danblick

From a security standpoint, it seems like there's a problem treating everyone's social security number as if it's some kind of secret key.

Has there been any real discussion about alternatives to the present system? How else could authentication work for opening a bank account?

I imagine that the present system survives (1) because of inertia, and (2) because it doesn't require much infrastructure and so it's relatively cheap.

Maybe the next step is something like putting a chip into driver's licenses and ID cards nationwide?

Your argument that credit rating agencies shouldn't be on the internet is really terrible.

People pay credit rating agencies to retrieve their credit reports, get credit scores, and open disputes, and the best way to do all that is online.

In a simulation environment you know ground truth. (Where objects actually were vs. what you saw.) You can play the same scenario from slightly different angles. Or different conditions (rain, snow, low light, bright light). You can look at counterfactuals (what would I have seen if I had done X). You can test the behavior of your system without actually setting a computer loose in the real world.

From "real world" recorded data you have none of these things.

This is the difference between observational and experimental data and it's a huge difference.

Miles of data is not the best metric. If you've ever tried to build a machine learning system you'd recognize some of the issues here. First off, where are the labels? How are you to interpret all that data?

Say Tesla has recorded video of some dude driving down his street in LA and their system detects a car door opening in his data stream. How can Tesla find out whether or not there actually was a car door opening? Human labelers (for billions of miles of data)? Multiple instruments?

If I were designing a system, I'd want a car with extra instruments on it that would help me validate what the primary instruments detected. The idea of just recording a bunch of live traffic seems pretty lame. (You have no way to validate any of the conclusions you draw from the data and hence no way to improve your ability to draw conclusions from the data.)

Did you see the article about Waymos test environment at Castle? That, along with a boatload of simulation, is exactly the kind of testing you need to do to develop this kind of a system.

Has Tesla ever demo'ed their simulation environment or other tests?

> they've already put in the hardware necessary for FSD so when the software is ready a simple OTA software update is all that's necessary for you're car to have full L5 autonomy

That's bull. Every other company working on fully autonomous driving is using Lidar for obstacle detection, but Tesla's cars have none (only radar and cameras). I see no reason to think their software is any better than anyone else's (in fact I would say Waymo almost certainly ahead of them there).

Why should I have any confidence that they will be able to get better results than anybody else using cheaper hardware?

Yes, it seems to prevent states from having their own safety policies. It makes an exemption for "performance standards" which I guess is omissions.

Page 2: http://docs.house.gov/meetings/IF/IF00/20170727/106347/BILLS...

It seems like the legislation would delegate responsibility for coming up with such policies to the Secretary of transportation and gives them 24 months to come up with such rules.

‘‘§ 30129. Updated or new motor vehicle safety stand- 10 ards for highly automated vehicles 11 ‘‘(a) SAFETY ASSESSMENT CERTIFICATION.— 12 ‘‘(1) FINAL RULE.—Not later than 24 months 13 after the date of the enactment of this section, the 14 Secretary of Transportation shall issue a final rule 15 requiring the submission of safety assessment certifi- 16 cations regarding how safety is being addressed by 17 each entity developing a highly automated vehicle or 18 an automated driving system. Such rule shall in- 19 clude— 20 ‘‘(A) a specification of which entities are 21 required to submit such certifications; 22 ‘‘(B) a clear description of the relevant test 23 results, data, and other contents required to be 24 submitted by such entity, in order to dem- 25 onstrate that such entity’s vehicles are likely to

http://docs.house.gov/meetings/IF/IF00/20170727/106347/BILLS...

I wonder what fraction involved smartphones?

https://mobile.nytimes.com/2016/11/16/business/tech-distract...

Hmm. This might be kind of blasphemous, but I think I'd recommend against the dragon book? As I remember, the emphasis in that book is on syntax-directed translation. I might argue that the less you're thinking about syntax and lexing/parsing crap the better. (For that reason: ignore other answers that tell you to learn about a particular parser generator (e.g. ANTLR). That's just fluff.)

Really, if you're coming to programming language design with the thought "I'm going to make an imperative, object oriented language" (with parallelism as an afterthought), you're doing it wrong. The world has enough of those already and you're going to invent something worse than what's already there.

Probably, instead of inventing a new language (say Matlab for matrix operations or Prolog for logical reasoning), you'd be better off implementing a library that handles the same concepts and embeds into another language (which is really what happened with Tensorflow or MapReduce to think of two examples).

(Grune's book "Parsing Techniques" is a great reference on parsing crap, but the secret is that if you design your grammar to be LL(1) you can parse your language using recursive descent: you only need a fancy parser if you designed a more complicated grammar (why'd you do that?))

