5 ms·

Thingd is building a structured database of every object in the world

42 points by freejoe76 16y ago

rglullis 16y ago
It's one of those ideas that seem so obvious and that has the potential to change the world, but it is highly dependent of having critical mass. Technologically, it is very simple to implement it. One doesn't need much more than something like CouchDB, and you could have users changing things even from the admin interface. With job4dev (shameless plug), we have something very similar, applied to job listings and company info. We created a "wikipedia for jobs", but lots of the curating needs to be done by hand. We are automating as much as we can, using as many APIs as possible, but there's still the need for human involvement. The problem is, you need to have people wanting to use this. It's the ultimate chicken-and-egg problem. I could see this working if there is some sort of incentive, or even if it ties itself to Mechanical Turk. Without that, it is no different than Freebase, dbpedia, or Aaron's Infogami.
terra_t 16y ago
How's this better than Freebase or dbpedia?
- affleck 16y ago
  freebase and dbpedia are tied to wikipedia wiki articles about concepts and they seem to be trying to do products.. different data domain I don't know which one is "better" but this seems to have commercial tie-ins where wikipedia might not.
  - rictic 16y ago
    Freebase was seeded with Wikipedia, but we're always loading data from new license compatible sources. We've got books from Stanford Library, music from MusicBrainz, films from Netflix, and a lot more. We don't have good coverage of product data right now though.
spoon16 16y ago
How does this compare to Freebase?
- affleck 16y ago
  freebase is structured wikipedia (concepts) this thingd looks more like structured ebay (objects)
stcredzero 16y ago
This reminds me of Bruce Sterling's "Internet of Things." http://en.wikipedia.org/wiki/Internet_of_Things http://en.wikipedia.org/wiki/Internet_of_Things http://video.google.com/videoplay?docid=-3857739359956666768# http://video.google.com/videoplay?docid=-3857739359956666768...
lubos 16y ago
Joel Spolsky predicted this project 10 years ago :) "When you go too far up, abstraction-wise, you run out of oxygen. Sometimes smart thinkers just don't know when to stop, and they create these absurd, all-encompassing, high-level pictures of the universe that are all good and fine, but don't actually mean anything at all." http://www.joelonsoftware.com/articles/fog0000000018.html http://www.joelonsoftware.com/articles/fog0000000018.html
Groxx 16y ago
>[signing in]: It's easy: just connect your Facebook or Twitter account [and no other means] So "easy" is not allowing me to make a login not tied to everything else. Anyone with an account care to inform me / us what you can do with an account? Just like / want it, or add thingds, or...? I'm interested, but the site contains no info about the site aside from the TOS and privacy policy. I should really make a throw-away set of accounts on a handful of social sites, just to get around crap like this. If they allowed straight OAuth or OpenID instead of Twitter-hosted OAuth, maybe I'd use it. But nobody does that.
- abraham 16y ago
  Plus they ask for an email address/password during registration which nullifies any possible gain of a quick registration. You can unlink any both Facebook and Twitter after registration.
  - mike-cardwell 16y ago
    After they've gathered whatever information they can from you by having access to your Facebook profile or Twitter stream.
- terra_t 16y ago
  BTW, who the heck are these people? I don't see any contact form, contact email, contract phone or anything.
jasonjei 16y ago
At the very heart of it, it could map every object to its BOM including carbon based lifeforms. Imagine if every object were a class, including humans. The atom could be a basic super class, and you could map out various dependencies and object superclasses (e.g, Oxygen < Atom < Base). And it might even work with non physical things (Capitalism < EconomicTheories).
- Raphael 16y ago
  Does it support multiple inheritance? For example, water is made up of two elements.
  - gnaritas 16y ago
    Water is composed of two elements, nature, like good programmers, prefers composition to inheritance.
    - nl 16y ago
      I have both a Mother and a Father. Obviously nature has a bit left to learn.
      - gnaritas 16y ago
        No, there's a time and place for inheritance, just not as often as composition.
- paragraft 16y ago
  The physical world is never going to fit one taxonomy neatly, let alone the conceptual one.
rxin 16y ago
Isn't this what semantic web is about?
- edwincheese 16y ago
  I would prefer a open semantic web approach rather than a privately owned big database
- rfugger 16y ago
  And equally impossible to do in a way that is globally useful.
  - edwincheese 16y ago
    I still believed (or hope) that semantic web is the future of Internet. If a very particular industry willing to open up their data in standard and extensible format (may be RDF or a simpler standard), like film theater listing their showtimes or governments release their statistical data in RDF rather than Excel only, many useful application could be developed on it and it is a big step towards semantic web already. And I think it is not too hard to do.
    - terra_t 16y ago
      If RDF's got a problem it's that it's too simple, at least in terms of the functionality it's offered. It's complicated as hell to do simple things in RDF, but that's a different issue, not the one I struggle with every day.
- Devilboy 16y ago
  I think they're brute-forcing the semantic web
- joshu 16y ago
  No. The "semantic" refers to the semantics of schema, not data. Everyone gets this wrong.
LordLandon 16y ago
If only 5th Cell released Objectnaut, they'd have a great start!
troymc 16y ago
Any website or web service with a database of "things" (with attributes) is a competitor to Thingd: WorldCat, IMDB, gdgt, Google Merchant Center, Zappos, CNET reviews, Etsy... Good luck Thingd --- you'll need it!
derefr 16y ago
If they don't stop at just describing the semantic knowledge that a person would derive from the experience of objects, but instead also include the process knowledge—in a rigorous, possibly algorithmic form—then it would be the world's biggest MOO.
ntoll 16y ago
This is very similar to FluidDB (Fluidinfo, the creators of FluidDB have recently closed series A funding with some great backers).
dirtyaura 16y ago
Thinglink ( http://thinglink.com http://thinglink.com ) has tried several approaches in this space. Originally they offered this very abstract kind of "URL for things" service, then they pivoted to be more a social network around things, then they pivoted to be a social network for designers and people who like designer clothes and other designer objects. Now, their final pivot was in my opinion marvelous, as it went to very basics and offered a super-simple service to tag things in photographs. It's still operating in a same space "to URLify real world objects", but it offers an immediately useful service, for example for fashion bloggers.
praeclarum 16y ago
All these tags mean the same thing to me. 3.75 inches 3.75 inches tall 3.75 x 3.75 x 5.75 inches 3.75 x 4.5 x 7 inches The problem with human entered / non moderated data is that these lists will become longer and longer over time and the database will not understand that they all mean the same thing. Joins will be impossible and its relevance fades.

