> From Brad Allen <[EMAIL PROTECTED]>
>There are a number of reasons why Timm Murray is wrong about
>permanence. His reasons are about as well-thought out as crackpot
>hubris, which makes me upset when looking at new protocols.
Happy troll to you, too.
>* USENET: it is a problem that USENET looses its messages.
> Timm says that "nobody complains". This is patently false.
> Everybody complains about it. It is a large problem. The main
> problem is that discussions never get more than about two weeks old,
> thus capping the inteligence and usefulness of USENET severely into
> the potty-training years of a young child. Another big problem
> which is similar but different is the issue that people who stick
> around get tired of the repeats. USENET is severely limited by its
> lack of permanence. When people select USENET servers, the top four
> qualities people look for are reliability, speed, which groups it
> gets, and how long the messages stay; usually the last is the most
> important.
Freenet does not remove documents based on the time since they were inserted,
the way usenet does, so the problem with a fixed cap on the length of
discussion is less likely to be an issue.
>* There needs to be a seamless method to access all data, regardless
> of purpose or publicness. If freenet is just for psychotic geeks
> who are doing strange paranoid things, then it won't go far. If
> it on the other hand attempts to be a data store for every
> conceivable type and use of data, then it can really flourish.
> It is especially important today to make new data systems appease
> everybody, or as large an important subset of that as possible.
> This goes for any data communications and storage system, not just
> Freenet.
I'm not sure what you mean here.
>Permanence on freenet could be obtained in a number of ways:
>
>* Marking a value of permanence on an object. It would be weighed
> against others.
Who would decide what the permanence value of a document would be?
>* Marking that usage counting should be done on an object, and then
> counting it. (The usage data could be spread ala USENET -- to
> neighbors who also have it. I have not read the standard, so I
> don't exactly know if this will spread in all directions.)
Freenet already prioritizes documents based on their popularity, although
exactly how to do so is still up for debate...
>* People could agree to store objects in shared lists of reviewed
> objects according to whether the review says to save it, and how
> important it is. So, for instance, if I make a review list with me
> and a few other friends, anything we review will be deleted in
> reverse order of its reviewed value. So, for instance, using my
> reviewing scheme that I share with my friends, SPAM is reviewed at
> about level .1, and reasonable messages are at about level .7. If
> the computer needs space, it will start by deleting the SPAM.
>* Storing the file locally forever. If you want something to last
> forever, you save it.
>* People who agree to save the same object forever or for a similar
> amount of time can group up. Multiple copies (for redundancy) must
> be saved. When people want to leave the group, the objects will be
> redistributed.
The problem with all three of these methods is that they are not compatible
with Freenet's routing. In freenet, a node does not decide which set of
documents it is to store, rather, most of the requests it recieves will be: a)
from a fairly small subset of all the documents on Freenet (it's keyspace) and
b) a very small fraction of all the requests on Freenet. If I decide to store
a given document forever, it will not be available to anyone but myself,
unless that document "belongs" in my node (not very likely) and will be
requested by others looking for it.
>The concept of unreliable communications being a stable medium of
>discourse is untenable to me. For a solid communications medium,
>discussions have to get past a certain foundational state in order to
>have value. Often, in a system with rapid deletion, the foundations
>to knowledge aren't even laid down before the destruction has already
>ruined any chance of forwarding that knowledge. This is
>unacceptable.
So speech is of no value?
>
>Brad Allen <[EMAIL PROTECTED]>
>
>P.S., calling me a newbie will not change the above facts.
True, but it's a nice feeling. Newbie! Newbie! Newbie! Ahh, much better.
>A middle ground must be made
>where the message gets widely enough distributed that a deletion will
>not hurt, but not so widely distributed that it will overburdon the
>system (such as in USENET's traditional method of distribution).
The lifetime of a document in Freenet is a function of it's popularity and the
amount of storage space available in the network. With sufficent storage
space, any document will last as long as it has a non-trivial number of
requests over time.
--
Benjamin Coates
_______________________________________________
freenet-tech mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/tech
