Posts

[Link] Introducing OpenAI 2015-12-11T21:54:47.229Z
[LINK] Terry Pratchett is dead 2015-03-12T17:46:55.269Z

Comments

Comment by Baughn on Open thread, July 31 - August 6, 2017 · 2017-08-06T12:39:58.241Z · LW · GW

Dyson opens up another interesting question with this. Is it better to survive forever with a finite subjective time T, or it is better to consume 2*T experience in a finite amount of calendar time?

Isn't 2*T obviously better? Maybe I'm missing something here...

Comment by Baughn on Open thread, Sep. 12 - Sep. 18, 2016 · 2016-09-13T21:26:22.833Z · LW · GW

Data recovery is a last-ditch effort that often as enough fails, and if it succeeds will only get you back kilobytes or megabytes of your most critical material. (Unless you're lucky enough that it's actually a controller failure.)

If you want to avoid disk failures, invest instead in backups.

Comment by Baughn on Open Thread, Aug. 22 - 28, 2016 · 2016-08-28T14:08:01.077Z · LW · GW

When it comes to writing styles? Absolutely. There's a ton of skills involved, and deciding exactly which thoughts you want to convey is only a small part of it.

Comment by Baughn on Suggestion: jog instead of walk · 2016-07-17T03:00:08.036Z · LW · GW

If jogging isn't convenient, then you can still walk faster.

Whatever your 'natural' pace is, you can easily up it if you consciously try. I guarantee it. Downside is, once you get used to it you'll start to get annoyed at all the slowpokes who seem to think they'll live forever.

Comment by Baughn on Open Thread, Feb 8 - Feb 15, 2016 · 2016-02-08T11:19:26.305Z · LW · GW

The Clockwork Rocket

Comment by Baughn on February 2016 Media Thread · 2016-02-04T15:12:05.591Z · LW · GW

Garakowa: Restore the World

It's a movie about a suspiciously humanlike pair of anti-virus AIs, and their attempt to protect a backup archive of human history from destruction at the hands of sapient viruses. It's compressed a bit more than would be ideal, but otherwise well done.

Practically anything else I could say about it would be a spoiler, so I won't. What happened to humanity? Why are they there? Where do the viruses come from? You'll need to watch it to find out.

The story is also interesting due to the central Friendliness failure being novel, fairly plausible, and extremely disturbing.

Comment by Baughn on AI safety in the age of neural networks and Stanislaw Lem 1959 prediction · 2016-02-04T02:59:21.702Z · LW · GW

But here's the thing: most of the suggested ways to solve this problem (including the one I use) involve keeping the massive weak constraint relaxation, throwing away all irrelevant assumptions, and introducing new features to get the structured symbolic stuff. And that revision process generally leaves you with hybrid systems in which all the important stuff is NO LONGER particularly opaque. The weak constraint aspects can be done without forcing (too much) opaqueness into the system.

Where can I read about this?

Comment by Baughn on AI Fiction - Crystal Society · 2016-01-27T16:25:31.506Z · LW · GW

We already have a media thread. An average of several dozen media links are posted every month[0], and if they were all top-level posts the site would quickly become useless. While the story is interesting, I don't think it's interesting enough to overcome that.

[0]: Have not actually counted.

Comment by Baughn on Open thread, Jan. 18 - Jan. 24, 2016 · 2016-01-20T03:48:12.307Z · LW · GW

It's circular, and square.

That's literally all there is. I can't imagine it visually, the way I usually would. Wonder why. :P

Comment by Baughn on Open thread, Jan. 18 - Jan. 24, 2016 · 2016-01-19T18:06:36.477Z · LW · GW

I also asked for a square circle which got the immediate reply that it is logically impossible.

I am now imagining a square circle. That's interesting.

