HNHacker News
TopNewBestAskShowJobs

mreid

461 karma · joined July 15, 2009

Machine learning engineer at Apple. Formerly an academic researcher in ML at the Australian National University and NICTA. Hobbies: running, synths, piano, go.
submissionscomments
mreid··on NYU Professor Scott Galloway: Ouch
> There is nothing disrespectful here.

I disagree.

The point is not that the course might be mandatory for a business student (it wasn't - the student picked it as an elective), nor that he shouldn't make and execute such a plan if he thinks it is in his best interests.

The disrespect was in telling the lecturer, "we both know this is a waste of my time so tell me how to get it over with as easily as possible".

It's like constantly checking your phone while talking to someone in person. Even if you are listening to what they are saying, you are signalling to that person that there are more interesting things than them which demand your attention.

mreid··on NYU Professor Scott Galloway: Ouch
My wife is a lecturer and had a related experience with a student.

Sometime early in the semester after the course outline had been presented to her students, my wife received an email in which a student put forward his plan on how he was going to game the course to get a pass with the minimum amount of work. It went something like this:

According to the course outline, I plan to just read the poetry and one of the books. If I do essays 1 on the poetry, essays 3 and 4 comparing the book to the poetry, skip essay 3 and get 50% on the exam then, by my calculations I should get a pass. Could you tell me if this plan will work?

I've heard of time management but I'd never seen such a blatant disregard for actually wanting to learn anything in a course before. It didn't occur to the student that telling a lecturer that her course was just an obstacle course to be rushed through as efficiently as possible was a little disrespectful?

mreid··on Warning: Your reality is out of date
Your comment about Cadillacs reminded me of the scene from The Wire where Snoop buys a nailgun: "Man says this here the Cadillac of nailguns. He means Lexus but he don't know it."

http://www.youtube.com/watch?v=w8zavPW3Bus

mreid··on Google did something seriously wrong
I agree with you that Google has their work cut out for them trying to get into this market, but that doesn't excuse their mistake.

If you think it does then I am very curious to hear how you felt when Microsoft leveraged their desktop strength to compete with Netscape in the browser wars.

mreid··on Rands In Repose: A Story Culture
I don't doubt that close friends are an invaluable source of meaning but I think you may have missed the point of Rand's post.

I'm pretty sure he is not arguing that there is deep wisdom to be found in 140 characters. Rather, through your interaction with many tweets from many people you can distil meaning.

The point of Twitter isn’t knowledge or understanding, it’s merely connective information tissue. It’s small bits of information carefully selected by those you’ve chosen to follow and its value isn’t in what they send, it’s how it fits into the story in your head. There are great stories to be found on Twitter, but you have to do the work.

You're not precluded from discussing tweets with your friend.

mreid··on Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo
“This is the sort of English up with which I will not put.” -- Winston Churchill
mreid··on Google stole Nexus name, says PKD family
I largely agree that it is a silly lawsuit but, to play devil's advocate for a second, there are a few factors that make it slightly less cut and dried.

Firstly, Google's phone is the Nexus One while the androids' model in "Bladerunner"/"Do Androids Dream..." was Nexus Six - presumably the last in a series beginning with something like Nexus One.

Secondly, how come the "Droid" has to mention George Lucas's claim to the term but the "Nexus One" does not need to refer to PKD?

Taking off my devil's advocate hat, I think that the name "Nexus One" was clearly inspired by Bladerunner (engineers/nerds working on a platform called "Android" was always going to lead to lots of robotic cultural references) but that does not mean the estate of PKD should be able to make a buck from it. The use of "Nexus One" in this way should be a sign that "Bladerunner" has become an important cultural touchstone, at least in tech circles.

mreid··on My lazy American students
Yeah. I wish someone else would do that.
mreid··on An Old-School Board Game Goes Viral Among Silicon Valley's Techie Crowd
For kicks, play a game of Settlers with a New Zealander and ask them if they have wood for sheep. ;)
mreid··on Mathjax
I guess mobile Safari is not one of the 20+ browsers that are supported at present. Anyone else not seeing rendered math on their iPhone?
mreid··on Tuning Concurrent Clojure
The change in typing is not really much of a jump when you realise that Haskell's type inference means you don't have to declare too many type signatures up front. A lot of type signatures in Haskell code are not technically necessary and are usual there for documentation.

