An article from Barbara Fister about the role and biases of large providers like Google and Facebook in curating, sorting, filtering their content, usefully contrasted with academic librarians’ closely related but importantly different roles. Unlike a library, such systems (and especially Facebook) are not motivated to provide things that are in the interests of the public good. As Fister writes:

“The thing is, Facebook literally can’t afford to be an arbiter. It profits from falsehoods and hype. Social media feeds on clicks, and scandalous, controversial, emotionally-charged, and polarizing information is good for clicks. Things that are short are more valuable than things that are long. Things that reinforce a person’s world view are worth more than those that don’t fit so neatly and might be passed over. Too much cruft will hurt the brand, but too little isn’t good, either. The more we segment ourselves into distinct groups through our clicks, the easier it is to sell advertising. And that’s what it’s about.”

These are not new points but they are well stated and well situated. I particularly like the point that lies and falsehoods are not a reason to censor a resource in and of themselves. We need the ugliness in order to better understand and value the beauty, and we need the whole story, not filtered parts of it that suit the criteria of some arbitrary arbiter. As Fister writes:

“There’s a level of trust there, that our students can and will approach a debate with genuine curiosity and integrity. There’s also a level of healthy distrust. We don’t believe it’s wise to leave decisions about truth and falsehood up to librarians.”

Indeed. She also has good things to say about personalization:

“If libraries were as personalized, you would wave your library card at the door and enter a different library than the next person who arrives. We’d quickly tidy away the books you haven’t shown interest in before; we’d do everything we could to provide material that confirms what you already believe. That doesn’t seem a good way to learn or grow. It seems dishonest.”

Exactly so.  She does, though, tell us about how librarians do influence things, and there’s only a fine and fuzzy (but significant) line between this and the personalization she rejects:

“Newer works on the topic will be shelved nearby that will problematize the questionable work and put it in context.”

I’m not sure that there is much difference in kind between this approach to influencing students and the targeted ads of Google or Facebook. However, there is a world of difference in the intent. What the librarian does is about sense making, and it accords well with one of the key principles I described in my first book of providing signposts, not fenceposts. To give people control, they have to first of all have the choices in the first place, but also they need to know why they are worth making. Organizing relevant works together on the shelf is helping students to make informed choices, scaffolding the research process by showing alternative perspectives. Offering relevant ads, though it might be dishonestly couched in terms of helping people to find the products they want, is not about helping them with what they want to do, but exploiting them to encourage them to do what you want them to do, for your own benefit, not theirs. That’s all the difference in the world.

That difference in intent is one of the biggest differentiators between a system like the Landing and a general-purpose public social media site, and that’s one big reason why it could never make any sense for us to replace the Landing with, say, a Facebook group (a suggestion that still gets aired from time to time, on the utterly mistaken assumption that they duplicate each other’s functionality). The Landing is a learning commons, a network of people that, whatever they might be doing here, share an intent to learn, where people are valued for what they bring to one another, not for what they bring to the owners and shareholders of the company that runs the site. Quite apart from other issues around ownership, privacy and functionality, that’s a pretty good reason to keep it.

Address of the bookmark: https://www.insidehighered.com/blogs/library-babel-fish/ghost-machines-loving-grace

A really interesting paper on making crowds smarter.  I find the word ‘confident’ in the title a bit odd because it seems (and I may have misunderstood) that the researchers are actually trying to measure independent thinking rather than confidence. As far as I can tell, this describes a method for separating sheep (those more influenced by others) from goats (those making more independent decisions), at least when you have a sequence of decisions/judgments to work with. The reason it bothers me is that sheep can be confident too (see the US election or Brexit, for example).

We know that crowds can be wise if and only if the agents in the crowd are unaware of the decisions of other agents. If there’s a feedback loop (more accurately, I believe, if there is an insufficiently delayed feedback loop) then you wind up with stupid mobs, driven by preferential attachment and similar dynamics. This is a big problem in many political systems that allow publication of polls and early results. However, some people are, for one reason or another, less influenced by the crowd than others. It would be useful to be able to aggregate their decisions while ignoring those that simply follow the rest, in order to achieve wiser crowds. That’s what the method described here seeks to do.