Comment by Baughn on Open Thread, January 4-10, 2016 · 2016-01-05T20:13:43.377Z · LW · GW

Google for 'crashplan data loss', and you'll find a few anecdotes. The plural of which isn't "data", but it's enough to ensure that I wouldn't use it for my own important data if I wasn't running two backup servers of my own for it. Even then, I'm also replicating with Unison to a ZFS filesystem that has auto-snapshots enabled. In fact, my Crashplan backups are on the same ZFS setup (two machines, two different countries), so I should be covered against corruption there as well.

Suffice to say, I've been burnt in the past. That seems to be the only way that anyone ever starts spending this much (that is, 'sufficient') effort on backups.

E.g. http://jeffreydonenfeld.com/blog/2011/12/crashplan-online-backup-lost-my-entire-backup-archive/


All of that said?

I'm paranoid. I wouldn't trust a single backup service, even if it had never had any problems; I'd be wondering what they were covering up, or if they were so small, they'd likely go away.

Crashplan is probably fine. Probably.

Comment by Baughn on Open Thread, January 4-10, 2016 · 2016-01-05T15:24:39.725Z · LW · GW

Use a backup system that automatically backs up your data, and then nags at you if the backup fails. Test to make sure that it works.

For people who don't want / can't run their own, I've found that Crashplan is a decent one. It's free, if you only back up to other computers you own (or other peoples' computers); in my case I've got one server in Norway and one in Ireland. There have, however, been some doubts about Crashplan's correctness in the past.

There are also about half a dozen other good ones.

Comment by Baughn on Open Thread, January 4-10, 2016 · 2016-01-05T14:13:04.527Z · LW · GW

I would be very interested in trying one of those. In particular, I frequently change up my writing style (deliberately), and it might be able to tell me what I'm not changing.

Comment by Baughn on January 2016 Media Thread · 2016-01-03T11:38:36.723Z · LW · GW

And frightening.

Comment by Baughn on has anyone actually gotten smarter, in the past few years, by studying rationality? · 2015-12-30T14:21:05.140Z · LW · GW

I feel like spending time learning "domain specific knowledge" is much more effective than "general rationality techniques".

However, reading Lesswrong is what prodded me towards getting better at spending my time effectively, really getting into a growth mindset. My only problem nowadays is that there are too many things I want to learn, and that's a much better problem to have; I know I can, I just have to pick and choose. I'm getting better at that, too.

Maybe the same would have happened anyway, but I don't think it would have happened quite as fast.

Comment by Baughn on Open thread, Nov. 16 - Nov. 22, 2015 · 2015-11-19T21:57:27.416Z · LW · GW

Ah, no. I do agree that uploading is probably the best path, but B doesn't follow from A.

Just because I think it's the best option, doesn't mean I think it'll be easy.

Comment by Baughn on Stupid questions thread, October 2015 · 2015-10-13T23:43:58.881Z · LW · GW

Maybe.

Determining which possibilities this is false for in our particular universe would take some time, and depends on the exact form of the laws of physics (which we don't know), so let's use a simplified example.

Take the Game of Life. While simple, it is in fact turing-complete; this was demonstrated by implementing a turing machine on it, which is the best way to demonstrate that sort of thing. (It's fun to put one cell out of place and watch it disintegrate.)

Take an infinitely large game of life. Start it in a random state, and leave it to evolve for an infinite amount of time. As you'd expect, a lot of things happen; in a universe that large, you will indeed see, for example, all possible simulations of Earth. So in that sense, "all things happen"...

But there are some states of the world which you will never see, no matter how long you wait. These are called Garden of Eden states).

There's a very good chance that there are also garden of eden states for this universe. They're likely to be states such as "The universe is tiled with a mandelbrot pattern of black holes"; states which are simply so unstable that they cannot naturally arise. There may also be less exotic states of that sort, but I feel less secure about claiming that...

And the Garden of Eden theorem, if it is applicable to our universe, states that it has Garden of Eden states of and only if time is non-reversible. As physics does indeed appear to be time-reversible, that's a bit of a problem. However, I don't know how applicable it is to our non-cellular physics.

