HN: The Good Parts (2016) • Hacker News Board

Re: HN: The Good Parts (2016)

#51

Post by intergalplan » Wed, May 05, 2021, 3:25 PM UTC

PaulHoule wrote: post #32

suprfsat wrote: post #27

Earlier quoted context omitted.

Color ratio of a typical comment on this web site: https://contrast-ratio.com/#%23dddddd-on-%23f6f6ef 1.25, which is well below the minimum of 4.5

They ought to have a button to put on a high-contrast stylesheet, in fact I bet some hacker has already done it.

I'd assumed anyone with visual impairment—but not outright blindness requiring a screen-reader—was using custom stylesheet stuff pretty much constantly all over the web. Controlling the appearance of sites client-site used to be a basic browser feature. I know that's receded into the background, but is something similar (either some accessibility feature, or a plugin) not the norm for those who need them? If not, that sucks, because control used to be so firmly on the client-side for those things and that was a good thing.

Re: HN: The Good Parts (2016)

#52

Post by nickthemagicman » Wed, May 05, 2021, 3:27 PM UTC

beej71 wrote: post #49

nickthemagicman wrote: post #6

"HN comments are horrible." Proceeds to list pages of awesome comments. Claims topics he's informed about have mostly wrong comments. Doesn't give any examples. HN comments are a conversation/discussion not a contest of who is right/wrong, and it's great to have differing viewpoints as long as those views don't degrade into logical fallacies like ad hominem, ,anecdotal, or others. There are way less logical fallacies…

I think there's a blend of good and bad. If you ever see a HN front page story where you have inside information about some event, you can rest assured there will be maddening amounts of completely wrong speculation written as if it were gospel by people with no connection to that event. And then people attack the strawmen with abandon. And for professional reasons, you grit your teeth, keep your mouth shut, and let…

I guess I've never had inside information to see this firsthand.

Or maybe my internet spidey sense is so attuned that I've learned to subconsciously tune out B.S.

So I've never really seen that type of thing happening.

Does anyone really believe unverified information people post on the internet ?

Re: HN: The Good Parts (2016)

#53

Post by bombcar » Wed, May 05, 2021, 3:29 PM UTC

bryanrasmussen wrote: post #26

sokoloff wrote: post #5

I don’t think the informed technical opinions are nearly as rare as danluu characterizes them to be, to wit: “it’s rare that a thread will have even a single comment that's well-informed”. I do agree that the farther you get from computer tech, the more frequently you’ll see patently wrong comments (aviation is my eye-roll inducing topic here), but on computer tech topics, I’d say more than 50% of substantive comment…

I do wonder if the definition of well-informed includes agrees with my viewpoint somewhere in it.

Look at the examples he picks - I take "well informed" to mean not only knowledgable about the subject matter but also providing an insight beyond a platitude.

Re: HN: The Good Parts (2016)

#54

Post by intergalplan » Wed, May 05, 2021, 3:34 PM UTC

anotha1 wrote: post #38

PaulHoule wrote: post #36

Earlier quoted context omitted.

I know teens who do it all on Tik-Tok. Marshall McLuhan was right to point out that "television" would erode the need for people to read and write.

That reminds me of people who "learn it all on YouTube." Maybe it's the best option for some people. Not for me, but I do love an audible book. In conclusion, I think these other mediums simply lack the iterations that we've put into text over the centuries, leaving text as the clear winner in my book.

YouTube is great for things where the directions for the specific activity you want might not include "basic" things someone who knows it well wouldn't think to list in written instructions, or some little technique or movement they do automatically and don't even think about. I love it for house and car DIY stuff, for that reason and also because it makes it clear which parts will be a pain-in-the-ass, and how much, to help me gauge whether I want to do it, or hire it done.

It's also great for technique with tools in the kitchen—not learning recipes so much, but learning the right way to process a certain vegetable really fast, or exactly what a proper whipping motion for eggs ought to look like, or how a pro moves the pan on and off the flame while cooking this dish, exactly what it looks like when they deem a certain cooking technique "done", all of that.

Re: HN: The Good Parts (2016)

#55

Post by the_local_host » Wed, May 05, 2021, 3:39 PM UTC

creata wrote: post #14

treebog wrote: post #9

I sometimes wonder how HN’s audience is shaped by the site’s deliberate inaccessibility to visually impaired people. I suspect it is one part of the site’s tilt toward the younger crowd

> the site’s deliberate inaccessibility to visually impaired people Can you expand? I naively thought that a site with markup as simple as HN's (zooming is flawless and it even works without JavaScript) would be very accessible to visually-impaired people.

