In the last post I have stated that random lists are boundary objects: They include a low level of list-ness, and can therefore often better be described as sets or groups (I used these two terms quite carelessly as interchangeable - I will stay with the term 'set'), which have a certain degree of list-ness. Now I want to look into this a bit more thoroughly. To look at boundary objects such as random lists can teach me more about lists in general.
My question is: What is the random in random lists? I pose that it can be threefold.
The first kind of randomness is given, when a list contains randomly selected elements. What we can see here is that all lists are sets (but they are of course also more than sets, as I will develop later). In a way you could say: Such a list/set is based on no selection criteria. However, to say this is true and not true. It is only true in the sense that there are little logic criteria for inclusion: It is not based on significant traits of the elements.
So why is it untrue? A list based on random selection still contains elements (otherwise it would be no list/set at all). The element happened to be included in a moment of random selection. The criterion is the random inclusion of the elements in a past event. To take an obvious example: A, B and C can be part of a list/set, because Joe decided by drawing lots that A, B and C are members of his team. But even if in such a situation, such a list/set is still defined by this random decision. As soon as Joe has made it, the inclusion in this event becomes a character trait of the elements of this list/set: This is the list of members of the team, about which Joe has made a decision through drawing lots.
It is also not true, because a total randomness of selection is hardly ever given. Even if you choose 3 random objects, you normally choose them from objects of a kind (three randomly selected videos out of a pool of videos, three football players out of a pool of football players, for example). So there was a larger pre-selection, and then a random selection out of this larger set.
What can we learn from this? Two things. Firstly, randomness is already relative on the level of the list as set. Yes, a randomly selected list includes randomness, but no, it is not fully random. Secondly, the inclusion in any list/set does something to its elements. It ads a trait to them, at least for the time as they are included in this list/set in a context, where such a membership is of relevance.
Now to the second form of randomness. This form starts at the other end. List can be of random order. The position of the element is not determined by some of its former traits. It is contingent. This is a totally different form of randomness. Indeed, many lists of random order are not lists of random selection.
However, just as in the first case, it is once again true and untrue, if you say that a list has a random order. If the order is totally random, it is not a list, but only a set. I have put forward in the last post the definition of lists as a form of assemblage, whose internal order can be potentially expressed by ordinal numbers. This means: Even a list of random order still orders its elements along a line. Even a non-signifying order is an order. But of course it is an order of lesser importance. This is why a list of random order is on second level a list, and on a first level a group or set. Inside the list, there is no escaping of this order. On this line, each element has it precise place. This place is always defined through two other elements: One is before, the one is after, and the element is in between.
If you look at what I just said a bit more closely, you will notice that I was arguing quite imprecise. To increment more accuracy into my argument, I have to introduce a further division: element versus position. A position is indeed part of the form of the list. A position is the specific form that multiplicity takes in lists. An element finds it place on a position. A position A is defined by its relationship to one or two other positions. If the position A is in the middle of the list, it is defined by two other positions, if the position is at the beginning or the end of the list, there is only one other position that defines it (and I will look at what else there is needed to define this later). It is also indirectly defined by other elements of the multiplicity: It makes a difference whether I am placed on a short list or a long list or an open ended list. Once again we can see, that a list ads traits to its elements.
Lets finally look at the third kind of randomness in random list: Lists, which express randomness. Their message is: ‘This might look like a list, and therefore imply some kind of significant order, but in fact this is not the case, in fact this is a total random order’. They try to break with the effect that all elements get ordered simply by the inclusion in the list. They try to undermine the ordinal numbers, which are added by any list.
The interesting thing now is, that lists can only do that, if they suggest a different order. The most prominent form how we do this is the alphabetically ordered list. An alphabetical order means: This is no ordinal order of significance. Take a phone book: It orders all owners of phone numbers after each other, and can thus be defined as a list that is based on ordinal numbers. However, the elements are placed on this list by a second order: The order of the alphabet. This is done to enable us to find a name quickly. But it is also done to express: This order is otherwise totally arbitrary. We therefore use alphabetical lists in general, if we want to say: This order means nothing. It is only a tool to find its elements quickly.
If you now look at such alphabetically ordered lists more closely, you soon see, that the alphabet is in this case just another form of expressing ordinal numbers. It is not based on the "Base ten" (0-9), but on 27 graphemes (this is of course the Latin alphabet, and there are even here further regional differences). "Ice" is turned into "Eighth-Third-Fifth". A position through alphabetical order is another form of expressing a position of ordinal numbers.
