We Must Defend Science in the Face of Political Attacks • Hacker News Board

Re: We Must Defend Science in the Face of Political Attacks

#131

Post by kaitai » Fri, May 17, 2019, 3:50 AM UTC

mises wrote: post #72

kaitai wrote: post #68

Earlier quoted context omitted.

"Biased" still has real statistical meaning. It is unfortunate that people treat these problems with an aim to justifying their virtue rather than creating a product that works (not saying you are doing this, just commenting that often the reaction is about personal virtue rather than actual technical accuracy). It is not more difficult to make a facial-recognition algorithm work on black people -- or Asian people, t…

It might actually be more difficult. There have been cases where soap dispensers didn't work on blacks because more infrared was absorbed. Not saying it's not solvable, but it's possibly harder. I guess you're right that you could make different models and use the correct one based on race. Correct me if I'm wrong, I'm no expert, but could this be solved by adding an extra layer to a neutral network?

For facial recognition I believe it's possible to use pre-trained networks and add extra layers for specific (visual) demographics. You're correct that for the engineering of soap dispensers an actually different method might be preferable. That however boosts a previous point; if the engineers behind the soap dispenser had a number of black members, would they have designed something that didn't work for them, or discarded it in the design stage?

Re: We Must Defend Science in the Face of Political Attacks

#132

Post by dang » Fri, May 17, 2019, 5:24 AM UTC

instantwhat wrote: post #130

dang wrote: post #128

Earlier quoted context omitted.

Would it were so easy! It's more like this. You get dozens of bug reports a day. Some report that XYZ are bugs while others report that no, those are features and you'd better add more, but ABC are bugs and you'd better fix those. If you change ABC or XYZ, there is scandal and outrage! Many new reports are filed. The bug reports come with cartoons showing you secretly engineering the bugs owing to your thoughts and f…

> What should we do now? :) I don't know what the solution is. I wish I did. My point is simply that there is definitely a bias on HN, and I think few people would seriously claim that it's right-leaning. A multi-axis perspective would probably be more useful: one could say that the pro-business/pro-startup bias is right-leaning, but one could also say that it's libertarian, and libertarianism isn't necessarily on th…

I'd like to see links to thoughtful, reasonable comments that have been flagged or downvoted into obscurity. As many such links as you (or anyone) can possibly come up with, I'd like to look at them. Best to email hn@ycombinator.com because this thread is sinking low in the feed now.

Let me ask you a question, though you don't have to answer it. Do you feel like HN is biased against the point of view you yourself favor? Because, in my experience, that's what everyone who feels that HN is biased actually feels. This is so reliable a phenomenon that I can predict someone's political orientation just from what they say about HN. I don't believe I've seen a single exception. That's shocking, and I'm pretty sure it's true.

I'm not saying it follows that things aren't biased—that would be a non sequitur. But I am saying that people's perceptions on this are unreliable across the political spectrum, and for the same reason: it's humans looking at it, and we all have the same wiring. Whatever that wiring is, it cuts much deeper than left vs. right divisions, because the behaviors are identical across that divide.

Re: We Must Defend Science in the Face of Political Attacks

#133

Post by 75dvtwin » Sat, May 18, 2019, 4:26 AM UTC

dang wrote: post #132

instantwhat wrote: post #130

Earlier quoted context omitted.

> What should we do now? :) I don't know what the solution is. I wish I did. My point is simply that there is definitely a bias on HN, and I think few people would seriously claim that it's right-leaning. A multi-axis perspective would probably be more useful: one could say that the pro-business/pro-startup bias is right-leaning, but one could also say that it's libertarian, and libertarianism isn't necessarily on th…

I'd like to see links to thoughtful, reasonable comments that have been flagged or downvoted into obscurity. As many such links as you (or anyone) can possibly come up with, I'd like to look at them. Best to email hn@ycombinator.com because this thread is sinking low in the feed now. Let me ask you a question, though you don't have to answer it. Do you feel like HN is biased against the point of view you yourself fav…

I think there are 4 categories of topics

1) ideological (including political (globa/local), or religious

2) technical

3) personal (meaning focus is, specifically, on subjective position)

4) moral (meaning, focus is on subjective, but relatable at a group/society/era level)

Problem is with ( 1 ), and may be ( 4 ).

