My new blog is here

Widgets and the data web Permanent link to this item in the archive.

Also known as, don't mistake the appearance for the idea.

Valleywag has an article on widgets entitled "HYPEBUSTING: Against widgets". I found it via Ivan at Snipperoo who - no surprise - disagrees with it.

I think that widgets are a particular incarnation of the web as data, but by no means the best example. Widgets are to the data web as MySpace is to social networking - a good first attempt.

Look at the RAM usage of Apple Dashboard and tell me there's not a more efficient way of displaying this kind of data for the user. A typical Dashboard widget uses 30-50Mb of RAM. What exactly is it doing? Downloading some XML or JSON data from a server and making it look pretty. Desktop widgets suffer because of synchronisation issues. How do I know when this widget was last updated? With an RSS feed in a news reader, it's pretty simple - it's got a date and time stamp attached which I can look at.

Unlike RSS, widgets can't be maintained en masse. While various RSS readers are now introducing attention tracking (how many times do I click on Scripting News posts versus Boing Boing posts?), widgets remain lumps of dumb JavaScript. I can't filter my widgets and say "show me only widgets I've interacted with in the last month". And they don't scale. I know people who are producing widgets for their podcasts as an alternative to RSS. Great. My podcatcher can track three hundred RSS feeds with enclosures quite a lot more efficiently than my widget platform can track three hundred widgets.

I know that if I'm hosting my own blog, I'd rather have a script sitting on my server which would pull in the data, cache it, transform it and serve it out as HTML than a JavaScript sitting on the client side slowing down their machine and being potentially incompatible with their devices (your widgets don't show on my phone, by the way). Client-side scripting has a purpose, but displaying lumps of data isn't one of them, especially when it can be done on the server side. Yes, I know that I use Grazr, but avoiding client-side scripting is good practice wherever possible.

Widgets serve only a very introductory role to the web of data. What they point to is something important - there are Internet applications that need customised interfaces and applications. I'm waiting for someone to build a Twitter client for my Palm Pilot so that it uses my GPRS connection rather than SMS to send and receive messages.

What we need to do is think of the different types of content delivery and then try to imagine data platforms that will work with them. RSS serves a constant stream model very well. Widgets serve a different variation on that - a "grab the latest" model, not a "grab everything" model. If you miss the weather yesterday, you don't want to hear about it today. Imagine other models of content distribution and then build applications to meet them.

The introduction of 'grey' answers in to the application space will be one of those things. Currently, it's a choice between "friend" and "not friend", "subscribe" and "unsubscribe". Instead, we'll be making reference to more nuanced judgments - "this person is interesting but I'm not sure about them", "I like this person, but I'm not wild about their friends", "track this, but only disturb me if something really important happens". Widgets - basically dumb lumps of JavaScript - will need to work out where they are going to stand on the data web.

Widgets aren't revolutionary. They're actually surprisingly non-revolutionary. They are at the very mild end of a shift which could be revolutionary - which is the idea of a data web. I was thinking yesterday of how interesting it would be if all companies made all their data available (obviously with some kind of open trust mechanism in place). The number of utterly useless people ('human resources', 'customer services' etc.) who would lose their jobs and be replaced with Perl scripts is quite unimaginable.

Imagine, for instance, if I wanted to go and buy something. I tap in to a little device what I want to buy and how quickly I want to get it. It then goes and pings the shop databases, gathers back prices, then picks out which ones I can travel to (it would know already that I don't drive plus the schedules of people I know who might be able to pick it up for me), works out the best method of transport to get there, checks that what I'm buying is compatible with devices or processes that I already use, checks whether it would be possible to write it off against tax, makes available money in to a secure trading account to pay for it - following a set of rules about where to take money from and who to borrow it from (if anybody). At each waypoint on my journey, it would tell me the most cost or time efficient way of continuing my journey (the bus isn't running, so get a cab or walk). If I get back home and find it doesn't work, it not only sends out standard complaints to all the relevant people, it helps organise a return trip to the customer service department and manages the financial transaction of refunding the money. Couriers wouldn't try to deliver when nobody is at home.

Think about this from the perspective of health. My medical records are on pieces of paper in an office in my home town. Except my dental records. They're everywhere, because I use the dentist far more often than I go to the doctor (I've had orthodontic treatment and a whole load of other stuff too). My dentist has changed numerous times - one retired, the other one went private, then I was living elsewhere for a year, then I came back and the previous guy had left, and then I found out I needed emergency treatment and I didn't have private insurance so I had to find an NHS dentist who would do the treatment. Each of those has a seperate set of pieces of paper in their office.

There are two ways of putting that in a database - either you put it all in a giant database which you then try to hook up to all health providers in the country. This is big and slow and bureaucratic. Why just within the country? What happens if I need emergency treatment while abroad?

The other way is you put all this data in the hands of the individual on an individual database. Each person would pay a small amount each month for the hosting space for their database, and there would be charities and subsidies for people who can't pay. Whenever a business or organisation or the government wanted to do something, they'd send a request to your database. You could set whatever rules on there you want. If you don't trust your doctor, you are only one click away from denying him access to your medical data. This becomes portable, cheap to provision for (so long as the database could store triples, it's ready to go) and virtually hack-proof. Why? Well, if you set up a national healthcare database, that's one huge honeypot to smash your way into. Once you are in you have the private information of sixty million people. But if you break in to my database, you only have access to my data. Okay, you'd know all the details about what DVDs I've rented and you'd know about the type of braces that I had when I was fourteen. But nobody else would be affected.

The web of data goes much further than widgets. They're an early fork in a very long road.

|

Inorganic brains for organic produce Permanent link to this item in the archive.

Peter has pointed to these statements by David Milliband. Apparently, something inside Mr. Milliband's brain clicked and he is now actually acting rather sensibly. The word "organic" don't mean a thing. I've been drinking lots of chemicals this morning. Quite a lot of dihydrogen monoxide combined with citric acid, glucose, fructose, sugar, ascorbic acid. Pretty scary, huh? Yeah, my pineapple juice does sound scary when you break it down in to it's chemical components. The organic movement are pushing superstitious vitalism on consumers while people outside of the Western world suffer from malnutrition.

Save your money - buy the non-organic food and drop the premium you pay to these phonies in to a charity box to install clean water wells in Africa. If you are buying 'ethically', actually try and figure out whether your ethical action does anything. Organic food does nothing except makes your conscience feel clean - without good reason.

These same holist-vitalist asshats are actively preventing experiments with genetic modification. We already experiment in a not particularly intelligent way with plants - we subject them to irradiation in order to force them to mutate and then keep the beneficial mutations. But scientists aren't allowed to directly modify the genome in order to produce modified species.

Organic food is unethical. I refuse to pay extravagant amounts of money to subsidise farmers for producing a product that adds no extra value. With fair trade, you pay extra to actually help people in need. With organic food, you pay extra to push forward a pseudoscientific agenda that harms the rest of the world by preventing innovation in agriculture. Will my parent's generation (the primary consumers of organic food) get that message? Unfortuantely not. The social pressures of the book club and PTA are too much to deny the holy gospel of organic foodyism.

Produce the food locally, make it taste great, try and benefit the poorest off in society in the process. The 'organic' movement adds nothing to that.

|

HomeTom MorrisOpiumfield

Last modified: Sunday, August 26, 2007 at 11:22 AM.

January 2007
Sun
Mon
Tue
Wed
Thu
Fri
Sat
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
 
Dec   Feb

This is my old blog. Please visit the new one.

Send me a voice message via Odea PayPal
 Subscribe

My podcast (RSS)