But the difference is of course crucial. A position based on the alphabet is one that is based on an insignificant or random trait of the element: The form of the signifier is arbitrary (at least in most cases). So there we are: An alphabetical list imposes an order on its elements, to make clear, that there are no other criteria of order at play.
The significance of this move becomes clear, if we look at another form, how we sometimes try to express random order of lists: The bullet point. The bullet point says: Each of these elements is separate, and because they are not numbered, they are neither ranked nor otherwise ordered. However, this is a much weaker form of breaking the order. It hardly ever succeeds fully, because t does not impose an alternative order. It just say there is none, which we then are inclined not to believe, and rightly so. I will look at the bullet point more closely, when I write about the relationship of lists and text.
To sum up: There are three kinds of randomness in random lists: Firstly the randomness of selection. Secondly: The random placement of elements on its internal order. Thirdly: Orders that express randomness. These three forms of randomness do not have to occur at the same time and all are random only to a certain degree.
In the next post I will look at open and closed lists and thus, amongst other things, at the relationship of lists to infinity.
What distinguishes lists form other forms of assemblages? Well, to start with a pretty simple observation: Lists are pretty simple forms of assemblage. They are nothing near the complexity of an organism, not even of a network. They have a certain degree of systemic quality: If one element enters or changes its place, it can effect to a certain degree other elements placed afterwards. But once again, this is rather simple, nothing compared to the systemic quality of an organism. Indeed, one of the most important traits of lists is probably their simplicity, thus their possibility to reduce complexity. They order all their elements along a line.
So what kind of line is it? Firstly, it is a linear line, without loops or circles. Lists can fork (and I will come back to that later), or structure themselves into parts along a line, mostly through sub-headlines, but they remain linear. Secondly, lists always have a start. This is why lists have such a close relation to ordinal numbers (= first, second, third …). Thirdly, lists have multiple possibilities to end. Sometimes they have a clearly defined end, sometimes not (in the latter case, they end, for example, with "etc..."). Fourthly they create some kind of order along the line. This means fifthly: In most lists, every item occurs only once.
One way to represent a list is to give its elements ordinal numbers. A list does not need to be numbered to be a list, but its internal order has to be of a kind that can be expressed by ordinal numbers A list can thus be defined as an assemblage, whose internal order is potentially expressed by ordinal numbers. Ordinal numbers have in such cases the function of a secondary list, which expresses the "list-ness" of the primary list. I will come back to this later.
Once again, I would like to reformulate this in relative terms: The more the internal order of an assemblage can be expressed in ordinal numbers, the more it becomes a list. Random lists are thus lists with a low degree of list-ness. I would argue that random lists are indeed primarily sets or groups, and lists only on a second level; all other lists are indeed primarily lists, but groups or sets on a second level.
But how about networks? Is a list only a specific linear form of network? A network that is stretched and reduced to a chain of relations ordered in one line? My answer would be: No. The key difference between a list and such a super-flat network is: The relations in the network are still "personal". They connect one particular element with one other particular element. A list is "non-personal". The elements are connected to each other though the inclusion and the specific place on the list, but not through relationships in the network sense. Yes, the two elements before and after define the place of the element in between. They might also ad meaning to it. Indeed, in some lists one element can be defined by all other elements (such as in lists ordered by internal algorithms, where elements exactly define the other elements). But it is never "personal" in the sense of a relation, as it occurs in a network.
So we see: A central problem of lists is the nature of their internal order. The next post will address this in more detail.
Nico Nico Douga’s ranking is one form of ranking. Rankings are one form of lists. So what is a list? What looks like a simple question turned out to be a rather complicated one. To develop a sytsematical approach, I moved up even one branch higher in a tree hierarchy of concepts: Lists are one form of assemblage. Then what is an assemblage? Oh dear, I think I cannot move up even further. So I take assemblage as a starting point, even though it sounds a bit Deleuzian. Leaving Deleuze to the Deleuzians, I want to come up with a very simple and non-Deleuzian definition: An assemblage is a plurarity of elements.
Let’s look at some examples. In a network the elements are connected to each other via relations. A group would be another example: Here, the elements are placed inside a boundary. A system is a more complex form of assemblage: All elements influence each other. Some forms of assemblages are even more complex: An organism would be one example, a machine another one, an organisation a further one. In such forms of assemblage the elements need to cooperate to a certain degree.
