The "cumulative advantage" study mentioned in the article is very interesting, and relevant to HN. Basically they built a music sharing site where people could download and rate music, and they'd see which songs became popular. The cool part is they randomly split the users into 8 subgroups, each with independent song voting and ranking. The top hits in one subgroup were very different from the top hits in a different subgroup, showing there's a huge random factor due to the feedback loop of people liking what's popular.
The obvious connection to HN is that people read and upvote the popular articles, so there's (probably) a lot more randomness and a lot less meritocracy in what makes it to the front page than people suspect. This matches my experience, where one of my blog posts will sink without a trace on HN, and then be hugely popular after someone resubmits it later.
It would be very interesting to do a "cumulative advantage" study on HN: split the users into sub-groups, and see if there's any correlation between the popular articles across sites.
I suspect there's also a "cumulative advantage" effect on comments, with popular members getting way more upvotes. Some high-karma member could quantify this by posting half their comments from a new account and A/B testing.
Society is full of positive (self-enforcing, runaway) feedback loops. It is not self-stabilizing. It tends to flip-flop between extremes. This applies to pretty much any field. It's the strongest argument against laissez-faire economics, yet it's amazing how many people are blind to it (well, that's exactly part of the mechanism that enables the aforementioned feedback loops).
> It's the strongest argument against laissez-faire economics, yet it's amazing how many people are blind to it
If you actually would read free market economomics, for example market process theory. You would learn that actually free market economists do not belive that the market is this always stable thing that is pareto optimal.
It is actually discribed much more as a lagging, thing where there are forces pushing the markets in the right direction in the longer run.
Also if you for example look at public choice theory, that is economic analysis to democracy, you will see that just because the market is not stable does not mean you want the state to stabalize it. The state faces insentives that make unlikly to do the right thing, even if we know what that right thing is. There you have to problem that you often get unintended consequences, that is even if you assume that the state is all nice and only wants to help you, there is still a question if the have the knowlage and forsight to actually implment things correctly. So in the end it makes it very lickly that the intervention actually has negative consequences instead of positive ones.
So, next time please inform yourself befor attacking a groupe of on the basis of what you think they belive.
tl:dr; free market econmists are not as dumb as you think they are, they actually have spend the last 150 years thinking about these problems.
Right, and then they (X) go ahead and dismantle all control mechanisms that prevent the self-reinforcing loops from tearing the world apart. Sure they are "enlightened". :)
(X) - By 'they' I mean libertarian politicians. They are the ones who actually matter, who can make a change. I don't care what anyone thinks in theory. All I care about is what happens where rubber meets the road, in the political arena - that's the stuff that could affect me.
> So in the end it makes it very lickly that the intervention actually has negative consequences instead of positive ones.
I'm not a big fan of folks who take their semi-religious, untested, views and use them to beat me over the head, under the pretense that they are well-informed and cultivated. The stuff you're repeating here is just the same ivory-tower circle jerk that you keep hearing in libertarian circles as if it's Gospel, yet it could not be more removed from reality.
You people live in this fantasy because reality has been kind to you. You've lived through uninterrupted plenty and now can't distinguish between what's really out there and your own narrative. Some hardship would wake you all up really quick - unfortunately, there's no way to do that without affecting everyone else. :(
Your only argument is that there are no negative unintended consequences and that is simply idiotic. Just look at American Forgin policy, blow back anybody? How many more times does america have to invade iraq to fix the last time they where there. Or insert any other country america has been evolved with (exeption might be Japan and Germany but those where EXEPTION see http://www.amazon.de/After-War-Political-Exporting-Democracy...)
The same is true with strang economic policys that trie to patch up unintended consequences of there last economic policys. If you belive that the housing crisis is responsible for the larger economic crisis or not you HAVE to admit that the goverment in a huge way engourage housing and private house ownership. There mantra was for a very long time 'Every american should own a house' even if there is no evidence at all that societys with higher house ownership rates are any better of then nations with the same level of overall wealth. Now they have to bail out the half private half goverment 'lets press down housing prices'-agency.
