I'll take a look at your notes soon. I have something similar to
share, not related to mechanisms or rules for sharing, but rather the
issue of part sourcing from starting with near nothing.
Equipment acquisition for homebrewers
http://lists.puremagic.com/pipermail/robotgroup/2008-June/009660.html
Also, we previously discussed a few topics that are similar:
An internet of packages of physical things using standard small containers
http://groups.google.com/group/openmanufacturing/browse_frm/thread/810303ae87fa1b2d#
Delivery system or something using DARPA-contest guided robots for
milk carton delivery
http://groups.google.com/group/openmanufacturing/browse_frm/thread/66c29329c67d6bfb/f93eb1261d13613f?lnk=gst&q=cartons#f93eb1261d13613f
Materials for free stuff
http://groups.google.com/group/openmanufacturing/browse_frm/thread/1ccac1fa29e05e30#
> - to facilitate the sourcing of parts and materials for free and open-source hardware
> projects
> - comparison to the FreeCycle network ( http://www.freecycle.org/ )
So, part of the problem with freecycle is that people don't know what
to call the things that they put up. Heck, this is a problem with ebay
and other auction sites, or anything that isn't from the original
suppliers: not everyone is going to bother to scan in the bar codes,
or figure out the proper ID numbers on all of their parts and stuff
they're putting up online. Meh.
I like your thinking re: tie-ins to other open source
geography/mapping topics. But I think that the issue of "what to call
stuff" is a really big one, almost a show-stopper, but other than
that, it sounds like you're thinking along the right track.
But then some questions start to come up-
* who is in control of the main ISBN-like registry? (much like the
issue of ICANN being in control of domain names for the name servers
that computers are defaulty configured to connect to)
* who gets to add/remove, edit, modify, revoke and otherwise work with
that "universal name" registry?
Granted, these issues aren't too big of a deal. For instance, maybe it
would be a built in feature that the entire index is exportable to for
any user to download (to be installed alongside the repository
management tools), and then it's thus forkable if anyone wants to take
it on, if something goes terribly terribly wrong from an
organizational point of view.
Take a look at URL shortening services too, like tinyurl.com and
is.gd. They allow some API access in some cases, and so they aren't
terrible, but at the same time it's still somewhat of a SPOF or single
point of failure for the system of URL shortening (or redirecting-
which essentially any ID-lookup service is going to do).
There was an interesting idea proposed at MIT a few years ago called
physical XML, and it was quite literally an ambitious project to make
an XML specification for keeping track of all forms of information
from design to the time that it gets to the end-user, and then all
through the life-cycle. So, you might as well just put those files on
an RFID chip, or in the case of open source bar code scanning and
optical character recognition (OCR), you could encode a lot of
information on a barcode sticker- maybe even all of the information
about the object. Registration hooks back to centralized servers is
going to be useful, but to me it just sounds like it's something that
will easily break. How do we avoid that?
- Unique IDs (hard)
- Tags (easy)
- Hierarchical Linnaean taxonomy, in Latin (cool, but difficult and
fairly useless. Cool though. In fact its so cool I feel slightly faint.
I think I'd better go have a lie down)
I've had quite a lot to do with databases... getting data from a system
where things have non-unique and overlapping names to one where they
have unique IDs is a real bastard.
Being able to uniquely ID every manufactured product would be a massive
massive society-changing innovation though I think. On a par with
standardising bolts/screws etc back at the beginning of the Industrial
Revolution... possibly even more so.
Being able to hit a button and see the entire genome of a product would
have huge implications for ethical sourcing / externalising costs etc...
and maybe that's a place to start... maybe producers who want to compete
on "purity of source" could start adopting uniqueIDing.
Unfortunately, I think such a system would also need its own
bureaucracy/police force... or maybe we can get away with the
global-panopticon effect.
It's an interesting idea. Yer corporates would hate it.
And you say you've had a lot of work done with databases? I don't
think so. In many cases, IDs are directly implemented in databases
with UNIQUE_ID fields, and so on. So, these IDs are just as easy as
tags- which are usually either an entire table devoted to
cross-referencing a certain tag ID to a certain thingy ID, or some
parseable column in a table.
Oh Ye of little faith.
> In many cases, IDs are directly implemented in databases
> with UNIQUE_ID fields, and so on. So, these IDs are just as easy as
> tags- which are usually either an entire table devoted to
> cross-referencing a certain tag ID to a certain thingy ID, or some
> parseable column in a table.
Yup, for my sins I am the Director of IT for The Association of Football
Statisticians and although there are DBAs out there the hems of who's
garments I am not worthy to touch, I'm still a DBA, so I know what
Unique IDs are... enough to know that there are certain jazz-purist DBAs
who believe that there should be no such thing as a numeric primary key.
I disagree on the grounds that life is too short.
Anyway, the reason IDs are not as easy as tags, is that tags are "a
matter of opinion" and the implementation can be crowd-sourced - in fact
the more they are crowd-sourced, the more useful they become.
Unique IDs are another kettle of fish entirely, because you need to
uniquely identify the object before you can put it into the DB. This can
be hard.
