Are closed social networks inevitable? (2010)
danluu.com
danluu.com
Did she ever succeed? Did she write her thoughts in a more digestible format and make those writings public? Does she have further commentary on her thesis on the capitalist or non-capitalist tendencies of modern software companies, or on other topics?
Mostly I'd just love to read more of her writing, if it exists.
Edit: I believe she is Rebecca Frankel, according to http://danluu.com/programming-blogs/#rebecca-frankel. Still interested in any pointers to more of her writing if it exists.
I went digging into a Buzz post about the Boston startup scene that Piaw referenced, and could only find a crufty unstyled version in the Internet Archive. If you can forgive the formatting, it's still excellent reading: https://web.archive.org/web/20100613173203/http://www.google...
Another is the part about people paying ad words for freedom or stuff they couldn't otherwise get. The fact that Google's system can reach more people and more cleverly than before is certainly a benefit. There were alternatives, though, like listing your product in online stores, Buy It Now on eBay with a link to a store, or mass marketing after buying listings of leads. Similarly take less staff than traditional marketing at big firms. The online ecosystem for ads also involves lots of fraud with estimates showing it is a good percentage of ad expenses. So, Google is actually yet another middleman in this sense that tries to lock you into their set of benefits and problems. They also try to shift the benefits toward themselves a bit more over time like any business. So, they're just middlemen with a bent toward lock-in rather than purveyors of freedom.
So, aside from the Google comments, other stuff was pretty interesting. Lots of historical and economic tie-ins to her analyses.
EDIT to add: I'm only through around half the post. It's pretty huge haha.
"When I was at Google I hung out "
She also apparently worked at Google. Explains the bias. :)
I'm very curious whether she still thinks and hopes the same things about Google that she did then.
I'm not following your extension of artificial scarcity to Google. Do you mean that because PageRank isn't public, it's artificially scarce? That doesn't seem to hold water....there's no anti-user conspiracy theory required to understand why PageRank is private, but rather the fact that it would rapidly render Google Search pretty useless, due to the strong financial incentive for publishers to "teach to the test". A federated Facebook with public APIs would add to user surplus/utility, while a system based on a public PageRank would rapidly degenerate into a measure of how good a website is at matching PageRank, which would waste a ton of resources _and_ make search quality much worse (again, destroying user utility).
Maybe you've misunderstood what artificial scarcity and rent-seeking mean. She's talking about it in the widely-used economic sense of "scarcity that's imposed solely to extract rents, inferior to an alternative method which would increase net utility". A federated Facebook and open-API Windows both fit this neatly[1], and a public-PageRank Google doesn't even come close.
[1] One could make the argument that Facebook needs to be closed for reasons that benefit the user, but I've never heard any compelling arguments for that.
Ahhh. That could be it. I didn't realize it was a defined economic term. I thought it meant creating an artificial monopoly to grab a market, lock users into the benefits, deny competition via that tech, and make a financial killing in the process.
If I was wrong, I take back the critique based on artificial scarcity but keep it based on monopoly. PageRank was simply too good to beat. Patenting it led to their billions and continued dominance. They did smart things on top of that success. They might have not pulled it off without being monopolistic about search, though.
"a public-PageRank Google doesn't even come close."
How? It let them dominate search, control much ad revenue, put players with most money on top results, create a major contender in browsing, and later dominate mobile market. All that came from PageRank being both good and patented.
> How? It let them dominate search, control much ad revenue, put players with most money on top results, create a major contender in browsing, and later dominate mobile market.
The context of my original comment was that a public Pagerank algorithm isn't an artificially scarce environment. I don't understand your "how" question, and I don't see what the rest of this paragraph has to do with artificial scarcity.
Duckduckgo doesn't have a public spec for their algorithm either, because again, it would render their results useless in a short amount of time. That example should hopefully illustrate how keeping one's ranking algorithm private has pretty much nothing to do with artificial scarcity or abuse of monopoly.
I already told you I retracted artificial scarcity claim since I didnt know it had established meaning in economics. I shifted to a similar argument based on monopolies. Google's PageRank wasnt just publuc: it was patented (monopoly), extremely effective, and more efficieng than contenders that sprang up. That nobody could copy or beat the algorithm is why they stayed on top despite tons of search and meta-search engines existing.
If it was public and not patented, search landscape + Google's valuation might look a lot different. The monopoly on the best technique is why they got huge rather than a temporary advantage.
Anyone can link to my site from whatever social network they use.
I may or may not be able to link to everyone else, though. So far, I have no external links on my site. They're too likely to be broken in a year or three. If I get email related to my site, I often post it to the site.