Different forms of assemblage are not mutually exclusive. One element can be, for example, part of many groups, not only one. Not only that. A group can also be at the same time a network. An organism can be network, an organisation can be a group, and so on. However, very often you have a situation, where one form of assemblage is dominant. It can make sense to look at an organisation as a group (for example if you are interested in corporate identity), but if you do so exclusively, and look at an organisation as if it would be only a group, you miss the essential trait. The same applies to all forms of assemblage: They are mostly one thing – a network, a group, or an organisation -, but they have often traits of other forms of assemblage incorporated and can thus be also looked at from a different point of view.
This is all very abstract. So let`s take one example: Tags, for example, create assemblages (it is not their only function, but it is one). I would argue: On Nico Nico Douga, their main basic function as assemblage is to create groups of videos (this differentiates Nico Nico Douga from other video platforms). Groups are often organised in complex overlapping landscapes. One thing is an element in many different groups. To take the example of tags: One video can have many tags. But tags are not only groups. They are also part of networks: tags have relations to each other. This means: When you look at the tags on Nico Nico Douga, you learn that group and network are not mutually exclusive forms of assemblage. They can overlay each other, and can exist in complex forms of mutual entanglement.
You can look at this complex forms of mixes also in a different way. Assemblages are often a specific form of assemblage to a certain degree. One particular example is here the system. Most assemblages have a certain degree of systemic quality: All elements more or less influence each other. Indeed, I was tempted to use the word “system” instead of “assemblage” as starting point, but as I personally think that not all assemblage are to the same degree systemic in the above described sense (mutual influence of elements), even though most are.
With this groundwork I can now look at lists as a linear form of assemblage. This definition sounds simple, but it does not stop here. In the next posts I will look at what kind of linear form of assemblage lists are.
The British media theorist and sociologist Adrian McKenzie has recently sketched out his idea about the Internet as a medium of lists. I found his talk on a workshop at Goldsmiths highly inspiring. It struck me that rankings in particular, and lists in general are indeed one of the major structuring principles of the Internet. As soon as you look for lists and rankings, you find them everywhere: Google’s search results, YouTube’s rankings, Wikipedia’s lists of alternative meanings, online gaming results, menus and drop down menus, mailing lists, discussion lists and bulletin boards, internet archives, link lists, tag lists and bookmark lists, contact and address lists, lists of friends on social networking sites, rankings of personal tastes (such as my favourite films), collided lists of taste such as Metacritic.com, and specialised ranking portals such as http://ranqit.com/… Rankings and lists, wherever you look.
This means: Lists are probably the most prominent form of representation on the Internet. And I am not talking here about “mere representation”. Think of Google’s power, and you know: Lists have consequences. In fact, there is a case to make that lists are influential on deeper levels as well. Much computer code is organised in lists of some sorts: sequences of statements, which are machine-, and often also human-readable (not for me, though). Computer memory, I am told, is organised by lists, where to find what (and much of the processing power on a personal computer is used for search on these lists). Metadata is usually organised in ontologies, which are in turns organised as hierarchical trees, and such trees are in fact nothing else then multi-levelled lists. Most metadata takes the form of lists.
In the recent 20 years of Internet theory, the concepts of list and ranking never got the attention they deserved. As soon as you get obsessed about the question of lists and rankings, you can read much media theory in a new light. Lev Manovich, for example, puts in his “Language of new Media” (2001) the database at the centre of his approach. In this context, he briefly writes about hierarchical and object-oriented database structures. According to him, hierarchical databases have tree-like structures and object-oriented databases “are organised in hierarchical classes that may inherit properties from higher classes in the chain” (p 228). As I will argue later, both forms of databases are in fact complex arrangements of lists. Manovich does not follow this lead. He is interested in the relationship of database and narrative (and argues that in new media, narratives become one special form of database). But in fact much of what he says has an even deeper root in the origins of most databases in the notion of list.
