« swipe left for tags/categories
swipe right to go back »
At Foundry Group, one of our investment themes is Glue. We’ve done a handful of investments in this area, including Gnip. Since Gnip’s launch last month, it’s been put into production in a number of cases – some obvious, some subtle. Part of the fun is watching the adoption of it evolve rapidly as we continue to build out the core capabilities of the what Gnip can do.
I had a long conversation with a VC I work closely with about the value Gnip ultimately provides to its various constituencies (data providers, data consumers, and end users) and how / where it expects to get paid in the long term. During the conversation, we covered a number of different potential areas, but I realized that my thinking could be much crisper. That’s normal for this stage of a startup as Gnip is still very early stage (we’ve done one seed round of investment and are gearing up for the next financing) but the exercise of defining a clear business endgame (vs. just a technology endgame) is extremely helpful and self referential, as it creates more focus on what we should actually be building.
There is nothing quite like an example. Yesterday, we had the TechStars 2008 Investor and Demo Day. EventVue – one of the TechStars 2007 companies – provided the online community infrastructure for the event. They automatically extracted all the data from the registration system and build an online community. As part of this, members of the community could add their twitter account and – if they had already been a member of another EventVue conference community – like me – would automatically have all their information already in EventVue and wouldn’t have to do anything.
The then created a techstars08 twitter account. This rebroadcast all the tweets from anyone at the event that had a twitter account set up in their EventVue profile. However, rather than writing the polling software to Twitter to continually check for updates in the twitter stream, the used Gnip for this.
EventVue had a data set (I don’t know the number – but lets say it was 100 userids) of twitters at the conference. They wrote a tiny piece of code that monitored Gnip’s twitter notification stream. Whenever someone in the set of 100 usersids appeared in the twitter notification stream, EventVue’s handler then queried twitter for that one discrete piece of data and then rebroadcast it on techstars08.
This took a huge load off of Twitter. It was much easier code to write for EventVue. It created a virtually real time twitter rebroadcast stream. I’m sure I’m missing at least one of the technical nuances – hopefully the guys at EventVue will write up a deeper post on what they did, how they did it, and why it was valuable to them.
Update: Josh Fraser, the co-founder and CTO of EventVue, has posted How Gnip rescued us from our twitter nightmare.
Look for plenty of more thinking out loud from me on our Glue theme as we bring some of the investments we’ve made into sharper focus.
In the past six months comments have moved to the forefront of the discussion around user generated content. While the various new commenting systems that have emerged have played a part in this, I think the broad activity around systems that enable small bursts of user generated content (Twitter, BrightKite*) and systems that aggregate a wide variety of user generated content (FriendFeed, SocialThing*) are playing a huge role in this and more "comment-like" data is being generated all over the Web.
One of the investment themes I’m most fascinated with right now is the one we call "Glue". We’ve made a handful of investments in the Glue theme at Foundry Group including Gnip, AdMeld, and Topspin. We’ve also been working with our good friend Eric Norlin – the creator of the Defrag Conference - on a Glue Conference.
I’m always looking for great, simple examples of Glue and I found one accidentally the other day. I put up a blog post titled Brilliant Op-Ed Crushing McCain On The Economy. I posted it on Sunday morning and then went out for a two hour run. I came back to about 20 comments on it in my inbox. Even though the post was done on my blog, I noticed the comments were from FriendFeed accounts being emailed to me by Intense Debate.
Here’s what happened. My blog is one of my FriendFeed services. A vigorous debate broke out on FriendFeed between a couple of people. I wouldn’t have noticed it until Monday when I checked my FriendFeed ego feed (I only do this once a day.) However, Intense Debate is "glued" to my FriendFeed account so any comments that show up on a blog post of mine on FriendFeed automatically show up in Intense Debate on my blog. It’s a small feature, but a brilliant one, as it brings the overall conversation associated with my blog post back to my blog where I actually want it.
There are now 46 comments on this particular blog post (unexpected – I don’t write that much about politics and it was a Sunday post.) Most of them are from the FriendFeed discussion, but some are from my blog readers. They are intermixed where I want them – on my blog. Even though they are coming from multiple sources, they persist permanently on my blog due to a tiny feature in Intense Debate.
Now – this is all much too complex still, but it’s why the Glue is so interesting to us. We are continually looking for unnecessary complexity in the metaverse and ways to build really large companies that (a) take advantage of the complexity, (b) simplify the complexity, or (c) both. If you make glue, email me!
* Yes – I’m aware that each of Intense Debate, BrightKite, and SocialThing are TechStars companies from 2007 – and I’m immensely proud of the progress each has made and the fact they are in the midst of what I consider to be a very interesting and vigorous segment of our little tech universe
Loic Le Meur totally nailed why glue matters in his post My social map is totally decentralized but I want it back on my blog. Following is the defining image from Loic’s notebook.
I’m not sure the answer (I want it back on my blog) is correct, but we are in the middle of yet another massive decentralization of data – this time very personal. Loic’s identity and content is spread all over the web (as is mine.) He wants it back in one place – one that belongs to him.
If we only had 10 web services participating in this grand decentralization of Loic, it would be no big deal. But there are now thousands – all which want to be able to play with all the others. With each new unit of data, and each new service, it gets a little messier (and a little more fun.)
In our continuing series discussing some of our investment themes, we’d like to introduce a topic that we’re calling “glue”. Glue is our term for the web infrastructure layer that facilitates the connections between web services and content companies. As this ecosystem becomes increasingly complex and as web sites and web based applications rely on more underlying services, this “glue” layer of the Internet is becoming more and more core to overall web infrastructure.
This is a theme that we’ve stuck with for a while. Seth draws the analogy to EAI and the emergence of enterprise glue in the 1998 – 2001 time frame. We think there are lots of different types of glue, so we’ve once again started a conference with Eric Norlin – this time called the Glue Conference. Gluecon will be modeled after Defrag – it’ll be a place for smart people to hang out and talk about glue, and hopefully stick together (sorry, I couldn’t help myself) as they work on creating some interesting things.
Take a look at Seth’s longer post on Glue and tell us what you think.