On Thu, Jun 11, 2009 at 05:45:54PM +0100, Chris Dent wrote:
> On Jun 10, 2009, at 11:54 PM, Noah Slater wrote:
>> This is an absolute show-stopper for any third-party site.
>
> I'm not sure I agree with this. Even if my photos are unavailable to, say,
> Flickr at any given time, everybody else's photos are likely to be available,
> because they come from some other bank, which is currently up. So over all the
> flickr service remains up, while not working for my content. This is more
> embarrassing for than it is for Flickr.

Flickr would never use a system that put the content they were serving in the
hands of someone else. Community approved content could be swapped out for
pornography. Slow servers, or non-responsive ones, would cripple the usability,
and perceived latency of the user experience.

This is why it is a show-stopper.

> In other words, perhaps most relevant for the autonomo.us crowd, when
> things go right or wrong with _my_ content, it's on me: it's my success
> or my problem.

When so much of the user experience is based on the perceived latency of the
network, and when community ratings and filters are so important, it becomes
Flickr's problem - not yours.

> Right now they host over two billion photos. Under my proposal they host
> no photos, they host urls of photos and some metadata. They _present_
> those photos that people want to see right now.

There is no chance they would take this risk.

And why should they?

It would very probably ruin the user experience. When I browse Flickr, load
times feel almost instant, which is very impressive given how many large images
are flying over the network. This is only possible because Flickr uses a very
advanced, and very fast, CDN. If these images were being served from other
people's personal servers, the page load times would suffer immensely, with
images occasionally failing to load entirely.

That doesn't sound like the type of site that I would want to use.

> If they wish to do any caching, they do caching based entirely on the caching
> protocols built into HTTP.

Either they cache the images locally, or they don't. Saying that they can use
HTTP caching to avoid these problems does not escape the fact this is a huge
amount of images to be constantly revalidating. It also doesn't change the fact
that the content can change beyond Flickr's control, which would circumvent many
measures put in place to protect people from seeing pornography.

Also related, each photograph on Flickr is available in up to six sizes:

  http://www.flickr.com/photos/amysimons/3602803206/sizes/m/

How would Flickr generate these, and keep them up to date?

> I realize this is very pie in the sky, but the sky is where ideas come
> from. I don't really expect people to walk out and do this, but if it
> stimulates people to think differently about content, then that's great.

Sure, it's useful to toss ideas about to see what comes of it!

Best,

-- 
Noah Slater, http://tumbolia.org/nslater
_______________________________________________
Discuss mailing list
Discuss@lists.autonomo.us
http://lists.autonomo.us/mailman/listinfo/discuss

Reply via email to