I hope I can continue to interoperate with other people like this. I have tried a couple other social networks like Hacker News and Twitter, but if they disappeared or acted in a way I didn't like, I could abandon them without much loss.
Do I have any reason to be scared that my current setup might become "closed" in the future? Not trying to be paranoid, just wondering.
> If I get email related to my site, I often post it to the site.
But only after consent from your the sender, right?
> But only after consent from your the sender, right?
I'm struggling to decide the right way to handle this. So far, if the message is substantial and I want to reply for follow-up, I also ask directly for permission. If it is a small message or I otherwise don't end up with consent, I'll remove any personally identifying information and/or paraphrase the message. You can see this in action at [0].
I'd love to know where most people think the proper line is. I usually consider direct email to be a rather private communication channel.
[0] https://www.gaxun.net/commentary/knuth-challenge/comments/
BBC is an exception to this - the way they publish their news pages means they have articles going back over a decade (with the original styling)
You are one of the few people right now using the internet in a decentralized fashion- I hope that the recent advances around cryptography, mesh nets and blockchain systems will soon make it practical for more people to join your ranks!
I'm actually not brave enough to open a few public ports on a server I own and manage and drive traffic there. It's a mix between "scared" and "it only took an hour to set up the entire thing the easy way."
But I can sleep soundly knowing that if any of the providers I'm currently using become incompatible, I can move things very quickly to a new location.
Especially mail is difficult to get started with now - domains that don't have years of (spam-free) reputation tend to be blocked a lot, creating a bit of a chicken and egg issue[1] to get past, and it is... extremely frustrating to figure out what hoops j-random-webmail.com demands you jump through. Also, setup is complex, mainly due to the accretion of of anti-spam half-measures that need to junk up your DNS if you want people to accept your mail.
But trust me, even with all that, setting up mail today is still much easier than it was in the Sendmail days.
Assuming you're interested, and you want to, I'd encourage you to try running more services. When experimenting, don't keep private things on your host, make sure you have your machine access covered (passwords, keys, magic customer service phrases, whatever) and keep an eye on it. If it is compromised, consider it a learning experience and recreate your config (I'm assuming you're running a cheap virtual host; if this is hardware, that's a bit different).
It isn't that hard to do, and I think too many people are much more scared of running their own services than is sensible or real.
Netizens, arise! You have nothing to lose but your shackles.
And maybe some time that you otherwise would have wasted on Facebook.
[1] A few sites still block my domain, despite being single-owner, always spam-free and online for almost 20 years. I don't feel bad about not having them as potential conversation partners.
Mostly because when running sendmail, well, nobody cared.
I'm proud to run my own mail and webserver, but you do have to be careful
If you want to look into what other people are doing with "my domain is (the center of) my social network", check out the community at https://indieweb.org – some clever ideas & tools around communicating updates etc between sites and integrating with existing social media where necessary.
I love port knocking for this reason. You can take a baby step between not opening a port at all and opening it wide to the world ... for now, just open it up to yourself.
My sshd and personal HTTPS wiki are hiding behind knockd. I like the idea that they are simply invisible to the rest of the Internet.
A bunker does nothing to help avoid the war itself.
Yes, a little bit, I think ... specifically with regard to running your own email server.
I have run my own personal email server for about 20 years and rsync.net and Oh By have their own email servers as well.
What I have noticed is that gmail (and AOL) consistently classifies my email as spam even though my configuration and spam scores are flawless. I am on 10+ year verified clean IPs, I have everything (DKIM/DMARC/SPF) configured ... and email to new addresses very, very often (although not always) goes to their spam folders.
Google/AOL/MSN/whomever are incentivized to not support private email servers.
If you have personnel and resources dedicated to making this work (as a medium/large business might) then you can keep up with this wack-a-mole email issue. If you do this on your own it becomes very expensive in terms of time and frustration.
Personally, I will not give up self-providing my own email[1] but in answer to your question, there is a lot of pushback against running your own SMTP.
[1] In fact, it is my ambition to begin self-providing my own dialtone ... providing my own VOIP service, etc.
"When I was young my father read to me “20,000 Leagues Under the Sea,” advertising it as a great classic of futuristic science fiction. Unfortunately, I was unimpressed. It didn’t seem “futuristic” at all: it seemed like an archaic fantasy. Why? Certainly it was impressive that an author in 1869 correctly predicted that people would ride in submarines under the sea. But it didn’t seem like an image of the future, or even the past, at all. Why not? Because the person riding around on the submarine under the sea was a Victorian gentleman surrounded by appropriately deferential Victorian servants.
