Kevin Lawver is giving a talk about the genesis of Ficlets.com. He's the dark shadow next to the slide. It's very interesting to hearhow the team was able to experiment and push this small project out ina matter of weeks.It strikes me that Ficlets may well increase the Overton Windowof what's reasonable to seriously consider at AOL. The agile process,tools such as Rails, OpenID, and Creative Commons have now beenlaunched, which moves discussions from "will it work?" and "can we getit approved?" to "will it work for this project?". Which I think is amovement in AOL's internal window of acceptable discourse and givesother people air cover.
Kathy, you have my sympathy and support.
It's great to see Netscapestart accepting OpenIDs. It's good to see the user experienceissues being worked out -- the best methods will win out over time. Dmitry Shechtman calls for theelimination of registration, though what he actually says is a bitstronger:
The usefulness of OpenID is void if yourservice requires all users to sign upregardless of whether they have an OpenID. This doesn’t mean that youshouldn’t ask them to provide additional details, such as an activee-mail address (although OpenID Simple Registration usually handlesthat for you). You can do it after the user signs in for thefirst time. Just don’t ask her to sign up.I understand this point of view, but disagree:
- Users expect to find "sign up" or "register" links when going toa new site; not providing them breaks their expectations and that's bad.
- If a user doesn't have an OpenID already, they do need toregister with your site. You have to make this just as easy as beforeyou added OpenID to your UI.
- Sometimes you really do need to gather an e-mail address, agreeto Terms of Service legalese, etc. before using a site.
Unfortunately, you don't currently know who has an OpenID and whodoesn't. Maybe there's room here for a service: gotopenid.org. Yousign in there once with your OpenID, and it cookies you as someone whowants to use OpenID wherever possible. Sites that want to offer anOpenID-streamlined experience could make a JSON call tohttp://gotopenid.org/check.js. This would tell the site that the userknows what OpenID is and prefers to use it for signing in.
On a side note, it looks to me by the screen shots that the AOL/AIMscreen name integration is simply using the AOL OpenID identityservice. We'll see on Monday; certainly that's the easy way to do it-- a convenience wrapper around an OpenID sign in.
Tags: openid, launch, openid consumer, aol
REST's most visible implementation -- the Web -- has clearly been insanely popular as a read-only space. There are many books on how to put documents on the Web. There are very few (no?) books on how to apply the full REST style to distributed programming over HTTP (inconceivable?). It's appropriate that RESTful Web Services starts with a manifesto:
We want to restore the World Wide Web to its rightful place as arespected architecture for distributed programming. We want to shiftthe focus of web service programming from an RPC-style architecturethat just happens to use HTTP as a transfer protocol, to a URI-basedarchitecture that uses the technologies of the web to their fullest.Does this remind anyone else of the clifftop scene from The Princess Bride?
[Scene: At the top. Fezzik, Vizzini and Inigo are looking down at
the masked man climbing the cliff after Vizzini has cut the rope]
Fezzik: He's got very good arms.
Vizzini: HE DIDN'T FALL? INCONCEIVABLE!
Inigo: You keep using that word. I do not think it means
what you think it means. [pause] My God! He's