be expensive. Furthermore, they are more likely to lose out to bigger players who can aggregate inevitable software/sensor upgrade/repair costs across an entire fleet.

The idea that everything will work out because working people can just become capital owners ignores the obvious fact that working people work precisely because they don't have capital.

10 anni fadiscuss

Right, but that's a captive market. No one goes to Sheetz for food; they stop there for gas and grab something to eat while they're stopped. So making the food ordering process a bit less pleasant is NBD.

McDonalds has to compete with other chains and prices are already bottom barrel. In McDonal's case, taking human interaction out of the equation seems like a losing proposition.

10 anni fadiscuss

The choice between NK/Iranian levels of isolation and complete free trade is a false choice. Plenty of countries have high tariffs for certain types of goods and do just fine. Even today.

More-over, I hardly think a tariff on the output of robotic labor applied only to expats who left the country for tax reasons is going to cause a trade war...

10 anni fadiscuss

> Why on earth...

1. Security

2. Friends

3. Family

4. Security

10 anni fadiscuss

> but it does sound from your checklist that a lot of those qualities could eventually be automated

Totally. But the current generation of dead-reckoning-and-maybe-a-bit-of-very-simple-sensing style of control for grinding/tamping/steaming/pouring won't be sufficient. I would be unsurprised if substantially improved sensors (perhaps even vision) are necessary before we see computers over-taking the best baristas in an abssolute sense.

And unfortunately I don't think there's much demand for that sort of investment for the reasons outlined in other posts on this thread (even if robotic baristas could smash the human competition in coffee making, they still wouldn't displace the neighborhood coffee shops.)

10 anni fadiscuss

Actually, Marx foresaw exactly "a renewing demand for labor" (although foresaw is kind of the wrong word; directly observed would be more accurate.) See "Capital".

He also foresaw the shift from manual labor to intellectual property as the essential component in production.

10 anni fadiscuss

> brewing the coffee and even foaming milk is something that a machine can do easily, and arguably better than most people, heck even latte drawing can probably be better done with a 2-axis drip nozzle than a barista with a bamboo

Is this true?

I'm mostly convinced about the brewing part (except that even the most meticulous machine will make mistakes, and a good barista will throw out a bum shot whereas a machine probably won't know the difference).

But I'm not convinced about the milk part. Do you have links?

(FWIW I totally agree with the tone and overall thesis of your post. And I really wish someone would deploy freshly ground-or-vacuum-sealed/fresh milk machines in airports/cafeterias/hotels/anywhere there's only a starbucks. But I also feel like better-than-average humans are still better than state-of-the-art machines.)

10 anni fadiscuss

A good barista pays attention to their work:

* A good barista will notice when they screw up the shot and pull pull another.

* A good barista will not burn or under-steam milk.

* A good barista will make sure you have the proper amount of milk and foam. (Getting a perfectly consistent result when steaming milk isn't really possible using a dead-reckoning controller or when using simple sensors like temperature AFAICT.)

* A good barista will make sure the drink looks as good as it smells and tastes.

  * A good barista will have no problem customizing drinks.

Note that I didn't say impossible. Just difficult. You can probably get to better-than-Starbucks quality without too much AI. Auto coffee shops that beat starbucks already exist. But I'm not aware of any auto espresso machines that can compete with the baristas at my local coffee shop. And even if they could, I'd probably only use them at work, at airports, and at hotels.

10 anni fadiscuss

And then I'll introduce tariffs on goods produced using your robots.

These are, fundamentally, political-economic questions. Not merely one or the other.

10 anni fadiscuss

IMO self-driving trucks won't displace humans behind the wheel. Trucks on public roads will be self-driving in the same way that planes are self-flying.

10 anni fadiscuss

Automating good baristas in a cost-effective way is actually a pretty hard problem, but automating your typical Starbucks espresso machine button pusher is trivial -- machines that can make passable cappuccinos already exist. But people don't go to Starbucks for the coffee; they go to Starbucks for the same reason they go to bars.

I think a lot of the service sector automation that will happen in the next 20+ years is analogous. Anything customer-facing is bound to stick around as long as people have spending money, while anything that's not customer-facing will slowly become automated.

E.g. The person at the counter in a McDonalds is the easiest person to replace -- all you really need is an iPad and a couple accessories. But if I were a betting man, I would bet that McDonalds will replace food prep jobs before replacing humans with kiosks in most locations.

10 anni fadiscuss

Yes, the entire institution's purpose is to help the FBI. There isn't a single biologist, computer scientist, chemist, or neuroscientist in the entire institution who isn't hell-bent on helping the FBI /s.

10 anni fadiscuss

> Another model is UBI and government as employer of last resort (or a equivalent strong labour movement).

Exactly. It's amazing to me that people claim there just isn't enough work. There is more than enough work -- there's always some way to make the world a better place. And in the USA there are fairly obvious things, from social services (in the broadest sense) to mundane upkeep to large infrastructure improvements.

The problem is that there isn't enough work that's motivated by profit motive in the near-term.

10 anni fadiscuss

Successful researchers are almost by definition under-paid when their research (area) reaches the top of a hype curve; the success of their research (area) directly causes a massive increase in their value.

10-15 years ago, researchers working on self-driving cars weren't in-demand engineers. They were academic robotics researchers primarily subsisting on grants rather than capital. And CMU isn't in the business of venture capital. They're in the business of doing the research that results in the innovation that venture capital then develops into products.

To put it in SV terms:

* Saying that NREC employees were under-valued during the early days of self-driving is kind of like saying that an early-stage startup was undervalued with ten years of hind-sight.

* Saying that NREC should have increased their pay for those employees to work on (new, different) research projects is kind of like saying that early-stage startups should match Google's salaries for their employees working on a (new, different) startup.

In other words: maybe scientists are underpaid, but the fact that successful researchers can leave academia and make more money is basically a tautology.

10 anni fadiscuss

Then why not just fingerprint everyone at birth (or shortly there-after)? Because that would be an blatantly unconstitutional violation of privacy.

Disincentivizing people from volunteering at a school by forcing them to surrender biometric data to the FBI for inclusion in a database that is normally only used to catalog information about convicted criminals is insane.