link to older bug; practical issues exist

[git.ikiwiki.info.git] / doc / news / openid / discussion.mdwn

diff --git a/doc/news/openid/discussion.mdwn b/doc/news/openid/discussion.mdwn
index c0447a13fdbc84e8d8896581cacd64560c55055f..d8c83f022ff422429c192466b40aed2674a15bc9 100644 (file)
--- a/doc/news/openid/discussion.mdwn
+++ b/doc/news/openid/discussion.mdwn
@@ -90,3 +90,34 @@ I just tried logging it with OpenID and it Just Worked.  Pretty painless.  If yo
 
 ###LiveJournal openid
 One caveat to the above is that, of course, OpenID is a distributed trust system which means you do have to think about the trust aspect.  A case in point is livejournal.com whose OpenID implementation is badly broken in one important respect:  If a LiveJournal user deletes his or her journal, and a different user registers a journal with the same name (this is actually quite a common occurrence on LiveJournal), they in effect inherit the previous journal owner's identity.  LiveJournal does not even have a mechanism in place for a remote site even to detect that a journal has changed hands.  It is an extremely dodgy situation which they seem to have *no* intention of fixing, and the bottom line is that the "identity" represented by a *username*.livejournal.com token should not be trusted as to its long-term uniqueness.  Just FYI.  --[[blipvert]]
 
 ###LiveJournal openid
 One caveat to the above is that, of course, OpenID is a distributed trust system which means you do have to think about the trust aspect.  A case in point is livejournal.com whose OpenID implementation is badly broken in one important respect:  If a LiveJournal user deletes his or her journal, and a different user registers a journal with the same name (this is actually quite a common occurrence on LiveJournal), they in effect inherit the previous journal owner's identity.  LiveJournal does not even have a mechanism in place for a remote site even to detect that a journal has changed hands.  It is an extremely dodgy situation which they seem to have *no* intention of fixing, and the bottom line is that the "identity" represented by a *username*.livejournal.com token should not be trusted as to its long-term uniqueness.  Just FYI.  --[[blipvert]]

+
+----
+
+Submitting bugs in the OpenID components will be difficult if OpenID must be working first...
+
+------
+
+# Privacy and Decentralization
+
+Maybe I don't understand OpenID well enough, but it looks like there are just few providers, most
+of which are huge companies or belong to such, and I don't trust them to verify me identity
+or to not track all my logins. I'll use OpenID only if I can make my own home server
+be my OpenID provider, and if doing so doesn't interfere with the design and security and
+privacy of OpenID, and doesn't require me to use centrally-signed certificates or pay to some
+company or anything like that.
+
+Is it possible to use OpenID in a way keeping the user in full control and allowing any user to
+have their personal provider without damaging the architecture behind OpenID?
+
+I'm worried, at least until the issue is cleared.
+
+-- [[fr33domlover]]
+
+> You can install an OpenID provider on your own server and use that if you
+> wish. I believe you will need an SSL certificate that `ikiwiki.info` trusts.
+> -- [[Jon]]
+
+----
+
+This poll is now 8 years old. Do we have enough data to make a decision?
+Can we consider adding `open=no` to the poll? -- [[Jon]]