I'm not the parent poster, but I hear where their coming from.

I'm not visually impaired, just in need of bifocals; I zoom HN to 150% and it works fine.

But I am kinda surprised when I turn the zoom off. Even 25 years ago that text would have looked small to me.

Re: HN: The Good Parts (2016)

#56

Post by flakiness » Wed, May 05, 2021, 3:49 PM UTC

OT: I hope Dan Luu write his blog more often. There is no post this year and he seems to be consumed by Twitter. His tweets are witty and I like it, but I miss his long form, super deep dive kind of writings. These are marvels. I also like the witty ones like this, and I would love to have it as blog rather than as tweets.

Totally off-topic: One of my favorite from his writing is "Sampling v. tracing" [1].

[1] https://danluu.com/perf-tracing/

Re: HN: The Good Parts (2016)

#57

Post by dashwav » Wed, May 05, 2021, 3:55 PM UTC

PaulHoule wrote: post #34

lewisflude wrote: post #19

Earlier quoted context omitted.

I mean, the site scales really well with browser size, default font size etc. I'd argue the real inaccessibility is lack of images, and general "boring" vibe that puts off entire demographics who post the type of content that might not be welcome here.

I think a large number of people are unaware that you can scale web pages up with Ctrl -, Ctrl +, particularly older people who complain that "text is too small".

I'm not even old (mid-20s), but I scale the site to 150%. HN is honestly one of the worst sites I have used with regards to default scaling across resolutions.

OTOH the simplicity of the site allows for very easy browser-based scaling, which means I only really think about how bad the text size is once per machine.

Re: HN: The Good Parts (2016)

#58

Post by claudiawerner » Wed, May 05, 2021, 3:56 PM UTC

platz wrote: post #40

PaulHoule wrote: post #29

[flagged]

Wow so you are basically linking people to depictions of teen and child pornography.

That stuff isn't visible on Danbooru unless you buy the 'gold account', which last I checked is a one-time $20 purchase. Browsing the pages will often have a notice at the bottom along the lines of "2 posts on this page require a gold account to view." - those are typically the 'contentious content'.

Re: HN: The Good Parts (2016)

#59

Post by ChrisMarshallNY » Wed, May 05, 2021, 4:10 PM UTC

dkarras wrote: post #46

ChrisMarshallNY wrote: post #7

I think we've seen this before. Comments, everywhere are terrible. Heck, this one probably is. But I like HN more than anywhere else (so far). Maybe rot will set in, and the sun will set on our relationship, but not yet. I find many comments to be quite correct. Not all, of course, but many. The ones that I like, are the ones that make me think "orthogonally."

Yeah I come here to read opinions of intelligent / informed people who disagree with me. I even subtly and respectfully try to rile them up sometimes to see what they have to offer.

I have found that the best way to get correct information is to confidently state incorrect information.

Re: HN: The Good Parts (2016)

#60

Post by PaulHoule » Wed, May 05, 2021, 4:10 PM UTC

lifeisstillgood wrote: post #48

PaulHoule wrote: post #24

Ranking 'quality' is tough because it has multiple dimensions and aggregating multiple kinds of utility is problematic, see https://en.wikipedia.org/wiki/Arrow%27s_impossibility_theore... developing a good ranking function for bag-of-words full text search is about carefully balancing the attraction of larger vs smaller documents in the collection. If you try to aggregate several bad search engines you get a bad sear…

Are you saying you do know the way to create a good ranking function? Does Arrows theroum stop that?

I know how to make some ranking functions that are good in some ways. They are not good in other ways so I don't break Arrow.

The shadow of Arrow however is that it is a lot of work to get a real improvement over a tfidf with random characteristics and you are limited by an asymptote which is less far away than you'd like: the rel teams at Google and Bing cannot beat the p=0.7@1 barrier for text search because 0.3 of the time they will get your intent wrong.

There are ranking functions in Lucene that can be tuned up as I described by making a test data set and using the methods developed at the TREC conference. That kind of tuning really works. At that conference you might see people fight over a point or two of AUC and I don't know if users can feel that but I am sure users can feel the 15 point and more difference we were seeing from tuning.

The odd thing is that hardly anybody does it and the knowledge seems pretty obscure. (e.g. I have talked to people at Lucene, OpenText and other full text search vendors and they are much more impressed with having 10,000 connectors than with the search results being good.)