Futurists consistently get their stories wrong in a particular way: when they say that technology changes the world, they tell stories of fabulous gadgets that will enable people to do new and exciting things. They completely miss that this is not really what “change” – serious, massive, wrenching, social change - really is. When technology truly enables dreams of change, it doesn’t mean it enables aristocrats to dream about riding around under the sea. What it means is that enables the aristocrat’s butler to dream of not being a butler any more — a dream of freedom not through violence or revolution, but through economic independence. A dream of technological change – really significant technological change – is not a dream of spiffy gadgets, it is a dream of freedom, of social & economic liberation enabled by technology."
As someone who actually read Jules Verne (instead of watching the Disney movie or reading a poorly-translated abridgement,) I would not characterize Nemo as a Victorian gentleman.
First of all Verne wasn't English, and Nemo isn't a Victorian gentleman, he's a rebel and a terrorist. He doesn't have appropriately deferential Victorian servants, he has fellow freedom fighters.
Arronax's servant Consiel is the closest to a "appropriately deferential Victorian servant" but the way he is written, he comes across as a borderline Asperger's Syndrome savant scientist, not at all like Mercury from "Bleak House."
I get that people make up stuff to support their thoughts and biases, but calling Jules Verne a Victorian writer is simply incorrect. He's a lot closer to Dumas than Dickens.
Piaw: "The magic trick that Facebook pulled off was getting the typical user to provide and upload all his/her personal information. It’s incredibly hard to do that: Amazon couldn’t do it, and neither could Google. I don’t think it’s one of those things that’s technically difficult, but the social engineering required to do that demands critical mass. That’s why I think that Facebook is (still) under-valued."
Rob: "@Piaw - it was an accident of history I think. When Facebook started, they required a student ID to join. This made a culture of “real names” that stuck, and that no one else has been able to replicate."
I added the emphasis. I think this exchange is significant in thinking about an open and better replacement. Why hasn't popped into my head.
I've been pointing this out for years, recently ran across another source, at Unz, who makes the same observation.
Facebook transitioned from "the cool kids" to "all the kids". They've been playing a highly defensive game, through pricey acquisitions, of seeing that they remain "all the kids", though with significant leakage forming.
No single other party has decided to enter into the frey with a decentralised social networking model (Google could have, didn't, and at this point I'm suspecting won't, more on that: https://plus.google.com/104092656004159577193/posts/3L3Z5GhJ...).
One thought that's been bubbling up my head is that we still don't fully understand networks, values, and costs. First it was Sarnoff (V = n), then Metcalfe (V = n^2), then Tilly-Odlyzko (V = n(log(n))).
I argue that V = n(log(n)) - kn
That is: there's a decreasing value for each additional user, but a constant cost function. Where log(n) < kn, adding additional users hurts you.
You also risk a smaller, higher-value network beating you out.
There are a few implications, one of which is that network size is highly dependent on 'k'. This strikes me as a very general characteristic, common for pretty much any network-type structure: social networks, phone systems (crank and sales calls), cities (crime, disease, congestion), computer chips (heat, noise, defects), and more.
There are also some psychological and perceptual limitations at play. I've been trying to find what I can on human limits of information consumption. Several sources (Stephen Wolfram, Walt Mossberg) suggest ~100 - 300 emails/day is a near upper limit, the NY Times comments moderation desk averages slightly fewer than 800 comments/day per person. That's about one every 36 seconds, straight.
I know my own bandwidth for articles, comments, posts, and books is limited, as well as my capacity for handling longer forms once I've started consuming shorter-form content.
I'm tumbling this around to see what I can get out of it. Not sure yet.
1. I may want to have different social graphs on different networks. Facebook for friends and family, LinkedIn for work, Google Plus for hell if I know, etc. Not only would I probably want different graphs for the different networks, I'd also explicitly want to avoid having those networks know about the social connections I have outside of it (e.g. I don't want to just tell LinkedIn to ignore my family, I specifically don't want it to have access to that info in the first place).
2. The social graph isn't the only component of the network effect. Even if we had a collectively-owned social graph, all of the content that you produce or consume is limited to a single social network. So even if Google Plus could see my entire Facebook social graph, anything I post on Facebook would not be on Google Plus (and vice versa). So the network effect is still strong, because everybody would still have to gravitate to the same social network in order to see each others' posts.
The only way this really works is if we have a federated social network, where all federated networks share the same social graph and have access to all the content. But while federation works for a messaging platform, it doesn't really work all that well for a social network because it also means all networks have to have the same feature set, and that's hugely restrictive. If Facebook was federated with Google Plus, Facebook couldn't introduce any new features since Google Plus doesn't have them. Or Google Plus couldn't introduce its Circles concept because Facebook doesn't (didn't?) have that.
