Recently reporters from two major national magazines contacted me in preparation for doing stories on Bitcoin. This reminded me that Wired magazine did a cover story on the Cypherpunks in its second issue. I think the LessWrong community is already larger and more active than Cypherpunks were back then, and potentially more influential, but there hasn't been much publicity on us. I'm tempted to suggest doing a story on LessWrong to one of the reporters. Is this a good idea, or bad?
More generally, do we want more publicity, and if so what's the best way to go about getting it?
ETA: Would it be bad etiquette to reveal the names of these magazines at this point, or even to say as much as I've said?
I don't really think we want major publicity. A community can only absorb so many people at a time, especially when it's a huge vast sprawling mess of material like LW is. Ideally what we would get is a steady stream of low-level publicity in our natural recruiting grounds like Wired - but not headliners.
(One of the worst things to ever happen to Wikipedia, in my belief, was getting headline publicity - as part of the Seigenthaler affair.)
Any suggestions how to get that ideal "steady stream of low-level publicity"? And what do you think about my idea of suggesting a story idea to one of the reporters (I think he tends to write multi-page feature stories)? If he actually took up the suggestion (which I admit, given what Vladimir_M said, is a big if), would that be an overall positive or negative?
(Since nobody said it would be bad etiquette to reveal the names of the magazines, I'll mention that they are Wired and New Yorker, and I was thinking of making the suggestion to the Wired writer. If anyone thinks this actually is bad etiquette, please let me know and I'll try to quickly remove this part of the comment.)
Did that incident cause a large influx of newcomers to Wikipedia? The article you link to only mentions that Jimbo Wales banned anonymous page creation during that incident, which as I understand was not to handle the newbies drawn by publicity but due to what happened 4 months earlier, when someone anonymously created the fake Seigenthaler biography on Wikipedia.
I don't really know. Social news sites and peoples' blogs seem like good ways to get this steady drip. If we got a link on front pages of Reddit or Hacker News every few days, that might be enough.
It depends on the spin. The Seigenthaler incident was so bad because it was completely negative and seemed to traumatize the higher-ups. They ran around like chickens with their heads cut off, doing something, anything to tell the press that 'we're fixing things!' And this lead to a general climate where BLPs are treated extremely harshly by Foundation diktat, which has in turn fostered a general highly negative attitude to new content, content you wouldn't find in the Encyclopedia Britannica, and anything not impeccably sourced.
(For example, page creation was turned off. Supposedly statistics were being collected to see whether it helped. Wales finally admitted in the Signpost a few years ago that they were lying through their teeth, no statistics were or are being collected, and the decision was never going to be reversed. I don't know what others think of Wales, but that was a breathtaking example of why I trust him and the Foundation as far as I can throw them and have never donated since.)
If the spin is 'here's a great site with lots of fascinating things to read', maybe that wouldn't be so bad. If it's more like 'look at these dangerous low-status techy fantasists'...
How about college newspapers, forums, meetups, talks, casual lunches and what ever else works. Colleges often act as small semi-closed social ecosystems so it is easier to reach the critical number needed for a self sustaining community, or the critical number of people to take an idea from odd to normal.
Yes, that incident did probably result in a lot of newcomers to the English Wikipedia. here is a graph of the number of editors of various Wikipedias. That incident became public in May of 2005. Note how the linked graph shows a take off from that point for the number of editors in the English Wikipedia (the red line). However, the other Wikipedias graphed (German in green, Japanes in yellow, French in blue) do not so such a jump. Those are also some of the largest other language Wikipedias. It is possible that around May 2005 the English Wikipedia hit the point where network effects and related issues caused a severe jump in user base and it is possible that the very large number of people who speak in English allowed that to happen. However, given the timing of the takeoff, the simplest explanation seems to be that the Seigenthaler incident made people pay attention to the project.
Eyeballing that chart myself, I don't see a "jump" around May 2005. It looks to me like the growth was pretty steady until Nov 2005, where there was a big jump over the next few months.
Ah, looking at the Wikipedia article on the incident more closely, it didn't become major news until November, so that explains that.
So the next question is, what evidence is there that the newcomers drawn by the incident was overall detrimental to Wikipedia? Again gwern's linked post does not seem to talk about this. ETA: Never mind, I retract this question after seeing gwern's other recent comments.
A working karma system, where low-quality contributions get downvoted, would mitigate this. Maybe the concern is that a large influx of people would degrade the quality of voting, but there are ways around that. Not letting new users vote for a while would work, but it would be pretty heavy-handed. Maybe there's a non-heavy-handed solution that also works.