Comment by Baughn on Mean of quantiles · 2015-09-11T13:07:44.259Z · LW · GW

(E.g., the trimean will have a better breakdown point but be less efficient than the mean; a worse breakdown point but more efficient than the median.)

What does "efficient" mean, in this context? Time to calculate would be my first guess, but the median should be faster to calculate than the trimean.

Comment by Baughn on Bragging thread September 2015 · 2015-09-08T11:29:22.542Z · LW · GW

Did my first serious bit of Minecraft modding, and learned how to use Blender in the process. It's not as impressive as the things I do at work, but it's fun.

Power poles

Comment by Baughn on Stupid Questions September 2015 · 2015-09-05T00:41:04.814Z · LW · GW

There's a good chance you'll have a second elephant failure while the first one is giving birth to a replacement, so at least use RAID6.

Or ZFS RAIDZ2. That's also great.

Comment by Baughn on Open Thread, Jul. 27 - Aug 02, 2015 · 2015-07-28T08:00:05.002Z · LW · GW

Where, exactly? All I've noticed is that there's less interesting material to read, and I don't know where to go for more.

Okay, SSC. That's about it.

Comment by Baughn on Open Thread, Jul. 13 - Jul. 19, 2015 · 2015-07-14T20:49:02.920Z · LW · GW

Nothing terrible was going to happen. As has been pointed out, collisions that energetic or more happen all the time in the upper atmosphere.

Comment by Baughn on The Waker - new mode of existence · 2015-07-13T10:13:11.617Z · LW · GW

But you're always stuck in one reality.

Let's take a step back, and ask ourselves what's really going on here. It's an interesting idea, for which I thank you; I might use it in a story. But...

By living your life in this way, you'd be divorcing yourself from reality. There is a real world, and if you're interacting solely with these artificial worlds you're not interacting with it. That's what sets off my "no way, no how" alert, in part because it seems remarkably dangerous; anything might happen, your computing infrastructure might get stolen from underneath you, and you wouldn't necessarily know.

Comment by Baughn on July 2015 Media Thread · 2015-07-04T14:39:04.319Z · LW · GW

Do you by any chance have those as MP3 or FLAC?

Comment by Baughn on A Roadmap: How to Survive the End of the Universe · 2015-07-04T14:12:31.094Z · LW · GW

"not impossible" == "possible". And this article doesn't show either one.

Comment by Baughn on Open Thread, Jun. 22 - Jun. 28, 2015 · 2015-06-24T17:06:01.149Z · LW · GW

So, some Inside View reasons to think this time might be different:

  • The results look better, and in particular, some of Google's projects are reproducing high-level quirks of the human visual cortex.

  • The methods can absorb far larger amounts of computing power. Previous approaches could not, which makes sense as we didn't have the computing power for them to absorb at the time, but the human brain does appear to be almost absurdly computation-heavy. Moore's Law is producing a difference in kind.

That said, I (and most AI researchers, I believe) would agree that deep recurrent networks are only part of the puzzle. The neat thing is, they do appear to be part of the puzzle, which is more than you could say about e.g. symbolic logic; human minds don't run on logic at all. We're making progress, and I wouldn't be surprised if deep learning is part of the first AGI.

Comment by Baughn on [Link]: The Unreasonable Effectiveness of Recurrent Neural Networks · 2015-06-05T08:59:01.002Z · LW · GW

If you're being generous, you might take the apparent wide applicability of simple techniques and moderate-to-massive computing power as a sign (given that it's the exact opposite of old-style approaches) that AGI might not be as hard as we think. It does match better with how brains work.

But this particular result is in no way a step towards AI, no. It's one guy playing around with well-known techniques, that are being used vastly more effectively with e.g. Google's image labelling. This article should only push your posteriors around if you were unaware of previous work.

Comment by Baughn on [Link]: The Unreasonable Effectiveness of Recurrent Neural Networks · 2015-06-05T08:56:09.053Z · LW · GW

The craziness it produced was not code, it merely looked like code. It's a neat example, but in that particular case not much better than an N-gram markov chain.

