I've created a new "landing page" on Friendly AI at Friendly-AI.com. This is similar to IntelligenceExplosion.com, Existential-Risk.org, Anthropic-Principle.com, and simulation-argument.com.
The site is less ambitious than the original plan for it was, but it serves its purpose.
Design courtesy of Lightwave.
Looks great!
I may be alone in this, and I haven't mentioned this before because it's a bit of a delicate subject. I assume we all agree that first impressions matter a great deal, and that appearances play a large role in that. I think that, how to say this, ehm, it would, perhaps, be in the best interest of all of us, if you could use photos that don't make the AI thinkers give off this serial killer vibe.
Here's my $0.02 on that page: Bostrom looks like he suspects you of something. Goetzel looks smug but ok. Hanson looks evil. Shulman looks fine, but getting rid of his red eye would take about 2 minutes in iPhoto. Chalmers looks like a hobo, but not scary. Hall, Salamon, and Yudkowsky look fine.
Eleizer's eyes seem oddly asymmetrical, which unsettles me.
Chalmers looks like that in person, its not the photo/
Hanson looks evil.
And I for one hope that picture never gets replaced with something more bland.
(Responding to an old comment but ...)
Wow. Judging by these pictures, and these pictures alone ...
Nick Bostrom is serious business.
Ben Goertzel is leather with a side of JPEG compression.
Robin Hanson is plotting something fiendish.
(And the backdrop for Robin's picture says "school pictures day".)
Carl Shulman isn't working; he's at a party with girls.
David Chalmers is on his yacht, the wind romantically blowing his hair.
(And you could be there at his side ...)
J. Storrs Hall is a math professor.
Anna Salamon is a perpetual student.
Eliezer is trying to extrapolate your utility function from a careful examination of your microexpressions.
Yeah, Anna's happy smile and Eliezer's piercing gaze sure make you think about the number of skeletons in their SIAI closet.
Anna looks like an ad for salad and Eliezer looks like a golden retriever puppy. However, Robin looks like he wants to negotiate about who should wear your skin.
(I think the "maintained by" notice at the bottom should have a definite article, as recently discussed.)
Agreed. Also in the FAQ, e.g. on this line:
Independent researchers associated with Singularity Institute: Daniel Dewey, Kaj Sotala, Peter de Blanc, Joshua Fox, Steve Rayhawk, and others
(I'm also flattered to see that I'm called a "researcher". I don't exactly feel like one - I just feel like someone who's scribbled together a couple of simple papers in his spare time. Of course, this doesn't imply that I'd object to being listed on the page.)
Typo in the "keywords" meta tag: "Friednly AI" - swap d and n.
What was I doing in the HTML source, you may ask? I was wondering if you'd used some sort of tool or service like unbounce to create the page, which looks a few notches better than the other pages you mention.
OK maybe this isn't important, but I'm confused by the "media" pages on friendly-ai.com and intelligenceexplosion.com. They both say "about the intelligence explosion", and they both list almost the same set of links.
On the topic of how the site looks in different browsers, and finding out whether the layout is borked on some browsers, you could use http://browsershots.org/.
At the moment though, it fails due to an internal server error when it tries to fetch http://friendly-ai.com/robots.txt. If you fix that, you should be able to easily see how the site looks in a bunch of different browsers on different operating systems.
In FAQ #6:
Friendly AI is a problem of cognitive science.
I think this un-argued-for assertion makes the site seem (and be) less rigorous. Unfortunately, I can't think of a better concise justification for why FAI researchers should read about cognitive science.
on linux, latest version as of a few days ago. May be my fonts?
That sort of thing happens when the element that is colored isn't constrained to the text element. Somewhere in there is an assumption about font size or dpi or something which is bound to be broken by some clients.