It would mitigate it, but it's not getting at what I'm really talking about. Our equivalent of Wikipedia's vandalism isn't the issue - downvotes suffice for that.
What I mean is in general the flood of people not familiar with the existing material, not reading it, asking old questions, engaging in half-arsed arguments, misinforming others, etc. A handful can be educated by the regular they run into; if 1 newbie replies cluelessly to an old comment or post of mine, I can take the time to dig up the links and refute/educate them. If 100 newbies do that to me, I might just quit LW entirely. And what if they simply contribute high quality content that drags LW far away from what I value it for? (A number of LWers feel this has already happened with all the more 'practical' stuff, and would like to see LW return to its AI/decision theory roots.)
And then there's the problem that we don't really have everything organized. There's the Sequences, but that's overwhelming (Rational Wiki comments of LW that 'read the Sequences' is our polite way of telling people to go screw themselves). The wiki is very slowly being organized, but it couldn't handle a flood.
I think that 'read the Sequences' is our polite way of telling people that there is a certain body of ideas that forms a common background for the discussions here, and that it does not serve either the newcomer or the community, to post here before being at least somewhat familiar with that background. This is our polite way of telling people to go screw themselves.
Perhaps The Sequences might be better organised, although I have no ideas about how they might be. Ideally, being shown a link to anywhere in The Sequences should have the same reputation for drawing people into spending hours reading as TVTropes and Wikipedia do.
I suspect that most of the time, telling someone to read (all of) the Sequences is overkill: telling them to read a particular Sequence, or even just a few well-chosen posts, should suffice.
I'm finding User:ciphergoth's ebook of EY posts is a much nicer way to digest the archives than strategies like "read the sequences". (people on irc have been intermittently harassed over the last week or so by my random thoughts on posts from 2007...)
I think the "press next till you're done" approach is much easier to handle than looking at the lists of sequences and desperately trying put them all into some kind of order, open 5 million tabs, etc.
No, just no. I wouldn't wish "read the quantum mechanics sequence" on anyone.
May I ask why? It might be worth rewriting those sequences to make them more accessible - you don't need to see the exact math to be taught most of this stuff, after all, and I think that sequence probably does too much "teach Quantum Mechanics" and too little "teach the lessons from QM that are relevant" (mind you, I absolutely love that sequence because I want to learn QM, not just lessons-from-QM, but I can also follow the math ^^)
I disagree extraordinarily strongly with the statement
and the implication that one can learn "relevant lessons of QM" without actually learning QM.
Here are my comments on the QM sequence, which summarize my current stance. I welcome objections or corrections.
For both "Amplitude Configurations" and "Joint Configurations", you summarised it just fine without mentioning complex numbers (except to note that this is the formal notation used). This would be an example of "being able to teach the lessons/concepts of QM, without teaching the math" - you can leave out the complex numbers entirely, and just present the ideas.
I'm only about halfway finished, but it hasn't shown any signs of getting particularly more mathematical.
Mind you, I absolutely love that Eliezer is teaching the math, because it's something I want to learn. But the basic idea of "drop apple, it falls" and even "falling objects fall faster over time" can be taught without resorting to calculus. I think that same level of abstraction could be applied to the QM sequence to make it MUCH more approachable. (Heck, if people were interested, I'd be game for writing it :))
You misunderstood the purpose -- these are my comments on the sequence, not a summary. I seriously doubt anyone could actually learn anything from my notes alone.
Let me try to figure out where our disagreement is: I believe you can learn the basics of "how gravity works" without knowing the calculus that was used to derive it. You can learn the basics of "how gravity works" without even knowing g = 9.8m/s^2, or the algebra necessary to solve that equation. You can teach the basic concepts of "things fall", "heavy things fall as fast as light things", and probably even "things fall faster the longer they've been falling" to a 5 year old.
Do you agree that a child can learn something useful about the way the world works from these non-mathematical lessons in gravity?
The difference between gravity and quantum physics is that by the time someone is ready to learn about gravity, they've lived gravity and experienced it their whole life.
Yes, they've "experienced" quantum physics too, but their intuitions about it will (almost certainly) turn out to be mostly wrong; therefore, mathematics is required.
People seriously thought the Earth was the center of the universe. They thought that light objects fell slower than heavy ones. My intuitive experience of the world is that it's flat and the sky is a hemisphere enclosing me. I can still teach the reality of gravity to a five year old, despite it being unintuitive. You don't have to have everyday experience to learn something.
