Re: Distributed Data

Kyle R. Rose (krose nospam at theory.lcs.mit.edu)
09 Mar 1999 16:13:35 -0500

> I totally agree on leaving out signatures. KISS is the name of the day
> to create a protocol that people will buy into.
>
> However: we still need a model for editing entries that arrive at the
> server. Or some other policy mechanism to keep the data clean and valid.
> I don't think we've seen anything like that proposed so far.

Here's one thing to consider: I was thinking about it last night, and
the first-come, first-served method of entering data may be the best
one after all, if we're trying to avoid trolls.

Why?

Because the hash function required to insert data into the database is
one-way. This means that a troll cannot reasonably be expected to
know the hash value of ANY CD that he does not possess -- even CDs
that won't exist for millenia! Since it is reasonable to expect that
only a person who likes the CD is going to rush out, buy it, and put
it into their DMI/CDDP+/whatever-enabled player, they aren't going to
enter crap into the database. (Of course, they may just be
brain-dead, stupid, or poor spellers, but I'm not sure there's a way
around this short of creating a trust system for "reliable" data
providers.)

For this reason, I would support a system that, by default, enables
users to add fields, but not change or delete existing ones. There
may actually be better systems, but I feel this is a strong argument
for the first-come, first-served approach.

Kyle

-- 
Kyle R. Rose                      "They can try to bind our arms,
Laboratory for Computer Science    But they cannot chain our minds
MIT NE43-309, 617-253-5883             or hearts..."
http://web.mit.edu/krr/www/                           Stratovarius
krose nospam at theory.lcs.mit.edu                              Forever Free