Radar Networks opens Twine to the world with version 1.0

Less than a year after its unveiling at last year's Web 2.0 Summit, and a mere eight months after closing a $13 million Series B funding round, Radar Networks' Twine today moves out of beta as a 1.
Written by Paul Miller, Contributor

Less than a year after its unveiling at last year's Web 2.0 Summit, and a mere eight months after closing a $13 million Series B funding round, Radar Networks' Twine today moves out of beta as a 1.0 Release, open to all comers.

The 'Semantic Web' with which it was so closely associated (an association that has attracted flak) at the outset is almost nowhere to be seen, and this is bound to incite a further round of criticism, nay-saying, and mud slinging. What many of those critics forget, though, is that this is quite explicitly billing itself as a consumer application.

If my mother, my brother or my children can 'see' the Semantic Web, it has failed - big time.

Talking with Radar Networks' CEO Nova Spivack ahead of today's launch, he was keen to stress the

"big focus in this release upon usability."

Twine is billing itself as

"a place to keep up with your interests"

A company briefing document suggests;

"Radar Networks is a venture-funded startup focused on 'interest networking' – the practice of connecting with others around the topics we care about most.

If a social network is about who we are interested in, an interest network is about what we are interested in.

The company’s first product, Twine, is the logical next step beyond a social network – It connects people around the content they find interesting."

The team has invested a lot of effort in easing new users into Twine, and streamlining workflow once inside. There's still some work to do; frankly the interest feed is a pain to keep 'caught up' when you are subscribed to a sizeable number of twines; especially given users' penchant for cross-posting items to multiple twines, most of which you're also likely to be subscribed to. It's fixable, though, and this release is a significant step forward from earlier iterations in the beta process. [Update, 0017 PST, 21 October: responding to this post, Nova Spivack tells me that enhancements to the interest feed will be rolling out over the next 24 hours.]

Twine 1.0 is definitely noticeably faster than previous releases, with Spivack suggesting that the site was

"1,000 times faster than last week"

Writing late on Monday evening in the UK as the Twine team add the last lick of Californian paint, I am still seeing the site occasionally slow to a crawl, but I'm not going to hold that against them. The site is technically still in beta as I write. If it's still slow after this post sees the light of day, then I'll complain.

As with so many 'social' sites, it can be difficult to clearly communicate value to a new user. Indeed, for many sites there is no value for a new user until they have invested significant effort in manually constructing their network. Twine is a little different, and the new signup screen encourages prospective users to enter some of their interests before actually signing up.

Straightaway, a prospective user is able to discover information that others have added to Twine. Behind the scenes, the semantic technologies that make Twine work are doing what they do best; without the user having to concern themselves. If interested in what they see, the visitor is then able to work through a straightforward sign-up process and begin to realise the additional benefits of connecting to other members and registering with subject threads ('twines') of interest in order both to post material of their own and to receive updates from other members of the twine.

Once a member, there are two main - linked - functions within Twine. The first is tracking and commenting upon content posted to twines by other people, and the second is bookmarking content that you discover out on the web.


New items and comments posted to twines of interest are visible in the Interest Feed that greets you each time you log in to Twine, as well as in optional email alerts, RSS feeds and the like. On the basis of user behaviour, Twine will also begin to recommend people and twines that may be of interest, and Spivack notes that an upcoming release will greatly enhance this feature by explaining why the recommendations are being made. In the same way as you can with Amazon, it would be useful to be able to declare non-interest in these recommendations, so that particular people and twines do not recur.


A simple bookmarklet enables Twine users to post items of interest into Twine. Around 50% of all twines are private and restricted to an individual or a group. The rest are public, and open to be read by anyone with a web browser. This example shows the result of trying to submit a page from the BBC. In this case, all of the text has been auto-generated by Twine, and all that I need to do is select the twine(s) and/or people with which I wish to share, and (optionally) add a comment of my own before saving. The result is as below (click to see the real thing), where you can see Twine's power beginning to express itself in the series of facets and tags down the right hand side;

Items can also be submitted by email, and in an upcoming release Twine will be able to directly consume RSS.

During the beta programme, Twine has grown in size, complexity and utility. According to Radar Networks, they have seen 500,000 unique visitors during the beta, 50,000 of whom are described as 'active' in adding over 1,000,000 items to 20,000 twines.

More than half of those users originate outside the United States, and they tend (around 75%) to be male, well educated, comfortably employed, and between 31 and 50 years of age; a pretty good demographic to monetise, in other words.

Turning to monetisation, Spivack suggested that;

"social networks do not monetise because you're basically there to communicate"

Twine, on the other hand,

"is different, because you're there to keep up on a topic. [You might therefore welcome] targeted advertising around that topic"

Spivack reports that the company is actively signing up a variety of partners looking to benefit from Radar's patent pending recommender system, and he expects the first adverts to begin to appear early next year.

Other features due in enhancements that are expected to roll out each month from now on include the release of an API and far more investment in making existing semantics or structure work that much harder.

As soon as November, for example, Spivack suggests that the company will release a new mining system that will use Natural Language Processing (NLP) to do a far better job of parsing information from pages that Twine users bookmark into the system.

During 2009, Spivack suggests that we will

"start to see the other 90% of our Platform."

Into 2009, users will gain the ability to create far more item 'types' (events, product data, etc,) and a public API that's already operational within the company will include capabilities such as the import of existing third party ontologies.

The API is apparently fully RESTful, and

"similar to Freebase."

One (unnamed) partner is using the API to integrate Twine into Microsoft Office. Powerset, anyone?

Despite alluding to similarities, Spivack was quick to stress that he has

"No interest in doing what Freebase is doing... building an encyclopaedic view of the world. [He would] much rather make it easy to pull Freebase data into Twine."

Twine has come a long way since I first saw it. As with all complex applications, some rough edges remain, but there is certainly enough utility for the avid hoarder of 'stuff' to get to work populating their twines today. Is it for everyone? No, probably not. But for all those people who want to track a professional subject, a hobby, or their favourite band, there's something here. For people who want to do those things, and who see the value of doing it along with similarly enthused individuals around the planet, there's even more.

The Semantic Web's technologies lie behind Twine. Sometimes you can almost see that, if you know where to look. Often you can't. Given Spivack's ambitions for 2009, the semantics in Twine are going to get a whole lot richer. The trick will be adding that richness whilst ensuring that the application continues to get demonstrably faster and more usable at the same time.

See Radar Networks' overview of Twine functionality in this short video, and listen to Radar Networks' CEO Nova Spivack talking to me about the Semantic Web several months before Twine was announced

Nova Spivack will be joining October's episode of the Semantic Web Gang to report on the first week of full operation, and to discuss the company's next moves.

Editorial standards