I'm confused why "experimental evidence" is less convincing than mathematics. I've taught the first half of the sequence to others without even mentioning complex numbers, so my anecdotal experience is that no, people do not need mathematics to correct their intuitions.
You're conflating experimental evidence (by which I imagine you mean the two-slit experiment and etc., correct me if I'm wrong) with everyday experience. The latter contains virtually no useful information about quantum physics. It entices us to think that matter is made up of particles, that observables take fixed values after being measured, and so on...
Well, I would argue that that's the least relevant sequence.
This comment caused me to read the entire QM sequence. Thank you.
Unfortunately it's also sort of misguided in purpose; the cosmological interpretation of quantum mechanics is the normality that quantum mechanics adds up to and doesn't need to be justified with nearly as much appeal to the secret powers of Bayes. I remember thinking this a few years ago; I remember thinking that people not immediately seeing it was evidence that compartmentalization was stronger than I thought amongst humans... I think I had a higher opinion of humans back then. Luckily Tegmark actually wrote up some math and has high status so I have someone to back me up for once.
That paper is dubious and confused. Their arguments revolve around infinite product states, representing an infinite number of causally disconnected copies of some physical entity. The whole argument is (i) in such states, the observable states of an individual entity appear as infinitely repeated factors with asymptotic frequencies equal to Born probabilities (ii) the basis decomposition of such infinite product states produces other infinite product states with the same property. From (ii), they wish to argue that the very notion of a cosmic superposition is redundant, and so there is no need for many worlds in the Everett sense. Or at least, they claim that there is no difference between the notion of one Everett world and many Everett worlds.
The first thing to note is that their whole construction really needs to be placed in a bigger context. The universe does not just consist of infinitely many causally disconnected copies of the same thing. Each copy is interacting with its environment, which (supposing the inflationary cosmology that they also assume) is in turn entangled with the degrees of freedom of its cosmological environment, all the way back to the beginning of inflation. There is no mention of entanglement prior to inflation, whether that is an issue, and how it could not be an issue if it is real. This lack of a larger framework makes it difficult to sensibly discuss what they have written. But their infinite product states really need to be embedded in some larger thing, the wavefunction of the universe, which is not a product state. The paper is dubious because it does not address this point.
The second thing to note is the total confusion regarding what the actual message of the paper is. Their technical argument is that a superposition of their special infinite product states is itself just another infinite product state. From this mathematical fact they conclude (page 9) that they don't know if it's a superposition any more. This isn't "adding up to normality"; this is like spinning on one spot so fast that you lose all sense of direction, and then concluding that all directions are the same direction, because you can no longer tell the difference between them.
To really take apart a paper like this one, it's not enough to show that it contains nonsensical statements; you have to figure out the philosophical genesis of the authors' mistake, and then show how they employed the formalism in the service of their mistake. In such a diagnosis, first you establish explicitly what constitutes valid reasoning about the aspect of the theory that they propose to address, then you show how they spin mathematically valid manipulations into wrong, tendentious, or meaningless statements about physical reality. This is a tiresome thing to do and I hope to avoid doing it in full for this paper. But meanwhile, I am quite confident that it belongs in the rogues' gallery of papers which falsely assert that they are now, finally, really-truly, solving the problems of MWI.
Noted; I will lower my confidence in its ultimate sense-making-ness.
Those sound like low-quality contributions to me. If they're net harmful, we should agree to vote them down.
Any one comment is not harmful. There are tons of those in LW already on the -2 to +2 range. What's harmful is all of them put together. The response is to either not have huge surges, or to collectively fight the surge by collectively imposing higher standards and voting much more actively.
And as with collective solutions, there are problems with defection... No individual benefits much from going through a new page and up or downvoting each comment. To quote Buffet, I think:
It isn't obvious to me that additional clueless comments are superlinearly harmful, that their harm outweighs the benefits of greater publicity, or that the problems with defection that you mention are serious enough to prevent a collective solution from working.
Can you think of other online communities that suffer or at least go through great and unpredictable change due to a high influx of new people?
The only one I can think of that's stayed relatively high-quality for a long time is Hacker News, and they actively discourage large influxes--for example, by flooding the front page with posts on Erlang internals when mentioned in the mass media.
Paul Graham also does very active experimentation with HN's reputation system, which I like: There are karma threshholds for voting down comments, higher ones for voting down posts; you cannot vote down a direct reply to your own comment; you cannot vote down a comment more than a few days old (this one wouldn't work as well here). The most radical change he's made is that only you can see the exact karma for your own comments (although comments below zero are progressively lighter shades of grey).