It is thus an interesting move that Matt Fuller included in his alphabetically arranged (!) anthology “Software studies" (2008) an article about “Lists". Here, Alison Adam traces the origins of lists back to the origins of writing: “Arguably, it is the business of recording lists, which marks our literate societies form pre-literate societies” (p. 174). For Adam, lists have a double character. On the one hand, they can be “a way of sanitizing and simplifying knowledge” (p. 175). No wonder, that we can describe not only the origins of modern bureaucracy, but also of modern science as an application of lists on a huge scale: “Through lists we order and control ourselves and the world we inhabit”. On the other hand, Adam points the out the principal openness of lists: Almost always we can add further elements, and a list “doesn’t have to impose a single mode of ordering”. It seems as if for Adam the first side of lists is bad, the second good, even though she does not say so explicitly: “The elasticity of the list, its capacity to surprise, means that LISP (a programming language based on lists - gb) resists the obvious Taylorisation that one might expect with such a powerful ordering tool" (p 177).
Adam’s article opens up a rich tradition of theorising lists in philosophy and sociology: From Foucault and Latour to Bowker/Star and Law/Mol. Even more surprising that these traditions seem to not have led to a theory of the internet as a medium of lists. Maybe it is exactly the old age of lists, which let us overlook their central role on the Internet: Lists were there since the origins of writing, and maybe even before, so why should we theorise them now? Closely related to this is a second reason: We were so impressed by the fact that the Internet is from a technical perspective a network that we looked mainly for network structures (and this, of course, had to be the first approach). A third reason can be found in the vague political implications of concepts such as lists or networks. "Network" sounds horizontal, democratic, open, participatory, even though this is not anymore the case, as soon as you take a closer look at them. List and rankings, in opposite, have a hierarchical taste. They are somehow linear, almost square, not as fuzzy and chaotic as networks. (Adam develops a more complex idea of lists, but she also relates their ability to control and order with bad, and their openness and elasticity with good, which is, I would argue, a rather simplistic view of a very complex relationship).
You can find such vague political implications in many forms. Not all are based in critical research. An annoying example for an apologetic version is the naïve form of Internet euphoria under the slogan “Web 2.0”. This aged buzzword collides a range of different internet phenomena (blogs, wikipedia, video communities, tags, social bookmarking, social networks, and so on), thinks them through under the paradigm of (social) networking, and values them as democratic, participatory, etc, which leads, of course, to some big mistakes. I am deeply convinced: Social networking should not be the paradigm for the rest of the Internet. I would argue that most phenomena cannot and should not be understood under this paradigm. And Nico Nico Douga is its most spectacular example.
But you also find such vague political implications in such forms of critique as Geert Lovink’s, who rightly so questions the concept web 2.0. In an interview, which he gave Die Zeit in 2007 to promote his book “Zero comments”, Lovink mentions the blogger’s obsession with rankings and lists, and finds this rather depressing. Rankings and lists have become, according to him, a technical mirror: We prove that we exist via Google. We exist through statistics. The concept of lists and ranking serves him to undermine the naïve web 2.0 discourse: Because web 2.0-ish applications are in reality based on rankings, they are not as democratic as they think they are. And yes, web 2.0 should be critiqued in this way, and lists and rankings should and can be means and objects of such critique. But just as networks are in itself nothing good, lists are in itself nothing bad. They are one form, how elements can relate to each other. And they can have a productive role, just as networks have. This includes their ability to order and simplify just as their openness.
When you look at Nico Nico Douga, you can learn about other functions of lists and ranking. Lists and ranking are, for example, one of the ways, how temporality is organised on the Internet. They are at the centre of the Internet’s public sphere. For me, they are one central corner stone for a theory of liveness, which I am working on these days (I hope to post more about this at some point here on this blog). If you look at the Internet as a medium ruled by metadata (which is, surprise, surprise, the approach of the Metadata project), lists and rankings move right into the centre of the Internet. They are a place, where power is centralised and exercised, but once again, this can be good or bad. And even more so, it can be contested, as we can see in the example of the recent ranking revolutions on Nico Nico Douga. The Nico Chuu did not fight to abolish rankings. They fought for a particular form, how rankings should be organised in their opinion.
To sum up: Lists and rankings are an undervalued dimension of the Internet, and they should be understood from scratch, without an approach that mingles aesthetic objections too easily with political ones. In the next post I develop a (Non-Deleuzian) idea of assemblage, which will be the base camp for what is to come. <gb>
August 20(Wed), 2008 Ranking (part 2): Nico Nico Douga’s recent ranking revolution
Like most video community websites, Nico Nico Douga does not have one single form, but a plurarity of rankings. The most important form is called, tellingly, “ranking”, but it is not the only one. The “ranking” button is placed in the upper right corner of the start page. When you click on it, you get a list of 100 videos, sorted by popularity. This lit underwent changes in the last week. Until some weeks ago, it was reset daily, and updated hourly. This meant: Every morning at 5 AM the list started anew, and it changed through the course of the day. The ranking related to all videos (and not only to videos of a particular genre) and was based on how many users have bookmarked the video during that particular day. Other forms of ranking were available through drop down menus and tags (I will describe these later in more detail). But the daily ranking was set default. It was clearly the most prominent for the Nico Chuu. And it was highly effective.