My personal view, is that if HN is to allow new submissions where (1) and (4) are present. Those submissions should go through a different flow, than (2) and (3).

These submissions should be marked as such, then reviewed by a moderator, to see if it fits the broader category of HN-allowed submissions.

And then, if the submission passes moderation, the discussion on those posts cannot follow same upvote/downvote model as ( 2 ) or (3). Because regular upvote/downvote model puts at disadvantage minority view.

-- --

WRT your question > Let me ask you a question, though you don't have to answer it. Do you feel like HN is biased against the point of view you yourself favor

Yes, I think selection of submissions, downvotes and flags favors disproportionally to anti-Trump, anti-Brexit, anti-Kavanaugh, anti-male rhetoric.

I particular share a view of environment issues, though I favor non-global, trade-based solutions. So I, personally do remember incidents where my submissions or comments were downvoted on that topic.

Interestingly, on more political conservative forums, that I participate on -- when I argue a pro-environment position -- I get some pushback, sometime ridicule -- but my comments never 'grayed out' or flagged.

-- Sorry for a long, response. But yes. I favor particular ideology and moral stance. Everybody has those positions.

I also appreciate, particularly, flexibility of mind and personal ability to adjust to something different if circumstances or new information dictates that.

I have yet to find any discussion that would cause me a shift in my positions on (1) or (4). I suspect this is same with other people, but being tested by questions, or telling others what I think -- is engaging, nevertheless.

Another thing, I noticed, that I tend to use 'reciprocal' tone. So the actual tone would depend on some selection/aggregation of other folks tone.

My overall complaint, if I may call it that, is flagging/graying out expressions that are conservative-leaning.

Re: We Must Defend Science in the Face of Political Attacks

#134

Post by instantwhat » Sat, May 25, 2019, 5:34 AM UTC

dang wrote: post #132

instantwhat wrote: post #130

Earlier quoted context omitted.

> What should we do now? :) I don't know what the solution is. I wish I did. My point is simply that there is definitely a bias on HN, and I think few people would seriously claim that it's right-leaning. A multi-axis perspective would probably be more useful: one could say that the pro-business/pro-startup bias is right-leaning, but one could also say that it's libertarian, and libertarianism isn't necessarily on th…

I'd like to see links to thoughtful, reasonable comments that have been flagged or downvoted into obscurity. As many such links as you (or anyone) can possibly come up with, I'd like to look at them. Best to email hn@ycombinator.com because this thread is sinking low in the feed now. Let me ask you a question, though you don't have to answer it. Do you feel like HN is biased against the point of view you yourself fav…

I appreciate your response. I don't expect you'll see this reply, since it's so old, but just in case:

> Let me ask you a question, though you don't have to answer it. Do you feel like HN is biased against the point of view you yourself favor?

Generally, yes, of course.

> Because, in my experience, that's what everyone who feels that HN is biased actually feels. This is so reliable a phenomenon that I can predict someone's political orientation just from what they say about HN. I don't believe I've seen a single exception. That's shocking, and I'm pretty sure it's true.

I'm guessing this isn't what you mean by that, but I'd like to point out a way to interpret it: If HN is left-leaning, and left-leaning people do not think HN is biased, then only right-leaning people would think HN is biased, and one could predict political orientation based on whether a person thinks HN is biased.

In other words, if left-leaning people think their position is neutral (e.g. because they're irreligious, and they think that religion is biased and that secularism is unbiased), and HN is left-leaning, then left-leaning people would think HN is unbiased, and right-leaning people would think it is biased.

Now I'm guessing you meant that both left- and right-leaning people think HN is biased against their political orientation.

But I challenge you (not that you haven't done this, but still) to look at submissions on polarizing issues (e.g. global warming, sexism, racism), look at the comments that are downvoted and flagged, and look at which viewpoints those comments espouse. Modulo comments that are flagged or downvoted for obvious guidelines violations, my claim is that the vast majority are ones that are right-leaning. I don't think confirmation bias is causing me to think this--I think it's a fair observation.

And if that observation is correct, then I think it's fair to conclude that HN is biased to the left.