42 points by freejoe76 16y ago

It's one of those ideas that seem so obvious and that has the potential to change the world, but it is highly dependent of having critical mass. Technologically, it is very simple to implement it. One doesn't need much more than something like CouchDB, and you could have users changing things even from the admin interface. With job4dev (shameless plug), we have something very similar, applied to job listings and company info. We created a "wikipedia for jobs", but lots of the curating needs to be done by hand. We are automating as much as we can, using as many APIs as possible, but there's still the need for human involvement. The problem is, you need to have people wanting to use this. It's the ultimate chicken-and-egg problem. I could see this working if there is some sort of incentive, or even if it ties itself to Mechanical Turk. Without that, it is no different than Freebase, dbpedia, or Aaron's Infogami.

terra_t 16y ago

How's this better than Freebase or dbpedia?

affleck 16y ago
freebase and dbpedia are tied to wikipedia wiki articles about concepts and they seem to be trying to do products.. different data domain I don't know which one is "better" but this seems to have commercial tie-ins where wikipedia might not.
- rictic 16y ago
  Freebase was seeded with Wikipedia, but we're always loading data from new license compatible sources. We've got books from Stanford Library, music from MusicBrainz, films from Netflix, and a lot more. We don't have good coverage of product data right now though.

spoon16 16y ago

How does this compare to Freebase?

affleck 16y ago
freebase is structured wikipedia (concepts) this thingd looks more like structured ebay (objects)

stcredzero 16y ago

This reminds me of Bruce Sterling's "Internet of Things." http://en.wikipedia.org/wiki/Internet_of_Things http://en.wikipedia.org/wiki/Internet_of_Things http://video.google.com/videoplay?docid=-3857739359956666768# http://video.google.com/videoplay?docid=-3857739359956666768...

lubos 16y ago

Joel Spolsky predicted this project 10 years ago :) "When you go too far up, abstraction-wise, you run out of oxygen. Sometimes smart thinkers just don't know when to stop, and they create these absurd, all-encompassing, high-level pictures of the universe that are all good and fine, but don't actually mean anything at all." http://www.joelonsoftware.com/articles/fog0000000018.html http://www.joelonsoftware.com/articles/fog0000000018.html