Do you want to start talking about drug policy? I always fought the idea behind drug policy was to prevent people from doing drug, bot ohoho. It lead to civil war in mexico, america forces operating all over south america, 1% of the american population in prision, inner citys with high murder rates and bad schools, huge healthcare cost because of bad drugs and so on and so on (I could talk days about this alone).
Tell me more about that 'semi-religious, untested view'. Most of these are well known and exepted by most of social sience. There are some harder one where there is more disagreement but overall it is absolutly clear that when goverment tries to fix something unstable they probebly arn't understanding all ramifactions there actions have.
You might now argue but if we had better 'experts' (and im sure you have a list of who those are) and gave them enougth money (at least another 10-20% of GDP plus some extra debt) and power (see how wunderful that works with the NSA) they could fix all those problems. The would finally stop all the lobbying. If those expert fail because they two dont know enought to fix the problems and also get payed by buissness then we should find even other expert and give them more money. If we could just give the right people the right positions with the right abount of money and stop lobbying and fix govermental system and voting then everything would be well and the goverment could solve all the 'problem' (things you dont like) without unintend consequneces.
> without affecting everyone else
That argument comming from you? Isn't it you who wants to take my money (or your country mans) in order to fix some problems you think you have? Isn't it you that want to take my money and give it to whatever industry is failing at the moment (because you know the 'poor workers'). Isn't it you that does not want actual laber market in teachers because then teachers would act with a profit motive and that would be bad. Isn't it you who want to prohibite people from trading freely, or consuming whatever they want?
Note im using 'you' here as in people who think markets need to be constantly corrected and profit motive is a bad thing, since you have not actually given any examples of what these 'self-reinforcing loops that can tearing the world apart' are. Seams to me that saving industrys that nobody needs anymore is actully the oppsite of what you want. The market corrects it self, maybe it does not correct itself in a way that you want, maybe you dont want unemployed people in detroit, maybe you think that american farmes deserve huge amount of money in order to survive. But if you then stop that mechanism and start paying people in detroit to do nothing, sure you stopped had good intentions but in the long run it will only hurt people there and everybody else that has to pay for it (and do not have the benefit of profiting from those people actually working somewhere profitable).
Can you provide one example of such a phenomena operating on anything longer than a very short timescale? I'm genuinely curious. Nearly every example I can think of is long-term negative.
edit: I'd venture to say anything longer than 5-10 years could be considered long term for this question. Under that measure, I'd say government spending has exhibited positive-feedback behavior over the last 50+ years. Anything else?
I wanted to argue originally that the original iPhone is a nice example, if you compare it to other 2007 high end phones like the N95, but a theoretically perhaps clearer example is the evolution of soccer formations, they evolve to beat the currently used formations. To pick a specific example, the 4-4-2 with diamond midfield is a formation which is good in two situations, beating itself and being perfectly suited to the 1998-2000 French international squad, which did win the World and European Championship. After this success suddenly everybody wanted to play a 4-4-2 diamond. Interestingly it is not very good against the dominant formation of the late nineties, the 4-4-2 with flat midfield.[1] The solution to this problem was the development of the modern full back, [2] and suddenly the best players are playing the position with the least influence. And this in turn leads to shifts in tactics, such that the full back position can more easily join the attack. [3] So I would argue, that if France did not have the specific players they had in the late nineties, then modern football would look very different than it actually does. Especially we would still see very unremarkable players at full back.
[1] A flat midfield has two wingers on each side, a diamond midfield only one.
Religion. Fashion ex Louis chairs. Language ex regional accents and new words / phrases. Manners and other social protocalls. Operating systems, social networks, and computer languages.
In Japan, this is even more pronounced. MFGs seek to ride the wave of teen uptake of new fast-moving products. These products tend to enjoy a short-lived, but intense attention by consumers.
The obvious connection to HN is that people read and upvote the popular articles, so there's (probably) a lot more randomness in what makes it to the front page than people suspect. This matches my experience, where one of my blog posts will sink without a trace on HN, and then get reposted later and shoot to the top.
Perhaps a more apt comparison would be how certain blogs (37signals, marco.org, Svbtle, etc) rocket to the front page based off of their TLD alone. I don't think this is a particularly bad thing -- this is how branding works, after all -- but I would definitely be interested to see an experiment that throws everything on Readability and hides the URL to see if things change.