Recommended book: The Reasoned Schemer. It's a cute (maybe too cute) book that shows how to implement a logic programming language (~datalog) using scheme as a base language. The Wizard book (structure and interpretation of computer languages) also has really cool examples that I think 60% of programmers I've worked with in industry don't fully appreciate.

Software developers are already highly paid with lots of perks. Do they need unions?

Exactly. It's possible there are other features that would have more predictive power, they just haven't figured out how to quantify them. The title of the paper could be, "We're still not very good at describing what makes people attractive."

Another issue: If you want to argue "we all deserve a share of productivity increases", then who is "we"?

Does the local community deserve to benefit from the automated cog factory? Does the state? Do all Americans? All humans? Those with coercive power?

It bugs me sometimes that you can argue "let's all share the wealth" but do it in a self-interested way by making a choice about who deserves what.

That chart is interesting, but I'm not sure how to interpret it.

I think the authors are suggesting, "people are not being paid in line with what they produce."

I think the chart may show something else, which is "an increasing amount of work is being automated (and goods are produced by no one in particular), and this benefits owners rather than workers."

If I make widgets and you build a robotic cog factory, why should my wages go up?

Weird. I was so sure that "Michael Jordan" would be the Berkeley CS professor in this context. I guess not!

I think this is an interesting topic, really.

Credentials are all about signaling. But what is it exactly credentials communicate? Is it, "person X learned about topic Y"? I'd argue it's a lot more than that. If you have a degree from Stanford, then the information "this is a person who was admitted to Stanford" has a lot of value in addition to just "this is a person who took Stanford classes". (How much would you care if someone told you they'd sat in on Stanford classes without enrolling? Not very much, right?)

There was an Econtalk episode on this:

http://www.econtalk.org/archives/2014/04/bryan_caplan_on.htm...

I think MOOCs are not particularly good as credentials in their own, but the projects you do in them can be very useful in getting a job. (This is kind of the "human capital" model discussed in the podcast.) Maybe their lack of exclusivity is part of the reason they're not more positive signals on their own.

Think about the projects you'll do and what you'll be able to say you did after the course is done.

Don't count on the credential itself to say too much.

There's an insane quote from John Ehrlichman about the origin of the war on drugs in the Nixon Whitehouse:

https://harpers.org/archive/2016/04/legalize-it-all/

"The Nixon campaign in 1968, and the Nixon White House after that, had two enemies: the antiwar left and black people. You understand what I’m saying? We knew we couldn’t make it illegal to be either against the war or black, but by getting the public to associate the hippies with marijuana and blacks with heroin, and then criminalizing both heavily, we could disrupt those communities. We could arrest their leaders, raid their homes, break up their meetings, and vilify them night after night on the evening news. Did we know we were lying about the drugs? Of course we did."

I wonder, is that credible? If so, it's kind of amazing how little has actually changed since those laws were put in place.

(Edit: okay, there are some reasons to doubt that quote) https://www.quora.com/Is-John-Ehrlichman-a-credible-source-w...

If I have an intellectual hero, it was Ericsson's post-doc advisor Herb Simon. In "Peak", Ericsson talks about work they did together on the mental representations used by expert chess players. I recently reread some of a lecture Simon gave in 1982 ("Reason In Human Affairs") and it is pretty much golden. Simon is the person responsible for the idea of bounded rationality, but in that talk he also captured ideas about information overload/"quality of information"/"media ecology" that were talked about later by Neil Postman and Nassim Taleb and are especially relevant today although they haven't quite sunk into our culture yet:

https://books.google.com/books?id=TYnI9tbHkCEC&pg=PA93&lpg=P...

Since you mentioned both deliberate practice and 10,000 hours: are you familiar with Anders Ericsson's criticism of the 10,000 hours idea? Do you think it's still a concept worth talking about?

Are you assuming travel is one-way?

That seems okay for probes and maybe problematic for people.

Although the poor slob plan comes to mind:

https://books.google.com/books?id=7ejGJaFjLccC&lpg=PA154&ots...

And not much lidar data, right? I don't work on this, but I imagine the quality of the data set makes a big difference.

From what I've learned, lidar and radar are typically used for object detection (i.e., avoiding other cars; perhaps augmented with camera data), while cameras are used for things like lane detection and traffic sign detection. If Tesla is trying to solve the same hard problems as everyone else, using weaker equipment, I wouldn't bet on them having a lot of success.