EDIT: it's not my fault. Here's your problem:
#link_boxes {
margin: 30px -40px -15px -40px;
padding: 0 40px;
height: 180px;
background: #f6f6f6;
}
the height: 180px is the problem assumption. I think If you just removed it, it should work in all cases. If I figure out how to make opera hack css teh way it can hack the html, I'll test it.
EDIT2: that doesn't fix it, also have to change this part:
#link_boxes a.box,
#link_boxes a.box:link {
width: 160px;
height: 150px;
float: left;
display: block;
background: #f6f6f6;
border: none;
text-align: center;
padding: 15px 11px;
}
pixel measurements all over the place. I hate web designers.
EDIT3: for a quick hack that should make it a bit more flexible, replace px measurements with something approximately the same in em measurements. I think em is properly tied to dpi and font size.
EDIT4: FAQ page is broken too. The 'Questions' part of the 'frequently asked questions' header word-wraps onto the table of contents. interestingly, it doesn't when I give the browser less room. The font size increments when I give it enough room to draw those damn margins, but again, the assumptions of DPI and font size don't let it fit properly. In this case the whole fixed content width idea should be thrown out. Let it fill the window. The whole point of HTML is that it reflows with different browser windows. At least there's no horizontal scroll.
I'll stop trying to break the site now. Sorry about all the hate.
I also see the FAQ page as broken with 'Questions' in the header appearing overlayed on the #2 and #3 items in the 'contents' list. With Firefox 8 on Linux at default zoom, and zooming down to make the fonts smaller than normal does fix it.
I agree with nyan_sandwich that things would be much improved if the CSS used ems instead of pixels, which are guaranteed to break if users have non-standard fonts or font sizes or their browser happens to have different enough default CSS rules.
How is it for you now? If you have any problems, can you please make a screenshot (and upload to http://imgur.com or somewhere else) to help us diagnose the problem?
It's much better now. The only issue remaining is that the 'Frequently Asked Questions" is just a tiny bit too wide to fit on one line inside the containing box, so the 'ns' of 'Questions' sticks outside of the gray box it is supposed to be inside.
I don't know. My CSS hacks are limited. If I had control over the server I could do it easy, but I don't know how to make opera modify the CSS.
EDIT: if you experiment a bit and get it right on your browsers, it should work as well on mine and everyone else's. That's the point of using em instead of px.
I can test it and let you know if you do that.
bah, still trouble.
"explosion" in "external links about the intelligence explosion" is still off the bottom. the FAQ no longer wordwraps all over the place, but goes off the side of the margin instead.
I don't know. Good luck.
Hi, can you make a screenshot of the "explosion" part (you can upload it to http://imgur.com), please, to help us diagnose the problem.
Also, is the "links and media" which you originally pointed out still a problem?
I found your answer to the sixth question of the FAQ especially helpful. It's nice to have a reference of great books and papers on essential topics. Upvoted due to that.
On that note, does anyone have a PDF or ebook of Rational Choice in an Uncertain World? If so, I'd very much appreciate it if you could share it with me. I've searched for it, can't find a copy online.
If you can share, or if you know of a more proper place to ask this question, please let me know. Thanks!
Nice!
From FAQ #5:
Independent researchers associated with the Singularity Institute: Daniel Dewey, Kaj Sotala, Peter de Blanc, Joshua Fox, Steve Rayhawk, and others
Would it be feasible to make this list exhaustive so that you can delete the "and others"? I think the "and others" makes the site seem less prestigious.
I also see the FAQ page as broken with 'Questions' in the header appearing overlayed on the #2 and #3 items in the 'contents' list. With Firefox 8 on Linux at default zoom, and zooming down to make the fonts smaller than normal does fix it.
I agree with nyan_sandwich that things would be much improved if the CSS used ems instead of pixels, which are guaranteed to break if users have non-standard fonts or font sizes or their browser happens to have different enough default CSS rules.
How is it for you now? If you have any problems, can you please make a screenshot (and upload to http://imgur.com or somewhere else) to help us diagnose the problem?