Comment by Baughn on June 2015 Media Thread · 2015-06-03T22:35:53.578Z · LW · GW

Log Horizon

If you liked the anime, you will likely find that this is better. If you felt that the anime was flawed, you may well find that the book is not, or not in the same way.

The story is slow, with a great deal of explanations and musings, especially in the beginning; it's trying to paint an entire world, and that shows. It is the sort of thing that is very difficult to adapt to an animated format. The book, however, was well worth the read.

Only the first volume is out yet, the second to come in July.

Comment by Baughn on June 2015 Media Thread · 2015-06-03T22:33:12.847Z · LW · GW

nostalgebraist has started work on a new novel, The Northern Caves. It's off to a slow start, but looks interesting so far.

Comment by Baughn on Devil's Offers · 2015-05-26T16:09:47.554Z · LW · GW

In addition to what James said, I'm reminded of the mechanism to change screen resolution in Windows XP: It automatically resets to its original resolution in X seconds, in case you can't see the screen. This is so people can't break their computers in one moment of weakness.

But you are absolutely allowed to break your computer in "one moment of weakness"; it isn't even hard. The reason for that dialog is because the computer honestly, genuinely can't predict if the new screen mode will work.

Comment by Baughn on Open Thread, May 25 - May 31, 2015 · 2015-05-25T14:32:27.925Z · LW · GW

I don't believe that it's mainstream transhumanist thought, in part because most people who'd call themselves transhumanists have not been exposed to the relevant arguments.

Does that help? No?

The problem with this vision of the future is that it's nearly basilisk-like in its horror. As you said, you had a panic attack; others will reject it out of pure denial that things can be this bad, or perform motivated cognition to find reasons why it won't actually happen. What I've never seen is a good rebuttal.

If it's any consolation, I don't think the possibility really makes things that much worse. It constrains FAI design a little more, perhaps, but the no-FAI futures already looked pretty bleak. A good FAI will avoid this scenario right along with all the ones we haven't thought of yet.

Comment by Baughn on Experience of typical mind fallacy. · 2015-04-27T20:18:47.316Z · LW · GW

I distinctly remember, at some point in my teens, realizing that other people sometimes thought like me and I could model their reactions as something more than inscrutable environmental hazards. So there's that.

Comment by Baughn on April 2015 Media Thread · 2015-04-05T20:06:19.613Z · LW · GW

On the flip side there's Luv and Hate, which is an (incomplete! still good) rewrite of the Muv-Luv Alternate story with a guest protagonist from... Supreme Commander. Including the ACU.

It's well-written, mainly character-focused with a few amusing combat interludes, and oh so gratifying after attempting to read the grimdark original.

It's also a quest. If this doesn't mean anything to you folks... don't worry about it, you can treat it as an ordinary story if you wish.

Comment by Baughn on Open thread, Apr. 01 - Apr. 05, 2015 · 2015-03-31T19:03:20.809Z · LW · GW

We won't run out of coal anytime soon. It has other issues, but I think that invalidates his conclusion—coal power plants are pretty cheap, and are already being built.

I'm also more optimistic about politicians. Ten years may be beyond their reelection horizon, but it's not beyond their "This place is going to hell"-horizon.

Comment by Baughn on Seeking Estimates for P(Hell) · 2015-03-22T20:37:44.238Z · LW · GW

Okay. I'm sure you've seen this question before, but I'm going to ask it anyway.

Given a choice between

  • A world with seven billion mildly happy people, or
  • A world with seven billion minus one really happy people, and one person who just got a papercut

Are you really going to choose the former? What's your reasoning?

Comment by Baughn on [LINK] Terry Pratchett is dead · 2015-03-13T01:23:08.967Z · LW · GW

If you have Alzheimer's, and you want to use cryonics, you should do your very best to get frozen well before you die of the disease.

This is problematic in all jurisdictions I can think of. Even where euthanasia is legal, I don't know of any cryonics organisations taking advantage, and there might be problems for them if they do. I'd very much like to be proven wrong in this.

