Two geeks who have both founded incredibly successful social start-ups talking about funding, coding, user experience, killer functionality, luck, social networking, scalability, how to define your competition, the difference between building a product and building a company.
Kevin Rose of Digg et al talks to Kevin Systrom, founder and CEO of Instagram.
They obviously enjoy each other’s company and the content is clearly better and more revealing for it.
The interview ends with a deceptively simple piece of advice. Namely to focus on solving problems rather than focus on technology.
“Far too many start-ups are technologies in search of a problem.”
As part of our ongoing work with Maxxium UK, we have been working with their wonderful Mixxit team to redesign and develop a new website.
Mixxit is the brainchild of celebrity mixologist Wayne Collins. Along with his colleagues Andy Gemmell and Patsy Christie they set-up Mixxit initially as a training and education programme, with the aim of inspiring bartenders to make quality mixed drinks and cocktails. But they are so passionate that every drink can be perfectly made whether you are in the bar and or at home with friends, they wanted to spread their knowledge.
Mixxit is unique in that every drink can be perfectly made in 5 easy steps. So how could we give people access to over 300 delicious cocktail recipes as well as offering guidance from three of the most prominent drinks experts in industry?
Welcome to the Interactive Cocktail Mixer; you can search by spirit, occasion, glass, mixer or even colour in your quest to find the perfect drink. The cocktail mixer is built in Flash but utilises the site’s search functionality to find the right cocktail. Happy mixxing!
EdTwinge is a service that monitors tweets about all the acts at the Fringe, analyses them for sentiment and creates a “karma” score for each one. These karma scores are used to create league tables for each genre of show at the Fringe.
So the audience becomes the critic in 140 characters or less.
It means that you can use the wisdom of the crowd, people like you, to decide what shows to see rather than the more dubious and subjective wisdom of a few “professional” critics.
In its previous guises the podcast has enjoyed between 400,000 and 2,000,000 iTunes downloads, and we’ll be working together to add value to each other’s offering and build each other’s profile. For instance the acts that are booked for the live podcast recordings will be influenced by the EdTwinge league tables.
Hopefully the relationship will help EdTwinge grow again in 2011. The site had 192,029 visits from 131,887 unique visitors during the 2010 Fringe.
This digital lark can be deceptively simple. For instance…
The new IRN-BRU ad is based on those olde seaside cut-out boards. Holiday makers would stick their heads through the cut out holes and have a (sometimes saucy) souvenir snap taken.
In the ad, once the live action protagonists put their heads through the holes, they become embroiled in an animated battle for possession of a can of IRN-BRU.
It’s a simple idea through which IRN-BRU can hopefully claim more than its fair share of summery goodwill.
And the idea for digital amplification of the campaign is equally simple.
You, yes you and a pal, can stick your own faces in the cut-out holes and star in your own version of the ad.
“Simple but not entirely original” I hear you cry.
And, yes, it does sound a bit like the kind of thing that Evian et al have done before.
But we took things a bit further than crudely pasting a still image into a moving picture film.
“Star In The Ad” isn’t just about being in the ad, it’s about performing in the ad.
Only we do the performing for you.
You, or you and friend/partner/lover, upload your photos and our application does the rest in terms of making you “act” in the film.
And here’s the science bit. (Art and science bit actually).
More to the point, here’s why this particular digital lark is deceptively simple.
How do you add a range of credible expressions to a still facial image and smoothly animate between them?
First up we tried to replicate the effect of the “liquify” filter in Photoshop. This filter effectively turns an image into liquid paint and allows you to fluidly manipulate it.
In Flash we do this with a DisplacementMapFilter. This takes the colour of one image and uses it to displace the pixels in another image.
Allow us to demonstrate with the help of our handsome model Charlie.
Charlie volunteered to have his face mangled (we mean displaced) and, given that he was already sporting a Movember ‘tache and a beanie hat, we figured we couldn’t make him look much worse…
On the face of it (geddit?) DisplacementMapFilters looked like a promising approach. We could raise an eyebrow or create a passable grin. But there were some problems challenges.
Setting up the colour maps was incredibly time consuming, not to mention very easy to mess up. And the maps were monstrously difficult to tweak.
BitmapFill is a relatively recent addition to Flash and it’s the basis for most of the 3D stuff you see in Flash at the moment – Papervision uses it heavily for instance.
