There are my notes on Lachlan Hardy’s (@lachlanhardy) session The Open Web.
What is the open web?
It’s a buzz phrase as much as a real phrase. Definitions and usage vary (some of them are a bit vacuous).
“My” definition: it stems from a common philosophical approach. It’s a shared interest in the progress and development of the web.
Open Web group on Google Groups. Diverse interested and membership. Have come up with four points that, they believe, are required for something to be “open”:
- An open specification, providing the freedom to implement. Unencumbered by patents and licensing.
- One (preferably more) open source reference implementations.
- Supported by more than one vendor.
- Public involvement in the evolution of the specification.
Some of these four points are contentious.
The big concepts and the philosophical concepts are hard and, as such, they’re evolving. This isn’t the definition he was using last year and will likely change.
Small pieces
Web standards
HTML, CSS, JS, DOM, Atom.
All have a baseline of useful support across major browsers.
XMPP
IM protocol that isn’t ICQ, MSN, AIM, etc. It’s an open standard used in Jabber, GTalk, etc.
They’ve been approved by open standards organisations (W3C, IETF, etc.) and aren’t controlled by a single entity.
Open specifications
Specifications openly available, implementation, but not standardised.
OpenID is a good example. Anyone can read and implement the specification.
Another good example is the microformats movement. Anyone can read and adhere to the conventions for semantic markup of events (hCalendar), business cards (hCard), etc. Anyone can implement an applications to read and write data according to the specification.
OAuth is another. Secure API authentication from desktop and web applications. Granting someone else (a 3rd party iPhone app) access to “something” (your account) on a service or server (Twitter).
WebFinger is finger over HTTP. Specified via implementation. Google and Yahoo! both support it (opted in). Use it to connect an email address to an OpenID.
Use OpenID to authenticate, OAuth to authorise, etc.
Activity Streams extends Atom to describe the activity performed by a user.
Portable Contacts provides a common access pattern and schema for contacts. A single unified way to transfer contacts, etc. between services (e.g. import “friends” from GMail when signing up for Twitter). Build around OAuth and vCard (hCard). BUild on a schema from Open Social.
This is the reason we love the web! No-one has built a single monolithic entity. Instead, it’s being built by unrelated people and orgs as a modular jigsaw puzzle. They are design and intended to be used together. They’re not standards, but they are open. OpenID, e.g., has a legal foundation, Microformats are sorting their IP out (victims of their own success), OAuth is forming and covered by the Web Standards Foundation (which is open to anyone else).
Cold Hard Cash
The goal is cheapter and better. Lachlan is not an OSS fanatic (he used to be a .Net developer). This isn’t an abstract philosophical ideal. These technologies are a better business choice as well.
More people have looked at and worked on open standards than the proprietary options. And you didn’t have to pay any of them. All of the work and development and progres has been done (and paid for) by these other people.
Open Architectures
Much less clearly defined.
URLs should be readable. We can look at them and understand them and modify them. Using verbs: http://omniti.com/does/degin/, http://omniti.com/helps/national-geographic/. Much more readable and usable (and great for SEO). But this requires planning and design.
Our information architecture should be reflected in our
APIs don’t need to be enourmous collections of methods, etc. (Ben Schwarz, Smoke) Use REST, it’s simple, straightforward and everyone understands it.
Example Time
- Simplicity: What can you do with 140 chars? Heaps!
- Ubiquity: Simplicity make ubiquity easy.
- Open API: Every one can access it (just fill in the form and get a key). They’ve got an enourmous ecosystem as a result.
- Microformats:
Gnolia
- Died and came back with a rebrand.
- Outsourcing identity: you don’t create an account with them, you sign in with an external identity provider (G, Y!, OpenID, Wordpress, AOL, LJ, TypePad, etc.)
- Microformats: hAtom, hCard, etc.
OpenID
- Literally dozens of OpenID implements to run your own server (or consumer)
Use the Open Web
The web connects stuff – people, places, services, information. It’s not a monolithic stack, but a series of components and communities which interrlate in new, interesting and changing ways.
Some things we should be doing more of:
- Use microformats in your content and it’ll be inherantly more accessible.
- Make URLs accessible and hackable.
- Put all that junk stuck in PDF and DOC and RTF into an HTML web-page (and microformat it).
- Offer relevant web feeds. An Atom feed is not just for blog posts and your latest news. They’re for anything that changes. isitchristmastimeyet.com (“NO”) has an Atom feed.
- Offer an API: there’s no reason not to.
- Offer web services. Your site shouldn’t just be a web-site. Should integrate with other services (“I want a Japanese restaurant here, now”).
- Offer and accept OpenIDs. Google, Yahoo!, etc. are providing OpenIDs, but most of them won’t consume. Once your users have identitied, you know things about them and offer them more and better services based on that knowledge.
- Offer OAuth access to user’s data. It’s a bit of a pain to implement, but well worth doing. Who are the third-parties? Some of them will be in your community.
- Personalise feeds. It’s not enough to offer a service to an audience. We should be aimed to provide services for Joe Bloggs, and Jane Doe, not “person in the English speaking world.”
Some of these things are hard and require significant effort, but there are many low hanging fruit, and every one who gives it a go and provides feedback can help make things better and stronger and easier.
Questions and Answers
What about privacy? Insurers, etc. are already collecting and mining and abusing this type of data.
The response from techies is invariably: that data is already available. These “open web” technologies are making things slightly more formal and make it easier, but it’s always been possible and it already happening. Just look at Facebook’s cooperation with third-party commercial interests.
What about multiple identities? Many of us have seperate identities for seperate interests and activies.
Many OpenID implementers, for example, support multiple “profiles” and allow you to choose between them as needed.