The speakers are trying to convey too much nuanced information at once for each point, and the text is presented in a small annoyingly dense format.
More editorial structure to make it easy to scan for the main points, and read through their related details as necessary would be required for me to have an interest in ingesting the content.
That isn't to say 'make it bite sized', that's saying "Give it structure, organization, and consolidate like things together".
The problem, for end users, seems to have largely been the classic: Security is Difficult issue.
The ideas of microkernels and, well, TCP, suggest otherwise.
But I think the larger point stands - the real change is in indexing the public internet, making it possible (in theory) for a small player to reach the larger world without the capital investment that would've previously required.
Reality is not living up to the early optimism over the so called "long tail", but I believe that only reinforces her point about the dangers of centralizing the internet.
Hence I think it is beyond the power of a company to monopolize the social sphere in Orwellian fashion, but monitoring, nudging, and creating an atmosphere - those things happen in any coffeehouse or bar. Small towns are known more for lack of privacy than the opposite.
In the past week I "returned" to Facebook after years of effective silence on it. The impulse was relationship-based, as many of these things are, but I had to decide on a method of engagement and decided that I was going to treat FB as a direct extension of Twitter, which I've kept up with - just tweet as usual but follow a second thread. I used to want to keep them separate to have parallel lives, but since I abandoned FB that "life" was already dead and I have learned a way of public living on Twitter that I am comfortable with.
Fortunately there is enough linkage between the two that this is straightforward. They have presented a tool that adapts to my preference, rather than an imposition or decorum. This, I think, is the direction that social software is moving towards inexorably - to separate or merge bodies, accounts, and personas as needed, across systems, according to various models of seeing the world.
I know I slowly stopped reading it as the spam and new customs (top posting, not trimming posts) crept slowly in, and most of the action switched to web forums (because of shiny HTML and what not).
Even if your ISP dropped NNTP, there were still free NNTP servers around, so I'm less than agreed about that expectation. And the Eternal September was in 1993. Usenet was "viable" well past ten years past that.
I think the best explanation is "new shiny Web" although about... half of the folks I used to read on Usenet are now on Other Fora and it's nowhere near the same - it's not a substitute.
See it at qbix.com/platform -would love to hear your opinions.
It is economically more efficient to centralize the resources but far less robust. SV could disappear tomorrow and a significant amount of the Internet's "revolutionary" capabilities would disappear. This doesn't appeal to SV's narcissistic sell of a special snowflake, but the power they wield in their capacity to damage any progress associated with the growth of the internet.
On the other hand, a distributed internet is far more diverse and robust yet the economic costs can be exponentially far more damaging and costly.
It is like this though. Any one of us can re-create our social graph in a heartbeat on any of these services at any point in time by importing our contacts and letting the network present us with our associations at the moment in time we want it to.
I've done this with linkedin, having zero presence there until the moment I wanted to network in a new area.
I've done this with snapchat
Telegram
anything I want at that point in time
there is no need for continued presence on any one of them since you can come and go as you please with new profiles and reconnecting to existing associates.
I read the open/closed in the article as less about that and more about the fact that e.g. I can't push a twitter post to Facebook without Facebook specifically putting that function into Facebook. I also can't combine the comment streams from Twitter and Facebook. That's the kind of openness that I thought the article was referencing.
That is what the social graph is about - making it possible to describe and discover the edges.
What types of edges might you want to have that existing models don't support?
All these metrics matter a lot in the social space. We can't get them (with, see below, a few exceptions like HTML - and boy is having the edges there a useful thing)
An article I ran acrosss a few days back pointed out how the nature relationships also matters. It starts off noting that gender is often given as binary (M/F), when reality can be more complicated (though most of the complications are a distinct minority case). That's something I ran across years back looking at healthcare data, where "gender" had in some cases five values: male, female, indeterminate, other, and unknown.
Relationships, though: "friend", "family", "co-worker", "ex-spouse", "enemy", ...
Any thoughts on capturing something such as this (or of similar complexity) on a graph? Or references?
I suspect international or inter-corporate relationships might be similar.
FOAF: https://en.wikipedia.org/wiki/FOAF_(ontology)
XFN: https://en.wikipedia.org/wiki/XHTML_Friends_Network
There's probably more. There's always more standards :)
How much does it cost facebook to handle traffic in India? How much would it cost for american-only phone networt to create and connect new network in India?
But now that they can use the internet, the networks become closed.
Interesting theory.
(DNR)