Instead of attempting to directly manipulate an uploaded image, the new approach involved the creation of a 3D model (or to be more precise a “flat” 3D – i.e. 2D – mesh) and the use of Flash rendering to map the uploaded facial image (now treated as a texture) to said model.
We created a plain, expressionless mesh onto which the original uploaded face images would be mapped. And we then created a series of distorted meshes corresponding to the variety of expressions required for each character to “act” in the ad.
Once the original face image had been mapped to the expressionless mesh we used good old Flash “tweening” to animate the image from there to the alternative mesh expressions, rendering the face at each frame en route.
At first we tried a simplistic approach to determining the mesh points, using ASCII (plain text) data to track the displacement of each vertex within the mesh. Unfortunately this didn’t give us the requisite degree of accuracy.
So we moved to the full 3D model approach (2D plane saved as a 3D model) and used the COLLADA 1.4 format to export from Blender (3D modelling tool) and import into Flash.
(Aside : we originally used C4D for the mesh manipulation, it being our 3D application of choice, but found that the COLLADA extension didn’t behave correctly in Flash and so we moved onto Blender, having tried and tested its COLLADA support on other Papervision projects.)
Whilst we created the first version of all the expressions in Blender, we actually built our own helper tool to make tweaking and fine-tuning quicker and easier. In fact for quicker and easier read amend in real time. Within the helper tool you could tweak an expression, press a button, assess the change and , if happy, save the updated mesh out to the main Flash application.
Grab from the homemade helper app.
The helper tool proved to be invaluable in terms of allowing us to create rapid iterations of the various expressions as we moved through the rigorous quality control process leading up to launch.
Which brings us onto the issue of eyes (the windows to the soul and the source of a great deal of soul searching and heartache for this project).
No amount of tweaking of meshes or uploaded images would give us realistic looking eye movements, and so we ended up having to cut holes in the mesh and place separate, animated eyes behind it (as hinted at in the helper tool image above).
That’s the somewhat geeky story of the back end of this campaign.
But the back end could only work its magic if we got the front end user experience right.
Given that all expressions would be the result of mapping between various meshes, and that all said meshes are based on an “anchor” expressionless mesh onto which the original face images are mapped, the quality of the end result is hugely dependent on the “quality” of the uploaded images. Basically, the closer the uploaded face images are to a full frontal, expressionless image the better.
So designing the front end user experience wasn’t just about reducing friction to maximise participation and minimise mid-process drop-out rates, it was about “gently” encouraging users to upload the right kind of images, be that via webcam, from file, or from Facebook.
At the time of writing it’s a little too early to say definitively how well this is all working, but early signs are encouraging. Watch this space.
(Clever coding and design stuff done by Fraser McCormick, David Hartmann and Charlie Bell.)
I finally had a chance to play with my new Storify beta invitation.
Storify makes it easy to create stories from the social web. At least that’s what it says on the tin.
And, having briefly road-tested it with the somewhat frivolous story below, I’d say that it delivers.
You can easily pull in content from all the major social platforms, plus Google and RSS feeds. You can embed links, annotate the content to add colour and context, and select a master image to thumbnail the story when you publish.
To someone who has laboriously constructed picture essay blog posts about Facebook and Twitter in the past, taking multiple screengrabs and resizing to fit the blog column width, Storify looks like a godsend.
Once you’ve finished the story, the application gives you the option to notify the people whose (Twitter) content you have featured by tweeting the link to them. A nice courteous touch that also has built-in word of mouth potential.
The only minor quibble from this first trial is that when embedding the story in this blog, Slideshare-style, the formatting isn’t 100% perfect. But that’s what being in beta is all about.
I hope to give it a more thorough run-out in the near future.
A few clicks on from another excellent Mashable update and we’re demonstrating and debating our own variance in social behaviour with chat show style banter on Google Buzz. Here’s the discussion thread…
Following an email around the office this morning, investigating what our individual thoughts were on Google Buzz, I received a few pretty cynical responses. To be fair, many people are still recovering from Google Wave: asking where and what now?
For me, it’s a bit different. Google is half my social equation. I have a Gmail account and am also a member of Tech Meet Up, Mobile Monday Edinburgh and Girl Geeks. TMU is a Google group, as is Mobile Monday and for Girl Geeks we organise information around Google Sites. I have a number of files in Google Docs, plus I’m in and out of Google Maps all the time. It’s fair to say I like Google.