Take... Football for example, say you've just inherited a foxpro
spreadsheet with 2,000,000 lines of player-results and the players
aren't uniquely IDed.
You have about 1000 lines with:
Simon Barry
S Barry
S. Barry
Si Barry
Simone Barry
Simon Bary
Barry, Simon
Barry Simon
etc etc and you know that there is more than one Simon Barry and there
is also a Barry Simon and there might actually be a Simone Barry.
We come up against this all the time. It is the most difficult aspect of
populating databases - the Chinese and South American ones are
particularly tricky.
The difficulty is in getting stuff into the database rather than
creating the structure... and I'm guessing that uniquely IDing
already-existing physical objects will be a hell of a lot harder than
people - especially if this involves importing of existing data-sets
with overlapping parts that don't follow the same system.
What? So some random bloke on the other side of the planet comes up
with a new tag, thus invalidating another tag that covers the same
thing that another bloke added the other day, making this an entirely
pointless, waste of time.
So how much do you know about folksonomy?
You're going off half-cocked here Bryan, which isn't like you at all.
Drunk too much coffee?
I suppose what I said could be misinterpreted to mean something else.
Let me clarify. In particular, it's useless for there to be redundancy
due to people tagging to their hearts content. I don't care about
their silly personal ontologies (and I'll be damned if I'm going to
sit here forever reconciling different tagging ontologies together
just to make certain packages work), it just gets icky and messes
things up and halts the software automation.
The redundancy is part of the mechanism by which it works... I think
you're seeing everything (as per the video conversation) through the
lens of machine automation. There's more to it than that - and for as
far forward as any of us can see into the future, humans are going to be
the ultimate vitamin part.
We're building ourselves a planet-sized symbiote... and writing off
tagging/folksonomy is writing off something that the wetware part of the
system finds useful.
I think you might be imagining trying to use tagging systems as unique
identifiers. That's not what they are - tagging / folksonomy is about
finding non hierarchical routes through the data. It's a short-cut to
the sort of machine intelligence that allows ebay
http://shop.ebay.com/?_from=R40&_trksid=p0.m38.l1313&_nkw=treadmill+motors&_sacat=See-All-Categories
to know that treadmill motors are good for wind-power generation.
To break it down to its nuts and bolts (and assuming we're talking
relational DBs), tagging is a crowd-sourced alias table. The 'parts'
table also needs a unique ID. The two aren't mutually exclusive.
> - Freecycle seems to be open to almost everyone; this project should
> perhaps only be open to *trusted* individuals
This is one of the greatest things about freecycle, it's openness. If I
can't get what I need because I'm not "trusted" then what's the point?
What basis is this "trust" derived from?
The reasons freecycle doesn't work:
- It places a high burden on the disposer to catalog what they are getting
rid of, answer a bunch of emails from (usually stupid/greedy) people,
arbitrarily select one of these people on the basis of near zero
information, exchange numerous emails about address and to show up at
certain times, then verify that it's gone, and then write another post
that it's gone.
- It burdens the receiver with massive amounts of irrelevant mail which
they must dig through feverishly in order to be the first to respond. They
have to show up at a certain time. (Why?) There's no categorization
system, no way of tagging nor time to tag anything before the item is
claimed. Most of the items posted are simply garbage.
- There is no automation, it's just a mailing list.
So instead i'd like to see some system where you just take a picture on
your cellphone of what you're getting rid of; this is geotagged and
uploaded to your account, where you can apply pick-up schedules and
preferences to multiple items at a time, and tag items, categorize them,
or write a description if you feel inclined. When you publish the items
they get sent via RSS and/or email, and also posted on the website.
You could limit item views to your circle of friends if you are a
tribalist bastard.
End-users can then tag the items and add descriptions, and place a 'hold'
request which would prevent other users from getting the geo-location for
a period of time. You could reward users for adding tags and descriptions
by (automatically) putting them ahead in the queue. This doesn't really
solve the "who deserves it" problem but at least it reduces the workload.
> copper, aluminium,
how do you avoid people who will immediately sell these to a scrapyard?
> salvaged/desoldered electronic components,
why would I expose myself to toxic chemicals and then give away the
results of my labor?
I'd really like to see more businesses involved in this. Freecycle sucks
because it's all household crap. Meanwhile down the road there is a
dumpster full of linear bearings and servo drives, but nobody bothered to
try to share them because it's "obviously trash" or it wouldn't be in the
dumpster right? Maybe there does need to be a monetary incentive to get
people to upload information.
This is easy enough to write... which is to say, I could do it. The
problem is getting enough people to use it for it to be useful... which
is to say, I've failed to achieve critical mass with web-apps before.
So. Rules (of thumb) for getting web-apps to work.
1) it needs to still be useful with a really small user-base.
2) it really helps if it helps people make (not just save) money.
3) it really helps if it helps people propagate the memes that they are
addicted to (or find interesting).
If you can think of a model that includes these then I'd be happy to
write it... or at least start that conversation.
By the way, have you seen this:
http://tgimboej.org/
http://www.curiousinventor.com/blog/44
Cute idea... and to be honest, I can see it working - given the junk
that I'd throw away - I know it's good stuff.
Nick