The paper is more concerned with describing its model than with describing or analyzing the experiment itself, which is a pity as I’d like to know more about the populations used and tasks performed, and whether it really is discriminating confident from independent behaviour. I’ve also done some work in this area and have written about how useful it would be to automatically identify independent thinkers, and to use their captured behaviour instead of that of the whole crowd to make decisions, but I have never implemented that because, in real life, this is quite hard to do. In this experiment, it seems quite possible that the ‘independent’ people might simply have been those that knew more about the domain. That’s great if we are using a sequence of captured data from the same domain (in this case, length of country borders) because we get results from those that know rather than those that guess. But it won’t transfer when the domain changes even slightly: knowing the length of the Swiss border might not well predict knowledge of, say, the length of the Nigerian border, though I guess it might improve things slightly because those that care about such things would be better represented in the sample.

It would take a fair bit of evidence, I suspect, to identify someone as a context-independent independent thinker though, given enough time, it could be done, it would be well worth doing, and this model might provide the means to identify that. I’d like to see it applied in a real context. There are less lengthy and privacy-invading alternatives. For instance, we might capture both a rating/value/judgement/whatever and some measure of confidence. Some kinds of prediction market capture that sort of data and, because of the personal stake in it, might achieve better results when we do not have a long history of data to analyze. Whether and to what extent confidence is related to independence, and whether the results would be better remains to be discovered, of course – there’s a good little research project to be done here – but it would be a good start.

Address of the bookmark: https://arxiv.org/abs/1406.7578

Landing-like software from CUNY, based on Buddypress, intended to provide a learning commons with relatively little effort or configuration. It’s a nice bit of packaging, slick, with good collaboration tools and a simple, activity-stream-oriented social network. Commons in a Box is definitely worth looking at if you need a site to support a bottom-up social community or network, and you don’t have a wealth of resources to put into building your own. 

I came across this software because it is being used in the University of Brighton’s newly reborn community site at https://community.brighton.ac.uk which, until it was killed off last year, used to run on Elgg.  I remain a fan of Elgg for building such things, which has a lot more options than BuddyPress available by default, richer access control, and a much more elegant technological design that makes customization more robust and flexible, but this seems to be a great simple solution that just works without demanding much effort, and that, thanks to its WordPress foundations, could be customized to do pretty much anything you’d want a bit of social software to do. 

Address of the bookmark: http://commonsinabox.org/

A sad article, if ever there was one. This is about Google’s in-kind response to Facebook’s depressingly successful attempts to be a bigger and better AOL/Compuserve (amongst other things, through its ‘philanthropic’ internet.org arm, that people in developing countries afflicted with it sometimes think of as the Internet). The general idea is that Google will host content, rather than linking to it.

This is not the way the Internet should go, and this is not in line with Google’s avowed intent to not be evil. On the bright side, in real life, though poisonous and virulent, it is not the way the Internet really is going: the Internet is, ultimately, self healing, both in technical and in social terms. It might look like a fairly closed system to people that generally interact with it through Facebook or Google Search (or any of hundreds of thousands of other less successful attempts to lock people in) but it is heartening that WordPress dwarfs all of them put together in terms of sites and people that visit them (more than a quarter of all sites), and Worpress sites are, to a very large extent, controlled and owned by the people that run them. And that’s just the most popular content management system: the Web is many times bigger and more distributed than that, and the Internet is vastly bigger still. And, of course, Google is not the only search engine. You can find the rest of the Web in many other ways.

So, though the article claims doom and gloom all round, I remain optimistic that common sense and decency (or indecency if that happens to be your thing) will triumph in the end, and the game will never be over. A few successful parasitic corporations/applications – Facebook (including its subsidiaries like Instagram, Whatsapp, etc), Google, Apple, Amazon, Twitter, Snapchat, LinkedIn, Yahoo, Microsoft, Pinterest, etc – are doing their darnedest to wreck the open Internet, and are definitely shaping much of it, killing open standards, and sucking billions of people into their locked-in lairs, but those billions of people are just a click away (often, from within those systems themselves) to what the Internet is actually composed of, especially the Web side of it. Sure, these are parasites that suck the life out of openness and diversity but, like all parasites, they would be more than foolish to kill their host. And, hearteningly, the network effect (especially the rich get richer Matthew Effect) works just as effectively in reverse, as any former MySpace or Friendster afficionado will tell you. Or AOL, for that matter.