Comment by Baughn on [LINK] Terry Pratchett is dead · 2015-03-13T00:35:30.226Z · LW · GW

It's a suspiciously pleasant way to go, but I see no reason to look more closely at this. Let's just be happy he got the end he wanted.

Comment by Baughn on March 2015 Media Thread · 2015-03-05T18:51:51.908Z · LW · GW

Saenai Heroine no Sodatekata.

It's an anime about... making a game... that appears fully congruent with the contents of the anime...

In short, it seems to be a metacircular anime. It's worth watching because of the way it plays with tropes, and the origin of those tropes; it's marginally annoying in that many of the tropes it plays with are of the harem genre. There may be something more going on in the background, but I haven't watched enough to tell. It may be especially interesting to people who have long experience with japanese animation.

The first episode is fully representative, so I'd recommend having a look if the above appeals.

Comment by Baughn on March 2015 Media Thread · 2015-03-05T18:46:00.963Z · LW · GW

The only other one that springs to mind is the one with the Nine-Brained Kyuubi.

Got any more?

Comment by Baughn on Harry Potter and the Methods of Rationality discussion thread, March 2015, chapter 116 · 2015-03-04T23:50:52.895Z · LW · GW

And Quirrel was Ravenclaw.

Comment by Baughn on Harry Potter and the Methods of Rationality discussion thread, March 2015, chapter 116 · 2015-03-04T23:48:44.174Z · LW · GW

That's a literal Idiot Ball reference, not to mention canon. I don't think we'll see it.

Comment by Baughn on Best of Rationality Quotes, 2014 Edition · 2015-03-03T19:40:46.549Z · LW · GW

That worked. Running "strfile rationality" on my system produces a different rationality.dat than the one you have checked into your repository, but "fortune rationality" works now.

(It doesn't need to be copied to a system path.)

Comment by Baughn on Best of Rationality Quotes, 2014 Edition · 2015-03-02T18:37:03.863Z · LW · GW

How do I use it? Fortune is being obstructive.

Comment by Baughn on Imagining Scarcity · 2015-03-02T14:04:30.148Z · LW · GW

Read some blogs. I'd like to read the good blogs, but I have no idea where to find them; any pointers?

Comment by Baughn on Harry Potter and the Methods of Rationality discussion thread, February 2015, chapter 111 · 2015-02-25T18:59:07.641Z · LW · GW

I wonder if that was all acting. If it wasn't, then Hermione may now be very hard to kill indeed.

This can only be a good thing.

Comment by Baughn on Harry Potter and the Methods of Rationality discussion thread, February 2015, chapter 108 · 2015-02-22T18:51:01.011Z · LW · GW

HE IS HERE. THE ONE WHO WILL DESTROY THE VERY STARS IN HEAVEN.

I've been wondering about this. Voldemort knows this prophecy, and I expect he realises it's not about him. So, how can he expect to be able to kill Harry, in a self-consistent universe?

Perhaps Harry's reading of that plan is all wrong.

Comment by Baughn on Harry Potter and the Methods of Rationality discussion thread, February 2015, chapter 108 · 2015-02-22T18:39:30.244Z · LW · GW

Moreover, it ignores Merlin's interdict. Those children had not learned how to disillusion themselves.

I think you're on to something here.

Comment by Baughn on Harry Potter and the Methods of Rationality discussion thread, February 2015, chapter 104 · 2015-02-16T13:04:55.372Z · LW · GW

With professional players, whose job it was to play Quidditch.

It won't survive children doing the same. More to the point, the snakes and ravens are deliberately and obviously exploiting the current rules, which will trigger all sorts of fair play instincts.

Comment by Baughn on I notice that I am confused about Identity and Resurrection · 2015-02-13T16:33:13.426Z · LW · GW

A rather large fraction of my discussions happen via IRC; I log every bit of it, and carefully back the logs up.

Occasionally, I go back and read some random fraction of the logs. It is usually a valuable experience. I am doing so right now, albeit without IRC.