Why HN Should Use Randomized Algorithms
danluu.com
Why HN Should Use Randomized Algorithms
1–10 of 41 posts
Re: Why HN Should Use Randomized Algorithms
#2I think it's a great idea that solves an actual problem and it would probably help smooth out voting rings that slip past the detector.
However, that would probably mean HN couldn't optimize or cache the front page. On the other hand, this is a feature for logged-in users only, so it might be feasible.
Re: Why HN Should Use Randomized Algorithms
#3Great idea. Would be interesting to see it trialed.
Re: Why HN Should Use Randomized Algorithms
#4I think it's a great idea that solves an actual problem and it would probably help smooth out voting rings that slip past the detector. However, that would probably mean HN couldn't optimize or cache the front page. On the other hand, this is a feature for logged-in users only, so it might be feasible.
You could have two versions of the site, one with random noise, one without. Or maybe even specify the noise.
Re: Why HN Should Use Randomized Algorithms
#5I think it's a great idea that solves an actual problem and it would probably help smooth out voting rings that slip past the detector. However, that would probably mean HN couldn't optimize or cache the front page. On the other hand, this is a feature for logged-in users only, so it might be feasible.
You could have two versions of the site, one with random noise, one without. Or maybe even specify the noise.
That's what I meant when I said it was for logged-in users only.
Re: Why HN Should Use Randomized Algorithms
#6It is unfortunate that you have to read the footnotes to understand the article - on a first read of the text I was completely lost.
This is, though, a good idea - at the moment, many interesting articles are lost because not enough people visit the new page.
Re: Why HN Should Use Randomized Algorithms
#7It is unfortunate that you have to read the footnotes to understand the article - on a first read of the text I was completely lost. This is, though, a good idea - at the moment, many interesting articles are lost because not enough people visit the new page.
> It is unfortunate that you have to read the footnotes
> to understand the article - on a first read of the text
> I was completely lost.
> ... at the moment, many interesting articles are lost
> because not enough people visit the new page.
Re: Why HN Should Use Randomized Algorithms
#8It is unfortunate that you have to read the footnotes to understand the article - on a first read of the text I was completely lost. This is, though, a good idea - at the moment, many interesting articles are lost because not enough people visit the new page.
> It is unfortunate that you have to read the footnotes > to understand the article - on a first read of the text > I was completely lost. I'm not the author, but I'm intensely interested in effective communication. You've taken the trouble to comment, and I was wondering if you'd be willing to walk through the item and explain where you got to in the main body that you thought was unclear, and where you were lost. I…
The first three are in English, sure, I'm talking about the last two paragraphs. I understood the problem, almost from the title - I was interested mostly in the proposed solution.
In particular, the last 2-3 sentences where a complex procedure is defined in a few inline code snippets - and then the author continues as if the result is obvious.
It reminds me very much of when I read overly mathematical scientific publications (which I do a lot, being a PhD student.) In such papers, the authors often expect you to stop and spend 5 minutes reading a few symbols (perhaps reading a couple of Wikipedia articles along the way) before continuing. This makes the articles highly unreadable - I sometimes spend 2-3 days slowing working through the particularly poor ones. (Obviously, I only do this for highly relevant/useful papers!)
This article isn't in the same level of awful, obviously, but it did give me the same "I'm lost now, I should go back" feeling - which I thought was worth feeding back to the author, especially since they had provided the explanation, but they'd hidden it in a footnote.
Re: Why HN Should Use Randomized Algorithms
#9...or we could have a much larger home page, so we don't have to deal with the end-of-page-1 discontinuity, and we rely on the end-of-reader's-attention-span discontinuity, which is much smoother.
Re: Why HN Should Use Randomized Algorithms
#10It's a good idea, though I wonder if there is a better way to do this without randomness.[1]
I mean it's just an exploitation vs exploration problem. You want to maximize the number of articles the average user will upvote (essentially avoiding wasting our time.) You want to use the time and number of votes to predict the probability that a user will upvote it. But you also want to do "experiments" to find even better articles even if it means wasting a few people's time.
I'm certain there is an elegant math equation that does this perfectly but I can't figure it out.