For example, this is perfectly valid Haskell code and doesn't mention a single type:

    factorial n = if n > 0 then n * factorial (n-1) else 1
The bigger conceptual shift will be from Python's mainly procedural style to Haskell's (or Clojure's) very strong emphasis on a functional style.
mreid··on Tuning Concurrent Clojure
Haskell is the obvious alternative. It is a very elegant functional language that is, in my opinion, executed well. Does it fit into your list of languages that miss the mark. If so, in what ways?

I also really like Clojure at the moment but have programmed in Haskell in the past and found there was much to recommend it. If you are genuinely looking for a good functional language you would be doing yourself a disservice if you didn't take a serious look at Haskell.

mreid··on What is math?
Great points, especially the one on multiple perspectives.

If you haven't read it already I highly recommend Thurston's essay "On proofs and progress in mathematics" - http://arxiv.org/abs/math/9404236 .

mreid··on [dead]
Yes. My tolerance is clearly not as high as yours. I made the jump several months ago. It's much nicer here. Let's hope it stays that way.
mreid··on Thank you, Rails
My guess is that it was a bit of self-deprecating humour.
mreid··on Why I'm Leaving Twitter and Returning to Blogging
So don't do it while you are working.
mreid··on Why I'm Leaving Twitter and Returning to Blogging
You're not the only one to have expressed this attitude but I find it strange. How long does it take you to write 140 characters or less? Also, unless people are actively following you you are not "spamming" them.

Your summary of what Twitter is good for is accurate as far as it goes but what you are missing is that, when they are timely and from interesting people, those small bits of information can be valuable.

I follow around 200 low-volume people on Twitter with similar interests to my own (machine learning, maths, stats, Clojure, ambient music). The number of times I have got useful advice and heads-up on recent articles, events and releases makes Twitter worthwhile the small attentional overhead I give it. On top of that, there is a nice sense of community that has built up over time.

Dismissing something based on its functionality alone without having used it means you won't experience the myriad ways people make use of that functionality. Emergent behaviour is powerful and difficult to predict.

"Computers? Sure, they're useful for quickly processing numbers represented in binary but it's not like they are designed for doing really complex things."

mreid··on Information Theory, Inference, and Learning Algorithms (free ebook edition)
I think to say the book has a heavy focus on frequentist statistics is a little misleading. Jaynes discusses a lot of frequentist methods but the emphasis is on doing so from a very Bayesian point of view.
mreid··on The Anatomy of Determination
I agree. I had a problem with this specific claim by PG: "Most people would agree it's more admirable to be good at math than memorizing long strings of digits, even though the latter depends more on natural ability."

Surely determination plus the right mnemonics can help anyone memorise long sequences of digits. I would have thought that reasoning creatively, carefully, and abstractly - i.e., mathematics - requires more natural talent, or at least a predisposition, than wanting to memorise digits.

mreid··on How I explained REST to my wife
You said: "REST offers conventions that make it easy to guess the URL of what you are looking for, just as you would guess where to click/navigate to get it."

In your earlier comment it does seem as though you are talking about guessing the URL.

What the other comment was alluding to with "Hypertext. As. The. Engine. Of. Application. State." (or HATEOAS, one of the main principles of REST) is that your application should never have to guess a URL. The possible URLs your application can next GET/POST/PUT/DELETE to should be given in the response to the last action it took.

There is nothing in the original description of REST that says anything about URLs being human readable. That's a pleasant side-effect of a sensible design but isn't the main aim.

As the original post describes, the importance of REST is to make simpler for machines (not humans) to manipulate services on the web.

You should probably read a bit more about the principles of REST before confusing people with your own interpretation of what you think it is.

edit: mistakenly thought HN understood markdown. edit: mistakenly thought HN understood HTML.

mreid··on Online (Machine) learning in Clojure
That's very neat. Would you mind posting a link to your version from the comments on my article? Thanks.

I learnt Haskell before I tried using Clojure and really like it as a language. It doesn't surprise me that it runs faster. I'm guessing the feature value look-ups don't involve the extra unboxing that Java's maps do.

mreid··on Online (Machine) learning in Clojure
I'm the author of the blog post and to answer your question, yes, I know about Incanter. In fact, I found out about Parallel Colt via Incanter. I've played around with it a little and it looks very cool.

I was initially going to build my algorithm using its libraries as a base but I thought a simpler first step would be to write it without pulling in too many extra dependencies.

← PreviousPage 4 of 4