Not everybody feels the same though. Here were a few thoughts from other Blondes about Google Buzz, starting with our tech team:
“It’s like twitter, but without any clients and unable to tie into any other of the microblogging/social media things…
It does follow the two people I have set up in googlemail already though.
Sadly, until they create some reason for it to exist my thoughts are:
“Just what the world needs, another twitter-clone.”
Oh it has “like” from facebook, which twitter haven’t managed to add yet.”
The bandwagon of Buzz bitchin’ rolls on…
“If everyone starts using it then it becomes really relevant. Haven’t seen a killer feature yet…
Google have been, shall we say, intermittent in their marketing of services so the jury is out about how big it will become.”
Having said that the current debate (going on around me while I’m typing this) is that if you are a heavy gmail user (I am) then it does appeal -as long as the take up is good.”
And then the rest of the planning team jump in….
“Google Buzz – initial reaction = “groan”.
Groan because as a digital marcomms professional I’ve got no option but to try it out, when my gut feel and the input (via Twitter) from people I trust are telling me that it’s going to be no great shakes.
Groan because I resolved this year to add more depth to my social graph (more emphasis on RSS, blog writing and commenting), and this feels like more breadth. That’s my personal take on the whole signal to noise issue.
Groan because there’s this stool called RSS, and there’s this other stool called Twitter, and I can’t see a valid role for anything between those two stools.
Groan because I only ever use Gmail for opening Twitter accounts. I really really can’t be arsed to have another following mouth to feed.
So I’ll watch with interest how it develops but at the moment it’s not even coming out of the box for me. “
And the tech team is still going….
“It just seems way to complex and unfocused.
Twitter
Share and discover what’s happening right now, anywhere in the world.
Facebook
helps you connect and share with the people in your life.
Simple. I get it.
Buzz
Go beyond status messages
Share updates, photos, videos, and more.
Start conversations about the things you find interesting.
Try Buzz in Gmail. (why do I have to do it in gmail by the way?)
You lost me at “Go beyond status messages”. Er.. not so sure I get it.
Really bad design as well. I find it really hard to scan. Twitter is so easy to just fly through and pick what’s interesting. Buzz is a mess and feels horrible to read.
Google seems to have forgotten their design roots. A million miles away from one input box to search shit.
Social network app fail.
A lot cleaner in t’iPhone though. Bordering on being useful.”
Before finally concluding…
“My friends are on Facebook (for the most part)
My friends aren’t on Gmail (for the most part) and if they are they’re not updating their status there.
So, for social/picture sharing stuff = Facebook.
I get interesting web/marketing/wotsit links by following people and searching on twitter.
I don’t get that from Buzz (yet, since there’s noone on it)
So for interesting links/finding out what random “famous” people are upto/whatever = twitter.
Buzz is only as good as the people using it – and currently noone is.
I expect it to go the way of Google Lively, to be honest. “
It appears I am the lone voice of Google Buzz support here. I don’t think this product… this service… from Google is anywhere near complete and of course I’m aware of the privacy issues, which have been flagged up; but I need something that bridges the gap(s) between LinkedIn, Twitter and Facebook.
Social networking online isn’t this simple. I don’t want my engagement with professional communities to be confined to LinkedIn. I want meaningful, human-to-human knowledge sharing, which goes beyond the offering of Twitter and I certainly want a richer experience than Farmville updates on Facebook. I’m not arguing that Google Buzz is the solution, but given the incredible insight Google now has into our online behaviour, I’m holding out for some improvements first, before I search for a conclusion. Here’s to growing out of Facebook and Twitter…
As well as some impressive examples of digital thinking in action, there were some interesting themes for the afternoon.
Routes to attention
Ideas judged against how they…
Get attention.
Keep attention.
Turn attention into value.
Digital Business Models
This was a major theme from Tom’s presentation. As well as investing in sustainable non-broadcast public service initiatives, 4iP is actively experimenting with alternative business models for online start-ups. As he said, if Channel 4 finds an advertising revenue model challenging with 10 million monthly uniques, 4iP will take some convincing that an ad-only model will work for a start-up.
Accountability
Matt talked about a 9 point (3×3 square) grid against which each project is measured.
The objectives of getting attention, keeping attention and turning attention into value run along one side. Along the other are three types of audience engagement : visitors, fans, contributors/distributors.
