Ask HN: What's next in *social* Internet?

Social media seems to be a hot topic but with a lot of tension from two aspects:

(1) There is a lot of talk about developing and delivering more in social media. The more might be from things new in each of data sources, data manipulation techniques, Web sites, or companies. There might be more in just the content or in social search to find such content.

(2) There is not much clarity about just how to have more in social media or just why to have it. For the "why", what users want it, and what would they do with it?

Here are two examples of some of the recent talk about social media:

First, here on HN is the thread:

"Sergey Brin: We’ve Touched 1 Percent Of What Social Search Can Be (techcrunch.com)"

at

http://techcrunch.com/2011/01/20/sergey-brin-weve-touched-1-percent-of-what-social-search-can-be/

Second, is the thread "Building Better Social Graphs" at Fred Wilsons blog A VC* at:

http://www.avc.com/a_vc/2011/01/building-better-social-graphs.html#disqus_thread

with a lot of relatively good relevant comments.

So, let's dig in a little: We can start with being more clear about our terminology. There are three terms:

(1) Social Graph:

So, in applied math, the relevant definition of a graph is a collection of arcs and nodes with each node much like a geometric point and each arc a connection between two nodes (usually distinct but sometimes the same). An arc may be directed or not: A directed arc is drawn with an arrowhead and is regarded as a one-way street.

In a social graph, each node is likely a person (or maybe the blog of a person) but might be a group of people. The people might be users of Facebook, Twitter, etc. or just people who don't use the Internet. Then an arc might represent friend on Facebook, a follower on Twitter, or some such.

(2) Social Media:

Examples would include Facebook, Twitter, etc., maybe even HN. Maybe a definition of social media would be some Web site where people interact.

(3) Social Search:

Given data from social graphs and/or social media, one could do searches of that data.

To continue: There can be a problem with the concept of a social graph: Too commonly it is left unclear just what the arcs mean!

So, consider a person's social graph: They may have an arc to each person (A) they went to high school with, (B) live on the same street as, (C) went to college with, (D) dated, (E) married, (F) worked at the same company as, (G) hired to plow snow from their driveway, (H) got their business card at a Consumer Electronics Show in Las Vegas, etc. So, the point of these examples is that there is enormous variety on what the arcs can mean.

So, to make progress, maybe usually we should be more clear on what the arcs mean!

Now to something more substantive:

In Fred Wilson's thread "Building Better Social Graphs", there were two strikingly different themes:

First, Wilson started the thread with a post where he wanted to be able to download each of his social graphs and then curate them himself.

Second, in the comments, the theme was strong that given the data on the social graphs, we should have computer-based means to process the data for curation, etc. Curiously, the goal of this processing was Wilson's "Building Better Social Graphs" by stronger means than just Wilson's manual curation! That is, Wilson's title was stronger than Wilson's post, and the comments were closer to the title than the post was!

So, from 40,000 feet up, it appears that many people have some vague, ill-defined, intuitive, poorly identified and articulated visions of making progress with social graphs. Each of the broad subject of social, Facebook, Twitter, and the Internet, is so big that we should take the visions, as crude as they are, seriously.

So, there are a lot of people (e.g., 500+ million users of Facebook, etc.); each such person has one or more social graphs; somehow there should be some value in that data; we might process the data automatically to obtain some useful results; and the data might be good for doing related searching. Yes, for an ad supported Web site, the data might also be good for ad targeting!

Again, we should not miss the likely importance: From higher than 40,000 feet up, for people, networking has long been very important. E.g., about 120 years ago, some wealthy families built some large, expensive houses in Newport, RI. Why? So, in the summers the wives could hold big parties, network, and have their daughters meet good (i.e., mostly just wealthy) husband candidates! Sometimes even the husbands would get on a private railroad car in Manhattan, ride up to Newport, and show up at the party for a few minutes before going upstairs to read a book or play poker with the other bored husbands! E.g., in careers, long a common remark has been, "It's not what you know but who you know.". Well, the Internet is the biggest network of them all; it can in effect reduce geographic distances to zero; it has the power of computers and software to process data, and it might become much more important than anything before in "who you know", etc.

Descending a little from 40,000 feet, and thinking about what software we might write, we can identify three important issues:

(1) Meaning.

An arc in a graph from the definition in applied math has essentially no meaning in any sense social or even practical. So, if we are to make use of data from social graphs, etc., then we should make some progress, if only rough, on what the arcs, or other data, mean.

(2) Purpose.

We should identify the purpose of the software. That is, what will be the output of the software, and why will users like that output? Or, what do users want, or at least would like if they saw it, that such software might provide? What the heck is the darned purpose?

We might start by articulating just what is the purpose of Facebook, Twitter, etc.! The original purpose of Facebook at Harvard was to get dates. For Twitter, maybe the purpose is to follow selected other people and, maybe, ingratiate oneself.

(3) Techniques.

Given the data, what data manipulation techniques will we have the software use to get the results good for the purposes?

I raise one more point:

The US has something over 300 million people. In some important respects, this number is not very large. E.g., it is easy enough for current computing and data base techniques to have, say, 1 million bytes on each person and still be able to store and process that data.

So, it can appear that there is a chance that we could have a single, grand solution in the space of social graphs and social search. If so, then we will guess that the present efforts in social media are only tangential or indirect solutions for a central problem not yet identified, articulated, or solved and that a single, grand solution might be possible.

So, consider roughly 1 million bytes on each person in the US. It turns out, although not discussed very openly, actually there are data bases that have a lot of data on nearly every adult in the US. So at least in principle, there is a lot of data now. Keeping this data restricted in walled gardens forever may be unreasonable.

Here is a simple example: Just take the printed phone books, type in the data (or get a DVD from someone), and sort the data on state, city, and street address. Then, given a person and their street address, a simple data base query will yield the names and telephone numbers of all of that person's neighbors. In particular, such a telephone book can provide a case of a social graph for each person in the book.

When this person makes a telephone call, the telephone records provide another social graph for this person. When this person uses a credit card, we get another social graph. Each merchant who accepts a credit card defines a social graph.

Net, our society is awash in data for social graphs from sources much more general than Facebook, Twitter, etc.

So, generally we can guess that we can approach asymptotically 1 million bytes of data on each person in the US where this data provides a fairly complete description of that person.

So, we have potentially a grand answer to the issue what data.

Then we can move on to what purposes? What will people what to do with this data?

Okay, in some broad sense, many of the uses will be introductions. Or, "I see that you also bought a Golden Retriever puppy at the North Hills Mall Puppy Store and also have a son in the third grade at North Side Grade School." Uh, nearly anyone else could devise better purpose scenarios that that!

So, for introductions, there already is an industry, that is, romantic matchmaking. So, introduction can be for more purposes than just romantic, and likely both the data and the processing techniques used could be quite similar.

So, from 40,000 feet up, the purpose of social anything and of social search may be introductions that could be accomplished by starting with available data and processing it much as in current romantic matchmaking services.

Yes, the introductions could be for many purposes -- hobbies, careers, looking for a supplier, customer, employer, employee, date, spouse, cello in a string quartet, etc.

At times in the comments on the Fred Wilson thread, it appeared that one of the purposes was just the usual one of some women just wanting to meet with, get to know, and gossip with as many other people as possible, for no definite reason! That is, we have to accept that one of the purposes may be just to 'meet' people for no other definite purpose!

So, where am I going wrong?

What more is there to be said?

Where can we be more clear on the data, purposes, processing, and future of social whatever via the Internet?