It's a good idea to have lots of checks in place.  Twitter reliably
returns unreliable data... ESPECIALLY BEFORE HOLIDAY WEEKENDS.  In
fact, for a couple of calls yesterday we got non-error responses
saying that the user had ZERO followers.  Then it was 5000 followers.
Now it is their follower count +- 5%.

I could deal with this if Twitter.com used the same bogus data, but
Twitter itself doesn't use its own API calls, so the DM count, the
follower/friend count... these are all internally computed and do not
match any existing API call.

On Sep 5, 11:35 am, Jesse Stay <jesses...@gmail.com> wrote:
> Fortunately it only affected a couple users, but I'd like to make it up to
> them.  BTW, this didn't affect the mass unfollow feature you saw Scoble and
> others using (that would have worked fine). This affected the "unfollow
> those who unfollow me" feature.  We have safety valves in place as well, but
> there are still a few users that get through that.
> Jesse
>
> On Sat, Sep 5, 2009 at 12:30 PM, Dewald Pretorius <dpr...@gmail.com> wrote:
>
> > Jesse,
>
> > Last night when this thing hit I actually immediately thought about
> > you and wondered how it impacted you.
>
> > I'm now thanking my lucky stars that I don't do mass unfollow. I do
> > have the "unfollow those who unfollow me" feature, but I have limited
> > it to a maximum of 10 unfollows every 8 hours, even if the API said
> > that more people have unfollowed the user. That safety valve has
> > seriously saved my butt this time.
>
> > Dewald
>
> > On Sep 5, 3:06 pm, Jesse Stay <jesses...@gmail.com> wrote:
> > > Again, I can't stress this enough - when bugs like this are introduced,
> > it
> > > is imperative that follow limits are also removed temporarily (or on a
> > > case-by-case basis) so we can make this up to our users.  I've already
> > had
> > > to issue refunds to a couple due to this.  If you need me to send you the
> > > usernames John let me know.
> > > Jesse
>
>

Reply via email to