Address of the bookmark: http://www.salon.com/2016/05/01/googles_new_media_apocalypse_how_the_search_giant_wants_to_accelerate_the_end_of_the_age_of_websites/

I quite like the word ‘reactions’ that Facebook is using to describe their new options to express feelings about a post. I wish I’d thought of it. This is a matter of much more than passing interest to me as it relates closely to something that occupied a lot of my time over some years of my life. In my own CoFIND social bookmarking system (that first saw the light of day about 18 years ago and underpinned my PhD work) I used to refer to something quite similar as ‘qualities’ – metadata (tags) to show not just that something is good or interesting but how it is good or interesting, that could then be used to rate and thus help to filter and rank a feed of bookmarked resources. CoFIND is an acronym – Collaborative Filter in N Dimensions – that refers to this n-dimensionality of ratings. Facebook’s Reactions feature is a simplified version of this: it’s about categories more than tags, but the thinking behind it is broadly similar. The differences, though, are interesting.

Fuzzy ratings

One of the things that is most notable about Facebook Reactions is that ratings are, like its Likes before them, binary: a simple ‘yes’ or ‘not-rated’.  In most versions of CoFIND (it iterated a lot), users could choose to what extent something was good/loved/annoying/interesting/etc through a Likert scale. Giving the option to choose the strength of a feeling seems much more sensible when talking about fuzzy values like this. I want to be able to signify that I quite like something, or that is is mildly amusing, especially if my intent is to communicate my feelings to others. Facebook’s Reactions are a coarse as a means of expression: it is quite appropriate that its emoticons are literal caricatures.  In all the methods I tried – radio buttons, clickable links, etc – introducing scalar ratings turned out to be way too complex to be usable, but web interfaces were not as rich in those days: I think things like popup draggable sliders (not dissimilar to Facebook’s interface) might make it more feasible nowadays.

Evolving metadata

Facebook Reactions are not just binary but fixed. CoFIND – I think, still uniquely – allowed individuals to create new qualities (reactions), which could then be used by anyone else. It was an n-dimensional rating system where ‘n’ could be any number at all. Qualities quite literally evolved for each community, with more used qualities surviving (being immediately available for use) and less used ones being relegated to backwaters of the system (effectively dying, albeit with the possibility of resurrection if added again). This allowed for such metadata to provide a mirror of the values that mattered most within a given community or network, rather than being imposed uniformly on everyone, and for those values to evolve as the community itself evolved. While I appreciate the simplicity of Facebook’s interface (CoFIND’s most fatal flaw was always that its interface was far too complex to be usable) I still think that user-created ways of emoting – what I have since called ‘fuzzy tags‘ – lead to much more useful reactions that matter within a given community, especially when users can choose the degree to which a fuzzy tag applies. When CoFIND was used in an educational setting, qualities like ‘good for beginners’, ‘authoritative’, or ‘comprehensive’ tended to emerge – they were pedagogical metadata. When used in other contexts, such as to discover what HCI students considered important in a website, site-ranking qualities like ‘slow’, ‘boring’, ‘artistic’ and ‘informative’ appeared.

Parcellation

One of the things I hate most vehemently about Facebook is that it same-ifies everything: a person in Facebook has a single unchanging (and permanently reified) identity, with a single network, a single facade, a single caricatured way of being in the world, notwithstanding the odd nod to diversity like pages and lists. Facebook’s business model relies on this, because any clustering or parcellation reduces the potential to connect, and connections are everything to Facebook. This makes me highly sceptical of its claimed ‘discovery’ that people are actually separated by only 3.57 degrees rather than six. Given that the system very deliberately drives them to friend as many others as much as possible, on most tenuous grounds of connection, this is hardly surprising. It shows not that previous studies are mistaken but the extent to which Facebook has manipulated human networks for profit. Apart from evolving to fit a single community, another of the things CoFIND did was to deliberately parcellate the environment, allowing different sets of values to evolve in different contexts. What is ‘good’ in the context of learning to read is not likely to be ‘good’ in the context of learning geometry, so different topics each evolved a (largely) separate set of qualities. This might not have been the best way to drive the growth of large networks, but it was a much better way to enable the self-organized emergence of meaningful communities. It also allowed individuals to express and embrace different facets of themselves, which in turn made it easy to accommodate changing needs and interests: essential in the context of learning, which is (if nothing else) about change.