I suspect there's also a "cumulative advantage" effect on comments, with popular members getting way more upvotes. Some high-karma member could quantify this by posting half their comments from a new account and A/B testing.
I could be incorrect, but I recall reading that part of the ranking algorithm is based on the commenter's average karma.
As I remember it, there was a feature implemented where leaders (veteran accounts with high vote averages) do get something of a boost when they comment, at least within the sequencing of the comments in any one thread. I think this was done to help improve the quality of the conversation, even before the removal of the vote totals next to each comment.
Definitely! This is especially a big issues with focus groups (I had to run a bunch for a few semesters for part of my marketing degree, and one of the first things you learn is how to disrupt an echo chamber: if X and Y say two very positive things about a widget, than Z is suddenly very disincentivized to say anything negative and everyone else is more inclined to join the echo chamber).
But I don't think that's whats being brought up with the article. Literary critics are more or less the 'first voters' on new books: and yet, the correlation between critical successes and commercial successes are tentative at best (while the first few votes on an article are always an indicator of future votes.)
This is mainly, I think, a function of accessibility: you see the rating of a link on HN before you even click on the link, while most (I'd say 90%+) of the people buying books do so unaware of what everyone before them thought about the book.
That study has been very influential to my own thinking about how things get popular. Another thing I found fascinating about the study was that 'quality' of the content did matter, but that it accounted for roughly half the correlation with success, the other half being random-chance snowball effects.
You could probably extend this theory of what makes things successful to a lot of other domains (music, books, fashion, ideas, memes, startups, people...)
However, it's possible the ratio of intrinsic qualities vs extrinsic luck varies or is not 50/50 in all domains.
With music (or books) you have to invest a few minutes or a few hours to really evaluate if you like something, so those factors may affect how much extrinsic "cumulative advatage" snowball effects count vs intrinsic quality.
I'd really like to see more empirical research of how success by "luck vs skill" actually plays out in other categories. Any one know of follow up work to that 2007 study?
It seems like this would be a reason to release more high-quality work rather than fewer excellent quality work. Although perhaps the excellent work would stand out more.
Some high-karma member could quantify this by posting half their comments from a new account and A/B testing.
As it happens, my wife is a HN member but has not found it easy to get traction for stories she submits. On a couple of occasions I have re-submitted an article that I thought was unfairly overlooked and it has attracted replies and karma in abundance. Anecdata to be sure, but based on this and other observations in HN and other online communities I think your hunch is likely correct.
If anyone feels like exploring this in R or somesuch, MetaFilter (where upvotes by other members are public) makes a large amount of its user pattern data available for analysis: http://stuff.metafilter.com/infodump/
When you said it was relevant to HN I thought you were going to say because the same phenomenon happens with start-ups. Being good isn't always enough, usually you need a little luck as well.
The other thing to keep in mind is that well over half the battle in becoming a bestseller is even landing a publisher in the first place. There's serious cumulative advantage inherent to being Stephen King, or JK Rowling, in that you can submit your "anonymous" work to your ordinary publisher and at least guarantee some amount of distribution.
A truly anonymous play would be if Rowling or King submitted their pseudonymous work to agents and publishers, unsolicited, without any trace whatsoever of their existing relationships. (This might not be possible due to contactual obligations, however).
> A truly anonymous play would be if Rowling or King submitted their pseudonymous work to agents and publishers, unsolicited, without any trace whatsoever of their existing relationships.
Plenty of real world examples of that effect. Look at bands that are huge in say, Japan or Germany but struggle to fill a small club in their home country.
The obvious connection to HN is that people read and upvote the popular articles, so there's (probably) a lot more randomness and a lot less meritocracy in what makes it to the front page than people suspect. This matches my experience, where one of my blog posts will sink without a trace on HN, and then be hugely popular after someone resubmits it later.
It would be very interesting to do a "cumulative advantage" study on HN: split the users into sub-groups, and see if there's any correlation between the popular articles across sites.
I suspect there's also a "cumulative advantage" effect on comments, with popular members getting way more upvotes. Some high-karma member could quantify this by posting half their comments from a new account and A/B testing.
link to more on the music experiment: http://www.nytimes.com/2007/04/15/magazine/15wwlnidealab.t.h...