Re: [libreplanet-discuss] Building a free/libre platform for reviews

[Erik Moeller]

2016-05-26 12:04 GMT-07:00 Aaron Wolf <wolftune@riseup.net>:

> Just writing to cheer you on. A truly FLO review system that doesn't
> have a conflict-of-interest in being used to advertise or to be a
> network-effect keeping people on one platform like Amazon… this would be
> a super valuable thing if it succeeds.
>
> If I could use it and publish my reviews under copyleft terms, i.e.
> CC-BY-SA, I would be writing LOTS of reviews for it.

Yay! Thanks for saying so :)

> I think besides the work of moderating it (you do know that Yelp is
> *full* of illegitimate paid reviews, right?)

Yeah. I don't think that I want to follow the model where a corporate
player (even a nonprofit) has to remove or vet reviews. That should be
done by the community using moderation tools, participatory processes,
etc.

I'm especially interested in how team membership could add to the
sense of community. Teams could apply shared values to reviewing
businesses, products, etc., whether it's free software values or, say,
concerns about animal rights, labor practices, etc. -- and perhaps
even award their own certificates. A little bit like a subreddit
(though I think we'd need a stronger code of conduct from the get-go),
or a WikiProject in Wikipedia parlance.

> the other challenge would
> be getting items to review to be listed to easily find so people don't
> have to add them (and likely see far more accidental redundant listings
> than Amazon even).

Indeed. I think ultimately it's a huge challenge, as you say, since
you're dealing with millions or even billions of records. To produce
something usable quickly, I think I'll start with making it possible
to associate a set of reviews with a set of URLs identifying the same
real world entity, and building intuitive tools for merging items.
Then build out the metadata layer for one "type" of thing people want
to review, so we can test the limitations and refine the UI, while not
yet bulk importing huge open datasets.

That said, the OSM and OpenLibrary data do indeed look promising, and
there's lots of work in Wikidata as well on metadata for creative
works etc.

Tentatively, I'm calling the project "lib.reviews" (yep, I'm using a
silly new TLD). More to come :)

Cheers,
Erik