You can read about the tortuous process of CoFIND’s development and the thinking behind it in my PhD thesis. I continued to develop CoFIND into the mid 2000s but, though the final version was a bit more usable and scalable (I rewrote it in PHP and changed a lot of the mechanisms, simplifying a fair number of things, including losing the fuzzy ratings) I’m still most fond of the final version that is described in the thesis.

Address of the bookmark: http://www.huffingtonpost.com/entry/facebook-reactions-update_us_56ccb128e4b0ec6725e42861?ir=Weird+News&section=us_weird-news&utm_hp_ref=weird-news

Interesting reflections from librarian Barbara Fister on Twitter’s precarious position, turning a ‘mere’ $191m profit on 300m users and not growing fast enough to assure survival. Turning away from the customer-as-product model, she instead makes a case for small, vertical market, paid-for social (but not too social) services, like Pinboard (tagline: ‘social bookmarking for introverts’) and LibraryThing (“MySpace for bookworms”).

I don’t see these as any more viable than Twitter. If successful, they will be purchased by one of the big few (giving the predator access to my data and social connections) or, if unsuccessful, they will eat my data and my social connections made with them. If they weather all that, the chances of them continuing to grow to meet my ever changing demands are minimal. Vanishingly few will continue to meet my needs for the foreseeable future and it would be foolhardy to bet money that any will survive indefinitely.

And indefinite survival is what is needed, not of the service itself but of the data, processes and social connections that support those data. It is bad enough having elderly files on my own computer that I cannot access at all thanks to proprietary formats. Having such data on a cloud service I may not even be able to access next week is much worse. And, of course, for a social system it is not just about my data but about the network with other people that, when the service dies or I wish to leave it, will be lost.  It’s better to be more of a customer and less of a product, but the endgame is the same either way. Cloud services are susceptible to a thousand and one woes, including unwanted service changes, renogatiated service contracts (seldom beneficial to the customer, once the data are locked in), acquisition, disappearance, failure to grow when needed, and a host of other things beyond my control. The proprietary, locked-in cloud is not the way to go, whether ‘free’ or directly paid-for. It gives the illusion of being open while actually being closed. We need open, portable, distributable and distributed standards for all of this: RSS, OPML, OpenSocial, trackbacks, pingbacks, WebMention and so on. And we need the ability to create, change, move and develop virtual spaces that belong to us, not to a service provider. This is why systems like Known, Elgg and WordPress (with appropriate plugins to support federation) are so important. I find it very encouraging that WordPress (open source and hosted versions) continues to grow and to dominate the social media landscape, powering over 26% of the entire Web. It makes usage of the likes of Facebook, Twitter and YouTube seem like a rounding error, and it is just one of many social systems owned by those that use them.

Address of the bookmark: https://www.insidehighered.com/blogs/library-babel-fish/when-300-million-active-users-isnt-enough

Compelling argument, 10 years old now, that Metcalfe’s Law and Reed’s Law are wrong, and that the correct value for a network should be n log (n). The reasoning is good: the problem with Metcalfe’s and Reed’s laws is that not all nodes in a network are equal. An analogy is made with Zipf’s Law (“if we order some large collection by size or popularity, the second element in the collection will be about half the measure of the first one, the third one will be about one-third the measure of the first one, and so on”) which reflects the uneven distribution of value in a network.

This makes sense to me, but could be taken further. It seems to me that there is no such thing as an ‘average’ network, so we must always examine the actual patterns in any given network to see what value individuals add, and we must always be prepared for some serious outliers that can greatly affect the overall network. If, say, a prime minister or president started to use the Landing, the effect would be quite spectacular (and likely catastrophic, for all sorts of technical and non-technical reasons). There are great risks in averaging things out and looking for statistical effects when observing any human system.

Address of the bookmark: http://spectrum.ieee.org/computing/networks/metcalfes-law-is-wrong