|
Apr
10
|
|
I think Tim O'Reilly has it wrong. His blogger's code of conduct, while well intentioned, is unworkable. What the web needs is a reputation system. And by reputation I mean the social evaluation of a person, group or entity. The Social Evaluation - that's the important bit.
Here's how it would work. If you want to be taken seriously, people need to know who you are; something like OpenID would be adequate for this. Strictly it doesn't really provide anyone with any level of trust that you are who you say you are - but that's OK; just knowing that you're the same entity that left comments on other sites, or contributed to a wikipedia article is enough. Next we need a way of recording a reputation against OpenIDs - people should be able to very simply score you on your level of insight, wit, general behavior, etc. on the web - this has been solved locally for ebay and slashdot; what we now need is some way to federate your reputation globally.
The need for anonymity can't be pushed aside in this system; there are very good reasons for allowing anonymity. Flaming people or competing products, pumping up your own views, etc. are not good uses of anonymity and unfortunately cast the use of Anonymity in a bad light. Like I said there are good reasons for anonymity to be maintained for those who really need it - I live in a country and work for an organization where I hope I'll never need to hide my opinion. But there are many circumstances where anonymity is required,
The idea of web-reputation is anything but new - ebay, slashdot, to name a few have been using it successfully for years to provide user-moderation; what's missing is some convenient standards for federating or compositing reputation. A couple of sites have been trying to just that - for example here's my Opinity page. Opinity is a service that allows me to have some control over my on line reputation and does some degree of reputation compositing; but it isn't perfect; and it isn't really widely enough used to be of much practical use (today).
Here's my prediction - as the web moves inevitably towards adopting user-centric identity; reputation (and federated reputation) won't be far behind.
What do you think ?
|
Apr
1
|
|
This is Part 2 (of 3) of my notes from O'Reilly Emerging Technology Conference, 2007 in San Diego. (see Part 1)
Day two (Tuesday) was supposed to start with Kathy Sierra - but she was forced to cancel due to her online death-threat incident. That was a shame in many ways and means I still haven't had the pleasure of listening to Kathy Sierra talk. Instead of Kathy, Jeff Jonas was back talking about data and how to get the most out of it. A bit of a repeat but he did go into a little more detail.
Next up was Werner Vogels who talked about Amazon Web Services. This was really interesting - Amazon seem to be really leading the way in delivering a web platform vision. It's taken them a decade and $2 billion - if you were thinking of trying this at home. Their goals are to turn huge, fixed costs into lower variable costs, provide unlimited scalability all with datacenter reliability.
Rightscaling is the term Werner used to describe the cost-efective scaling - I heard an example of a startup laying down just $85 for their first 3 months of IT operations; yet could easily dip into the (virtually unlimited) pool to scale up as demand required.
Werner really only talked about the infrastructure layers, which (if you're not familiar with AWS) consists of EC2 (Elastic Computing Cloud), Messaging (Simple Queuing Service) and Storage (S3 - Simple Storage Service) - they use metered pricing for everything - eg. $0.15/Gb/Month for storage or $0.10/Instance/Hour for EC2).
The way they do provisioning for EC2 was pretty neat - you basically assign an AMI (Amazon Machine Image) to the physical units you want to provision; the AMI defines the OS and software you want installed (ie. you might choose Fedora, Apache and PHP or Fedora and Ruby on Rails). The AMIs are stored on S3 and you can either define your own or use pre-configured images.
The deployments can be tiered (for security) in the same way you might in your own DC - ie. only the Web or Proxy tier would be exposed to the public internet; whereas the database tier would be completely inaccessible.
I had to skip the rest of the morning due to a couple of meetings but Phil Windley has some great coverage of Jane McGonigal's talk on Creating Alternate Realities.
In the afternoon there was a pretty good demo from Apple showing how easy it is manipulate pictures and video. All you need is a moderately powered super-computer - ie. like the one you're reading this text on.
Jeff Hawkins (inventor of the Palm Pilot and Treo) talked about Hierarchical Temporal Memory - the idea of HTM is to replicate some of the human brains congnitive functions - they learn from exposure to sensory data, discover cause and make predictions based on the sensory data and inferences. Some of this seemed a bit far-fetched - but hey, this is Jeff Hawkins (the inventor of the Palm Pilot and Treo).
Note to self - I must give "On Intelligence" another try - I wasn't sufficiently motivated to get past the first hour.
Much later in the day I attended a couple of (related) BoFs. The first was lead by Joe Gregorio covering APP (Atom Publishing Protocol) - I was interested in knowing where APP was in terms of adoption - it seems there's a bit of a chicken and egg situation - there won't be many clients supporting APP until there are some services and the services won't support APP without clients. Apparently APP (even before the RFC is final). There was some interesting side-chat about APP / Atom / RSS over Jabber / XMPI - apparently IM services aren't censored any where near as much as HTTP and VOIP - so it may well be the only truly universal transport.
Matt Mullenweg of WordPress claimed that WordPress would be supporting APP real soon - so hopefuly the cycle will be broken and the Web will get it's universal publish button. Anyone know when Roller will support APP ?
Next (in the same room) was the Microformats BoF run by Kevin Marks; who, significantly, works at Google. Most of the people in the room (me included) were merely there to learn a bit more but there was one guy (sorry I'm lousy with names) who was using rel-tag extensively. There was some discussion about how well microformats fit with APP; how microformats could deliver some of the power that Metaweb promises. This is another area I need to investigate a little more.
