Ming the Mechanic
The NewsLog of Flemming Funch

Saturday, July 10, 2010day link 

 Strong Elastic Links
picture There's something fundamentally messed up about the way we store and use information. Most of our information connects really badly with related information, and with the stuff the information is about.

I've talked about that before, like here: Connected Information, so I'll try not to repeat myself. It is however, somewhat difficult to convey my point. I've tried writing and rewriting this as an article a couple of times, but left it unfinished. It still isn't coming out very clear, but I'll leave it at that.

I want information to be linked, by unbreakable elastic links, to what the information is about.

The type of links we know on the web are useful, way more useful than no links. But they're but a pathetic shadow of the type of links we potentially could have that truly would be useful and reliable.

I'm sure it is not only me who have found some interesting article on the net or in a magazine about something new and promising. Say, self-driving cars or super-efficient solar cells. And then, months later, when I try to search for information about how that project might be going now, there's no trace of it. Some journalist did some kind of investigative job and wrote about something. On the web it might even include some clickable links to more information, like another article or a company website. When I come back some months later, those might still be there, or they might not. It is quite likely those links would point to some frozen information from that same time period. What happened later might remain a mystery, unless I have the time and resources to do a fresh piece of detective work.

The links we use on the web are like addresses on an envelope that we put in a mailbox. They indicate some kind of coordinates for a recipient. "He's over there!" But he might not be. The address might have changed and become invalid, or it might now be occupied by somebody else who has no relation to the person I'm trying to reach. The links don't follow the target when it moves. Likewise, web links aren't very good at linking up real people or real subjects.

Part of the problem is that the web links are one-way pointers. They just point in the direction of some virtual place. That place doesn't easily know that they're being linked to, because there's no link the other way. So, even if they wanted to, they couldn't easily update others on the status of what they linked to. Even if they could, it would still be a cumbersome thing to do.

Links shouldn't just be some address. They should actually link the two things.

The reason you have problems with spam is because the contents of the e-mail messages you receive don't really link up with anything. There's an address for the sender and the recipient, and addresses for servers that have processed the e-mail. All of that can be arbitrarily made up by anybody, because the e-mail doesn't actually link to the sender and the recipient. It can say all sorts of stuff that isn't at all true, or it can say things that were true at some point, but which go out of date later.

Imagine that you could attach a link to something, and that link, without a doubt, would maintain the connection, no matter what.

For the moment, never mind how it could be done, but imagine that between all people, all groups, all subjects and all media about any of these things, between all of those there would be unbreakable links. Hard links, so to speak, or strong links, but elastic, as they will "stretch" to any length no matter how the nodes move around and transform.

You probably know what school you went to in a certain year. That school is a rather finite entity. It should not be a matter of archaeological detective work to retrieve the information of who the principal was, and what became of any of the teachers or any one of the students. The school was an unmistakable entity. It was there, very physically, it had buildings, it was paid for, it stayed there for a long time. The same with all the people who were there. Every single one was unmistakably a real, living, breathing person. There's really nothing fuzzy about it at all. But in accordance with the way we typically treat information, it has been saved in a very fuzzy manner. If you go search for your school in search engines, there is likely to be some doubt about what school you're talking about, and whether it even exists. It is going to be very hard to locate a list of teachers or a complete list of students, if one exists. The information was kept on pieces of paper, which might have been mislaid or lost or falsified, and maybe never digitized. Even if you found the list, you wouldn't know if it was the right one, and even if you did, it is only a list of names and maybe addresses and maybe a photo. Most of these people have moved, many of them have changed their names, some have died, etc. It would be a huge amount of work to track them down, and you'd probably have to give up on quite a few of them.

We've gotten so used to sloppy, unlinked information that we find it quite natural and normal that information gets lost or that it is hard to reconstruct or that nobody knows if it is true or not. We even find a certain comfort and security in all this fuzziness. There's no government that is sure how many people there are in the country it governs. And that's despite that they really want taxes from all of them, and they don't want illegal immigrants, and everybody needs an ID. And the subject matter, persons, is in no way vague. It isn't difficult to decide if somebody is a person or not. They're very finite and the number of people is finite.

The moment you commit information to little bits of paper and sloppy handwriting and filing cabinets and vague references to other storage places, the game is lost. The link between the information and what it is about is no longer there. It isn't much better if the same system is simulated with computers. Useful information can often be reconstructed, but there's nothing that guarantees that.

In the electronic world, we should by now be able to do much better. There's absolutely no reason to store our information in the same sloppy manner, lists of names and addresses in files that can be lost and falsified, or, worse, in free unstructured text form that also is stored in fairly random places, without real links to the subject matter.

What I'm asking for is, in part, two-way links, as one can pull the string from either side. But it is also unbreakable links, not just pointers. Not just signs that point in the general direction of the other piece of information. Rather, something like an electrical wire. The moment somebody cuts it, an alarm goes off. Or a quantum entanglement kind of mechanism, where you just can't mess with it without it being noticed.

How can one practically implement it? I didn't say I knew how, just that I want it, and that everything we do with information would totally change if we had reliable links. But it is not like it is an unsolvable problem. It would in no way be impossible to provide each living person with a unique encrypted ID code. There are certainly issues of politics and of privacy, and of identity theft, but they could be solved if there were any unified wish to have unique IDs. As it is now, it is in most places a no-brainer to acquire multiple ID numbers or to disappear to somewhere else.

The same applies to things, places, organized groups, subjects, etc. Information is as sloppily kept as for people, or more. A car at least has an ID number, but it is only used by government agencies, not for recording your photos or car trips or anything else.

A lot of stuff might deserve being very loosely joined, but not facts. A piece of information that is or could be a fact when recorded shouldn't later be a matter of searching and guessing. You should know its level of correctness by the way it is linked, not by some forensic text analysis.

Our shared information system has Alzheimer's. Real events instantly get converted into vague guesswork and conjecture and interpretation and stories and remixed soundbites. And then we expect to pour all of that stuff together, have a machine sort it all, and then we'll discover how really smart we are?

We'd probably get somewhere faster if we at least could keep most of the objective stuff straight, and then we could use our imagination and reasoning abilities for more important stuff than merely trying to reconstruct what is going on.
[ | 2010-07-10 13:01 | 159 comments | PermaLink ]  More >

Main Page: ming.tv