jmount

62 ms·

751.
by jmount 11y ago
My take on this: http://www.win-vector.com/blog/2010/02/postels-law-not-sure-... Though I am beginning to think it coming down to "do you want to make it easy to run on a dev-box or easy to run in produc
752.
by jmount 11y ago
"Blocking noise" with more sound will make you deaf.
753.
by jmount 11y ago
Another possible reason would be: Macsyma was already in LISP.
754.
What is a good Sharpe ratio?(win-vector.com)
2 points by jmount 11y ago|0 comments
755.
Why does designing a simple A/B test seem so complicated?(blog.revolutionanalytics.com)
1 points by jmount 11y ago|0 comments
756.
Neural net image salad again (with code)(win-vector.com)
1 points by jmount 11y ago|0 comments
757.
by jmount 11y ago
Not normal, not effective, probably leads to paranoia (so watch your back).
758.
by jmount 11y ago
That is neat. Just for fun I tried to figure out how to set up Caffe on EC2. I got it to work, but I didn't get CUDA up and it probably would have been faster with Anaconda. But for what it is worth: here are my current notes https
759.
by jmount 11y ago
R has lazy evaluation (but only at function arguments). That makes it an easy place to play with some of the effects of lazy evaluation (but not get too bogged down with issues like IO).
760.
by jmount 11y ago
At best reschedules tech worker layoffs. Likely the bleed them out slower to avoid layoff reporting laws.
761.
Betting with their money(win-vector.com)
2 points by jmount 11y ago|0 comments
762.
by jmount 11y ago
"Using U+2019 is inconsistent with the rest of the standard" I agree with the article, just my negativity is such I would say the correct statement is more like: "Using U+2019 is inconsistent with good use, making it consist
763.
by jmount 11y ago
Cute. You probably can shorten the argument a bit by accepting Rice's theorem ( https://en.wikipedia.org/wiki/Rice's_theorem ) instead of explicitly building the Rice widget (which is pretty much how Rice&#x
764.
I do not believe Google invented the term “A/B test”(win-vector.com)
1 points by jmount 11y ago|1 comments
765.
Wanted: a perfect scatterplot (with marginals)(win-vector.com)
1 points by jmount 11y ago|0 comments
766.
by jmount 11y ago
I found the way to sell unit testing to the academics is the following. The unit test isn't a proof that the code works on all inputs, it is a demonstration that there exists at least one set of inputs the code works for. As we know:
767.
R in a 64 bit world (indexing large structures)(blog.revolutionanalytics.com)
1 points by jmount 11y ago|0 comments
768.
by jmount 11y ago
Your comment seem phrased to imply the current Ellie system (featured in the article) doesn't read sentiment (when you write "will have to be" about incorporating sentiment- that usually reads that an existing system like Ell
769.
by jmount 11y ago
I "did the /etc/hosts" thing for a long time. But it seemed like some things hung and tool longer sometimes. Overall /etc/hosts was a big improvement. Where are the good lists and what is the current thing y
770.
by jmount 11y ago
26 square miles? Seems way too big for a simulated population of 35,000. San Francisco is 47 square miles and has a population of around 800,000. edit- I take that back. Los Gatos is 11 square miles and about 30,000 people. So it is in
771.
by jmount 11y ago
The author is right- square error isn't completely natural. However one great property is that it gets expected values right (so using it helps you develop unbiased estimators). I have some notes on this here: http://www.w
772.
by jmount 11y ago
Béla Bollobás "The Art of Mathematics" has a very nice treatment of this problem (the history, multiple proofs) as problem number 33. And Aigner, Zeigler "Proofs from THE BOOK" also has a good treatment.
773.
by jmount 11y ago
http://www.juliandibbell.com/texts/bungle_vv.html
774.
by jmount 11y ago
Isn't this just what companies like to call "a voluntary layoff?"
775.
by jmount 11y ago
So factory votes $1 each time (small commitment, forcing some fraction of the down-winders in unless they can form a coalition).
776.
by jmount 11y ago
My point is: it is a bad situation, equilibrium or not. And some down-winders have to vote if the factory votes a little bit each time. Or even if the factory threatens to vote (and then doesn't). If all the magic is in the nominati
777.
by jmount 11y ago
With all due respect, it sounds like while it might fight flaws common to other voting systems, pay-back based voting will have its own gamesmanship. Consider 1,000 voters upwind of a factory and 1,000 more voters downwind. Suppose the fac
778.
by jmount 11y ago
Very glad to hear the idea has one fewer advocate than I though t (my 2010 rant on the issue: http://www.win-vector.com/blog/2010/02/postels-law-not-sure-... ).
779.
by jmount 12y ago
The post got truncated by a bad character in my blockquote paste (sorry!). It should be legible now (and include two figures).
780.
I still think you can manufacture a (slightly) unfair coin(win-vector.com)
3 points by jmount 12y ago|1 comments

More ›