Groxx 16y ago

>[signing in]: It's easy: just connect your Facebook or Twitter account [and no other means] So "easy" is not allowing me to make a login not tied to everything else. Anyone with an account care to inform me / us what you can do with an account? Just like / want it, or add thingds, or...? I'm interested, but the site contains no info about the site aside from the TOS and privacy policy. I should really make a throw-away set of accounts on a handful of social sites, just to get around crap like this. If they allowed straight OAuth or OpenID instead of Twitter-hosted OAuth, maybe I'd use it. But nobody does that.

abraham 16y ago
Plus they ask for an email address/password during registration which nullifies any possible gain of a quick registration. You can unlink any both Facebook and Twitter after registration.
- mike-cardwell 16y ago
  After they've gathered whatever information they can from you by having access to your Facebook profile or Twitter stream.
terra_t 16y ago
BTW, who the heck are these people? I don't see any contact form, contact email, contract phone or anything.

jasonjei 16y ago

At the very heart of it, it could map every object to its BOM including carbon based lifeforms. Imagine if every object were a class, including humans. The atom could be a basic super class, and you could map out various dependencies and object superclasses (e.g, Oxygen < Atom < Base). And it might even work with non physical things (Capitalism < EconomicTheories).

Raphael 16y ago
Does it support multiple inheritance? For example, water is made up of two elements.
- gnaritas 16y ago
  Water is composed of two elements, nature, like good programmers, prefers composition to inheritance.
  - nl 16y ago
    I have both a Mother and a Father. Obviously nature has a bit left to learn.
    - gnaritas 16y ago
      No, there's a time and place for inheritance, just not as often as composition.
paragraft 16y ago
The physical world is never going to fit one taxonomy neatly, let alone the conceptual one.

rxin 16y ago

Isn't this what semantic web is about?

edwincheese 16y ago
I would prefer a open semantic web approach rather than a privately owned big database
rfugger 16y ago
And equally impossible to do in a way that is globally useful.
- edwincheese 16y ago
  I still believed (or hope) that semantic web is the future of Internet. If a very particular industry willing to open up their data in standard and extensible format (may be RDF or a simpler standard), like film theater listing their showtimes or governments release their statistical data in RDF rather than Excel only, many useful application could be developed on it and it is a big step towards semantic web already. And I think it is not too hard to do.
  - terra_t 16y ago
    If RDF's got a problem it's that it's too simple, at least in terms of the functionality it's offered. It's complicated as hell to do simple things in RDF, but that's a different issue, not the one I struggle with every day.
Devilboy 16y ago
I think they're brute-forcing the semantic web
joshu 16y ago
No. The "semantic" refers to the semantics of schema, not data. Everyone gets this wrong.

LordLandon 16y ago

If only 5th Cell released Objectnaut, they'd have a great start!

troymc 16y ago

Any website or web service with a database of "things" (with attributes) is a competitor to Thingd: WorldCat, IMDB, gdgt, Google Merchant Center, Zappos, CNET reviews, Etsy... Good luck Thingd --- you'll need it!

derefr 16y ago

If they don't stop at just describing the semantic knowledge that a person would derive from the experience of objects, but instead also include the process knowledge—in a rigorous, possibly algorithmic form—then it would be the world's biggest MOO.

ntoll 16y ago

This is very similar to FluidDB (Fluidinfo, the creators of FluidDB have recently closed series A funding with some great backers).

dirtyaura 16y ago

Thinglink ( http://thinglink.com http://thinglink.com ) has tried several approaches in this space. Originally they offered this very abstract kind of "URL for things" service, then they pivoted to be more a social network around things, then they pivoted to be a social network for designers and people who like designer clothes and other designer objects. Now, their final pivot was in my opinion marvelous, as it went to very basics and offered a super-simple service to tag things in photographs. It's still operating in a same space "to URLify real world objects", but it offers an immediately useful service, for example for fashion bloggers.

praeclarum 16y ago

All these tags mean the same thing to me. 3.75 inches 3.75 inches tall 3.75 x 3.75 x 5.75 inches 3.75 x 4.5 x 7 inches The problem with human entered / non moderated data is that these lists will become longer and longer over time and the database will not understand that they all mean the same thing. Joins will be impossible and its relevance fades.