It could just be an extension of the not being able to make posts until you get 5 upvotes idea.
I agree. I'm trying to find if there are any studies on groups abilities to absorb new members, that might help us decide exactly how much publicity we should aim for. Unfortunately I don't know the proper keywords for this.
You might look for business culture studies. I've seen quoted around various places that apparently (maybe it was McKinsey?) a company can't grow its headcount faster than 5-10% a year without losing the company culture/spirit. (I'm not sure how one could prove that, as all the obvious metrics involve things like surveys and self-reports.)
I think people might be forgetting that we (or at least Eliezer) is already drawing a lot of publicity towards LW, in particular through HP:MoR. And, while I don't know the numbers, I feel that it alone already attracted a lot of people, whose quality is... well, probably somewhat average. And we are okay.
The problem is that unlike Cypherpunks or the Bitcoin people, LW is not a group whose agenda can be summarized meaningfully in a soundbite. This normally means that there isn't going to be any media interest in it, except insofar as it might get entangled with some other news story by accident.
The only other way for a blog/website to get publicity is if it's interesting enough to gain a very wide audience. However, the audience size that's necessary for traditional media to take note is probably incompatible with the intellectual standards that are supposed to be followed here.
Taken right from LW's homepage. That sounds to me like a very good start, and it's only a couple of sentences.
How's that for a soundbite?
If we want Less Wrong to achieve its goals of increasing the sanity waterline or attract minds to AI, we will need more publicity.
But at the same time, we don't want to lessen the average quality of posts. Putting Less Wrong in a standard newspaper would definitely do that. If we were to do publicity, we should probably do so in a place where there is already some self-selection for intelligence/rationality. Wired would probably a good magazine for this.
I'm not sure, I get the impression the people reading wired are people who like to think of themselves as rational, but aren't necessarily.
Do you mean something like "...people reading wired are people who would bristle at being told they aren't already rational, because their self image depends greatly on seeing themselves as rational"?
I mean that people reading wired tend to suffer from the Dunning-Kruger effect.
It's a good thing we don't do that!
We'd be better placed to make use of publicity if Eliezer's book was on the shelves.
The first time I saw that I read it as "We'd better be placed to make use of publicity when Eliezer's book goes on the shelves." Which I think is also a valid point.
That's a big part of what I meant!
Do you mean publicity for LessWrong, the "community for refining rationality" or for SIAI/FAI cause?
As someone who cares about LessWrong in itself, and also about SIAI/FAI, I'd be interested in answers to both interpretations of "we" in the question.
Also, I think any in depth story about LessWrong is bound to talk about SIAI/FAI and the connection between the two.
I think for LW as a community the type of publicity you're thinking about is just fine - we get some newcomers, they try the site, post some crap, get downvoted and either learn or leave. I guess the system will handle it, and we'll end up with a bigger community in the end.
With SIAI I'd be a bit more careful, I'm concerned that they are too easily put in the "crank file". Some journalists might easily go for the weirdness low-blow. Others (like the guy who did the main Singularity article for the Times) are more thoughtful and would be beneficial.
(This is why I asked the question.)
It seems all the other comments are voicing concern about bringing on an Eternal September phenomenon to LessWrong. That would be horrible.
But there must be so many potentially valuable people out there, especially if LW is serious about having a diverse membership as I've seen claimed before. If it's only low-profile ads in Wired that lead here, will LW get the diversity it desires? Perhaps there's some way to tune the karma system such that only the right eyeballs stick around, and the wrong ones leave quickly?
I ended up suggesting the story to the Wired writer, and yesterday got an email back from him:
1) It is very often the right response to a question to break it into sub-questions. Here, I think the opposite is true. "How do we want our publicity?" or the like, to keep focus on our steering from the present to one possible future*, perhaps with "doing nothing actively" being the default.
2) The gap between "main" and "discussion" seems large. Perhaps the discussion section could have subsections, or a new "this post is a public rough draft and work in progress" section could be created.
*Don't even go there.
Do we want more publicity... ? Well, we want to "improve human rationality", "raise the sanity waterline", right? Would more publicity help with that? I think we all agree the site benefits its users.
Could the site maintain its quality with a major influx of new users? I think so... we have limitations/minimum karma for upvoting, downvoting and posting, which should limit any ill intentioned or misguided disturbances. Hacker News is a good model btw, they managed to preserve site quality during major user growth by periodically raising the karma requirements for downvoting.
Can the site cope with A LOT more traffic? ... dunno, what say the site administrators?