Also, kind of the point of this article is that Waymo has a test area where they can get real data along with the ground truth. That seems much more valuable than unlabeled/random data.

I recently finished Tim Wu's "The Master Switch", which is a history about the early radio/telephone/television/film industries, consolidation in those industries, and the effects on free speech.

One of my favorite examples from the book is the story about Hollywood and the Production Code in the 1930s. This was a voluntary code that the major studios obeyed for their own commercial interests (they wanted to avoid boycotts by groups like the National Legion of Decency). Because the film industry was centralized (with a small number of companies controlling all film production and distribution and most of the national theaters), this private decision by a handful of companies led to effective censorship in nearly all film in the US for ~20-30 years, all without any government action whatsoever.

I think we're in the same boat here with Google and other major internet infrastructure providers. Because a huge amount of power over the internet is concentrated in the hands of a few companies, those companies need to be very careful about the way they handle censorship.

"In an information industry the cost of monopoly must not be measured in dollars alone, but also in its effect on the economy of ideas and images, the restraint of which can ultimately amount to censorship."

"The story of Daniel Lord and the Legion of Decency goes to a central contention of this book: in the United States, it is industrial structure that determines the limits of free speech."

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

"Orwell warns that we will be overcome by an externally imposed oppression. But in Huxley's vision, no Big Brother is required to deprive people of their autonomy, maturity and history. As he saw it, people will come to love their oppression, to adore the technologies that undo their capacities to think." - Neil Postman in Amusing Ourselves to Death

What you are proposing:

(1) Is bad economics. In most cases, getting rid of occupational licensing would create wealth. To take law/lawyers as an example, see http://www.econtalk.org/archives/2011/09/winston_on_lawy.htm...

(2) Would have the practical effect of taking away my livelihood and ability to work. Would have radically altered my life, personally, for the worse.

Great video. My major takeaway: vertical farming is not economical for many crops due to the cost of electricity. Another takeaway: vertical farming might work well for some crops that are mostly water by weight and have low energy requirements (e.g. bazel, lettuce).

I wonder how the numbers work out. I guess they'd want to avoid (1) limiting themselves to certain crops and (2) limiting themselves to specific geographical areas in order to make their market as large as possible.

From what I've heard of the Jobs/PARC visit I don't think I'd characterize it that way. In Walter Isaacson's "The Innovators" he says that Xerox had taken an ownership stake in Apple and some higher-ups decided to give them valuable tech.

Also, Alan Kay had a Quora answer on this:

https://www.quora.com/What-was-it-like-to-be-at-Xerox-PARC-w...

I haven't heard of Plenty before. At first I thought they wanted to build indoor farms in cities, but I guess you could still build them someplace else (where real estate is fairly cheap) and make turnaround faster by taking advantage of indoor technology instead of using fields. [The articles mentions LocalGarden and FarmedHere, which I guess is where my impression came from.]

Is that where they hope to make their money - getting consumers to pay for fresher food? Or else, can they grow better produce indoors? Just trying to understand their value proposition.

Not long ago I was reading about a different thing called Farmbot. There's a discussion online about how much area you'd need to grow all your own food. They calculated something like 250 square meters using black beans. I was really impressed by how much space food production takes up. It's funny to think that just to survive, each of us living in a city depends on a big chunk of land somewhere else.

https://farmbot.io/2015/10/14/how-much-food-can-farmbot-grow...

Your comment reminds me of reading something, I think maybe from Judith Donath, about how throw-away accounts tend to be bad for online communities.

Something like ~ https://www.wired.com/2014/04/why-we-need-online-alter-egos-...

Companies have an incentive to make their employees more productive, but other things being equal, you'd prefer to teach your employees non-transferable skills rather than transferable ones, since non-transferable skills can't be used on the job market to bid up wages.

There's a comment here about how conferences are great to help your developers make connections in industry. That seems like a mixed bag -- why should companies help their best developers to find their next job somewhere else?

There's an interesting Econtalk episode on the book "Learning by Doing" where this came up:

http://www.econtalk.org/archives/2016/05/james_bessen_on.htm...

[To summarize: when the textile mill was invented, skilled individual labor became much more valuable (because skilled workers were much more productive), but wages didn't rise much over a period of 50 years. This finally changed with the introduction of interchangeable parts in machines, which made job skills transferable, which allowed skilled employees to demand higher wages from their employers.]

Maybe it's not about grocery per se but rather about meal delivery?

Yes, you could abuse that power to allow double-spending. For what it's worth, I'd say I'm a bit concerned about this, but it seems many aren't.

https://news.ycombinator.com/item?id=14485430