However, the programmers and content managers of Niwango were worried about the high turn over in the daily ranking. The daily reset on the default ranking introduced, in their opinion, too much newness. Popular videos became popular very fast, but just as fast they dropped of the ranking. Not all Nico Chuu watch Nico Nico Douga every day. Lighter users tend to miss important videos, so the programmers and content managers thought. As a consequence of such reasoning, a new form of ranking was introduced. Niwango made some fundamental changes: Firstly, they set the weekly ranking as the first default. Secondly, they added (only in the Japaneses version) on the right-top corner of the rankings site another button: "Set as default". This means: You can set a specific form of ranking as your own personal default. Thirdly, they stopped to update the daily and weekly ranking – what you now get, is basically yesterdays, or last week’s ranking. Fourthly, they introduced a new, additional form of ranking: The hot list. It is not anymore based on bookmarks. It is based on the videos that users are playing right now (to be more precise: in the last 10 minutes).
Additional to these fundamental changes, they also re-designed the interface. Formerly, alternative ranking were accessible in a drop down menu (bookmarks by the day, bookmarks by the month etc…, then views by the day, views by the month, etc…). After the re-organisation the interface of the default ranking shows eight buttons, placed next to each other: (1) “Day”, (2)“Week”, (3)“Month”, (4) “All”, (5) “Bookmarks”, (6) “Views”, (7)“Comments” and (8) “Hot list”. The items (1) to (4) are set alternatively, as are the items (5) to (7). The item (8.) is separate. As default you arrive on the item (2) and (5). With one click you can change either the rhythm (1-4), the category on which the raking is based (5-7), or you can directly move to “hot list”. With the additional button "set as default" in the upper right corner you can furthermore change these settings, and start, for example, with the items (3) and (6).
The purpose of this new form of ranking is clear: The changes in the interface make the ranking more usable. The changes in the update function and the fact that the weekly ranking is automatically suggested as default aim to decrease the turnover in the ranking. The button "Set as default" introduces a new form of user-controlled personalisation. And with "hot list", the programmers suggest a new form of semi-synchronous collective experience.
What then happened, took Niwango by surprise. A storm of protest broke loose. The Nico Chuu protested on all channels. They used the Nico Nico Douga bulletin board, Twitter, Mixi, Blogs and 2channel. Some Nico Chuu coined their anger in hard terms: “Nico Nico Douga ended” was one of the comments. Others threatened to stop their premium accounts (=they would stop to be paying subscribers). The almost uniso thrift of the protest was: Get the old ranking back! Two arguments were used most often: Firstly, a weekly ranking will lessen the chance of new videos to get attention. Secondly, a daily update does not provide the Nico Chuu with enough new videos. Especially for heavy users, the ranking is not anymore fluid enough.
Niwango reacted quickly. It introduced a further alternative form of ranking: The hourly ranking. This ranking is not only updated, but also reset hourly. Such a ranking did not exist before at all. It is faster then the former daily ranking (which also was updated hourly, but collided the data during the whole day). It is slower then the hot list, but based on the bookmarks, and not, as the hotlist, on views. The category “hour” was introduced on the new interface – it is thus only one direct click away from the ranking, which is set as default. The default ranking remains weekly reset and weekly updated. This is a compromise of sorts: The deep Nico Chuu get their hourly ranking, but not a daily ranking which is updated hourly, and the weekly ranking is still set as default.
I am not sure whether this compromise is the ideal solution. I will give some reasons for this later in this text (in part 20). Keeping my role as a researcher (even though I have also become a Nico Chuu of sorts, with the according passions), I want to refrain from quick opinions. Before I give my analysis of this latest version of ranking on nico Nico Douga in part 20, I will have a closer look at what lists and rankings are in the parts 3-19. I want to start in part 3 with a look at some of the media theoretical literature and ask, why rankings and lists have drawn so little theoretical attention. This will be the topic of the next post. <gb>