Using this grid, objectives and metrics can be entered in each of the nine squares. A simple but highly effective way not only to retrospectively report on a project, but also to run actionable ongoing diagnostics during a project.
All in all a useful and thought-provoking session.
EdTwinge is a Twitter-based, crowdsourced, realtime Edinburgh Fringe rating application.
It monitors tweets that mention any of the acts at the 2009 Fringe Festival and/or the most commonly used Fringe hashtags. It then matches the content of these tweets against an extensive database of positive and negative words and phrases.
The site publishes two scores for each act at the Fringe. The first is “Noise”. This is simply the number of tweets that have mentioned the act in an Edinburgh/Fringe context. The second, and more important, measure is what we call “Karma”.
A full, geek-friendly explanation of how Karma is calculated can be found on the Edtwinge site. But basically it is a measure of the net positive sentiment about each act, which is based on a robust statistical analysis. This statistical analysis ensures that the Karma score is as reliable as possible. For instance an act mentioned in, say, five tweets, all of which are positive will have a significantly lower karma score than an act mentioned positively in 60 tweets out of 70. Phrases are prioritised over words, so “shit-hot” would be correctly identified as a positive statement for instance.
The top ten tables for the Fringe as a whole, and for each genre of show, are primarily derived from Karma. Noise only comes into play if two shows have the same Karma rating. In this instance the act with the higher noise score would rank above the other.
The site also allows the user to search and view karma and noise scores for any act, and to view in chronological order the verbatim tweets that underpin these.
EdTwinge is also the result of a garage-band style collaboration with some very talented, sparky and creative people.
Mike Coulter – social media exponent at Digital Agency (the original idea was his).
Andrew Burnett – another social media exponent and expert traffic driver.
Jim Wolff – digital misfit (his words not mine) who joined Leith part way through the project.
There is no paying client behind EdTwinge. It’s been a fun and fruitful diversion from the day job for those of us lucky enough to be involved. And what we’ve learned about tag-team style collaboration, baton-passing Twitter account shifts, and fleet-of-foot digital seeding and amplification has been as valuable as the technical, under the bonnet of Twitter stuff.
Early days yet (3 and a half days into the Fringe at the time of writing), but there has already been a significant amount of positive commentary and the site appears to be performing well. A big thank you to our friends at Stripe for securing some really excellent profile for the project.
I’ll post a more in-depth analysis of results and learning in a couple of weeks but average time on site is currently running at over 4 minutes on the back of an average of 5.22 pages viewed per visit according to Google Analytics.
Follow us (EdTwinge) on Twitter for regular updates. And/or embed your own EdTwinge Top 10 widget, like this one that we prepared earlier…
I was fortunate enough to see Guy Kawasaki on his feet yesterday, talking about the Art of The Startup at Edinburgh Informatics. He’s a Silicon Valley venture capitalist and ex Macintosh brand evangelist. He also part of the team behind Alltop, the online magazine/news aggregation site that answers the question “What’s happening?”
Over the course of an hour he talked through 10 themes that summarise his view on what makes for a successful start-up. There’s little point in me recounting in detail what he said because Ewan McIntosh was live blogging and has pretty much nailed it on the 38 minutes blog.
So enough about the content, what about the delivery?
This was most definitely a gig, and had several things in common with the musical kind (all good).
1) VIP, access all areas style wristband.
2) Guy played a great set. It was clearly not the first time he’s spoken on this topic, but his familiarity with his material made for a compelling 60 minutes. He was most certainly not going through the motions.
3) His material was good. Interesting points, well made, with examples that were new to everyone. Peppered with candid and personal anecdotal asides. The equivalent of a continuous series of crowd-pleasers with no duff tracks off an unfamiliar new album.
4) If his ten themes were the equivalent of ten songs, then the linking banter between them was spontaneous and tailored to the audience. He wasn’t afraid to go off piste for a while.
5) He’s a natural front man. OK so there was no band behind him but he owned the stage.
6) His slides were his supporting act and he was the headliner. Like all really good presentations, this was about him not his Powerpoint. How many times do you see the reverse in action?
7) Encore. Courtesy of Mike Coulter in the Q&A, we were treated to an encore in which he speculated on the monetisation of Twitter. He suggested that Tweets could be treated like text messages. Say 250 free Tweets per month and $5 for more than that. The audience seemed to react well to this idea.
