davmre

48 ms·

271.
by davmre 13y ago
Garbage pickup is one of the many jobs that will probably be obseleted by self-driving trucks. So, at least in the medium-term future, we'll likely pay less than we currently do.
272.
by davmre 13y ago
Engineers and architects are paid up-front for their services. Musicians aren't. Do you have a suggestion for how they could be? It's true that musicians sort of used to be paid up-front, in the traditional label model where a r
273.
by davmre 13y ago
Yes, you could represent the score of a tile as the cost of an edge leading into (or out of) that tile. I don't know if it's possible to generalize any of the JPS ideas to this setting (my guess is that it'd be tricky), but b
274.
by davmre 13y ago
JPS assumes a very particular geometric structure, i.e., a grid with uniform weights, while A* is much more general and works on any graph with arbitrary edge weights. So there are lots of problems where the JPS tricks don't apply, but
275.
by davmre 13y ago
I don't understand where the cost comes from. Can't you double-spend at will until one of your transactions makes it into the blockchain? Until that point, as far as the Bitcoin network is concerned, you haven't actually spen
276.
by davmre 13y ago
The existence of the court is not secret, only its proceedings.
277.
by davmre 13y ago
Linear programming as it's usually defined allows for the solution to involve arbitrary real numbers, and it's that flexibility that allows general LPs to be solved in polynomial time. The restriction to integers turns this into an integer
278.
by davmre 13y ago
FWIW, the $10 million number is annual revenue. So they'd only need about 170k users at $5/mo ($60/year) to be making that much.
279.
by davmre 14y ago
> the probability that a specific Zerocoin belongs to you is 1/n I think you're missing the subtlety that the parent was trying to get at. Imagine that everyone redeemed their Zerocoins exactly five minutes after minting them; it'd be t
280.
by davmre 14y ago
This would depend on the distribution of times at which other people are minting and redeeming Zerocoins ... but under reasonable assumptions, you'd probably want to go with something like the exponential(1/N) distribution, since that's the
281.
by davmre 14y ago
All excellent points. My understanding is that fee waivers are granted more or less automatically for any plausible candidate, and no school would ever take anyone to collections over a denied fee waiver request (they just wouldn't process
282.
by davmre 14y ago
I don't think we disagree? My claim was that many bright low-income kids don't know about or understand the opportunities available to them. Obviously it's not because they're stupid -- we're talking about kids smart enough to go to Harvard
283.
by davmre 14y ago
Yes and no. Obviously there are some rich kids whose parents give them spending money, and don't need to work at all during school. But plenty of parents, even from wealthy families, expect their kids to buy their own booze/gas/concert tick
284.
by davmre 14y ago
No offense, but your comment is a symptom of the very problem the article is highlighting. You say: "A rural kid who makes it to a university will almost certainly have to work an almost full-time job just to cover their living expenses, bo
285.
by davmre 14y ago
They actually do rent manual cars: I searched a random day in April and 3 of the 10 cars available are manuals ( https://flightcar.com/cars/SFO/2013-04-02%2012:00:00/2013-04... ). This seems like a great benefit if you're looking to rent a
286.
by davmre 14y ago
Sure, lots of existing ML algorithms have efficient big-data implementations. But for new algorithm development, my (admittedly limited) experience is that the Matlab-prototyping stage usually comes well before the implement-at-scale stage.
287.
by davmre 14y ago
AGI has cool ideas, and is in some sense the "right" theoretical framework for AI, but it's not clear that it gives any kind of practical path forward for AI research. The main problem is that its basic idea -- an AI performing Bayesian inf
288.
by davmre 14y ago
As an ML researcher, this article isn't persuasive to me for a few reasons: - Computing power is getting exponentially cheaper even as computing requirements increase. The resources available to a university lab in the future will be much g
289.
by davmre 14y ago
Did you read the article? Harvard (apparently) violated their own policy for faculty email, which, among other things, states that "The faculty member is entitled to prior written notice that his or her records will be reviewed, unless circ
290.
by davmre 14y ago
No. But if Ferrari a) had no dealerships in your country, and provided no legal way for you to import, and/or b) decided they would only sell their $150k cars to owners of $1M houses, then (assuming you don't own a $1M house) they'd be refu
291.
by davmre 14y ago
Bounded supply + rising demand = rising price, no?
292.
by davmre 14y ago
Many people don't have the option to pay for HBO, either because they live in a country where it's not available, or because they don't have cable TV and it's not reasonable or affordable (for many people) to pay $60+/mo to Comcast just to
293.
by davmre 14y ago
Have you ever been to UC Berkeley? As a current grad student here: it's a fantastic research university and a beautiful campus, but the undergraduate experience is terrible. Classes are giant and there's very little faculty/student intera
294.
by davmre 14y ago
Here's one reason: a lot of state schools (Berkeley, for example) charge a lower tuition rate for in-state residents. Americans can gain state residency after a year or so of living there in a PhD program, but international students are alw
295.
by davmre 14y ago
It's a real department with seven profs. Lots of non-majors take CS classes, but even so, class sizes are relatively small (usually in the 10 to 30 range). That's really one of the major selling points for the LAC model of education.
296.
by davmre 14y ago
We actually had 13 CS majors in 2010: http://www.cs.williams.edu/people/alumni-directory/ . That went up to 20 majors in the class of 2011, so the gradient is positive!
297.
by davmre 14y ago
The method I described gives nonzero probability to every positive integer. The method you describe a) will never terminate and thus b) gives zero probability to every positive integer (since there's no stopping condition, even if you ever
298.
by davmre 14y ago
Put another way, you can't pick numbers from an infinite sample space because you can't create an infinite sample space. That's not true; there's nothing wrong with infinite sample spaces. It's just that there's just no such thing as a un
299.
by davmre 14y ago
This is actually exactly Google's strategy. The Siri-competitive voice search in Jelly Bean is really just an interface to Google search, which has been extended to answer a lot of the same basic questions that Siri can handle (for example,
300.
by davmre 14y ago
Yes, but Rails is a known quantity; everyone is aware of what's provided for you in a Rails app and can judge accordingly. It's harder to judge someone's work when you know it was written as a school project but you don't know which parts w

More ›