What I’ve lost by not looking for investment

Photo by Eric Gjerde

Yesterday I laid out why I'm not looking for investment. I'm comfortable in my choice, but it wasn't easy. The investment process brings a lot more than cash; here's what I'm losing by avoiding it.


Convincing someone to give you money forces you to get very specific, and flesh out your business plan. Defending it against hostile questions makes it stronger. Once you've raise money, having somebody who remembers your promises breathing down your neck keeps you focused. My biggest weakness is that I'm too prone to stay in my comfort zone, working on technical experiments and neglecting everything else you need to build a business. I try to self-police that tendency, but there's nothing like having an outsider's view to keep you on track.


I spend a lot of my week dealing with paperwork, doing website housekeeping, taxes, system administration, arranging travel and meetings, and all the other tasks that could be handled by somebody else if I had funding. I'd have to spend time managing investors instead, but at least the rest of my schedule could be devoted to customer support and product development.

Advice and Influence

Most investors want to convince you they're 'smart money', but from what I've experienced, not many actually bring much to the table beyond dollars. When they do though, it can make a massive difference. They can give advice that saves months of development time, and set up deals that bring in crucial revenue. The key is identifying the people who have the will and the ability to help like that. The only way to do this is by looking for a track record, and knowing entrepreneurs you trust that they've worked with before. A well-known brand is definitely not enough. In my experience the superstar firms are less eager to roll up their sleeves, they're less hungry, though the clout of their name as a credential can make up for a lot of that when it comes to making deals.

Why I’m not looking for investment

Photo by Matt Katzenberger

The short answer? I don't have a good enough idea.

Living in San Francisco, in the current angel investment bubble, with a (tiny-but-nice) acquisition by Apple under my belt, and big plans to change the world, I sometimes feel like a freak of nature because I'm not hunting for funding. Why did I make that decision?

I spent the first year after I left Apple aiming to build a VC-backed company. I had a plausible story, built a complete prototype, deployed pilot installations, and sat in numerous boardrooms pitching partners. The trouble was, it was a terrible idea. I was following in the footsteps of Tacit, VisiblePath and multiple other enterprise email analysis failures. I had no comprehension of the torture of the enterprise sales cycle. The idea of analyzing email fundamentally creeped people out, no matter how strongly I protected privacy. In the classical model of how the process works, I would have got all this feedback from the VCs I pitched to. Aside from being pointed towards Tacit by one partner, I learned everything by painful trial-and-error. VCs loved the idea, they were deluged with email messages and made their livings through their personal connections.

The lesson I drew from that was that I was an idiot. I should be listening to customers and discount what I heard from investors. Get customer traction, revenue, and ideally a profit, and the opportunity would sell itself to potential funders. Obviously that also cut down the range of opportunities I could chase, but what I'd discovered in my unfunded year was that it's possible to build a lot with very little cash. By building fully-functioning prototypes, I could learn what users responded to, as well as what was technically possible.

The trouble is, after a couple of years of operating like this, I'm addicted. I've had many failures, along with a few partial successes, but I'm become so much better at understanding what people need, what's possible, and executing a solution. My failures are becoming faster and far more interesting. I may be living on Ramen noodles, but it's an amazing opportunity to learn. Just the process of fund-raising itself would take a massive amount of time away from that, and then explaining every month where their money's going, why I'm fishing with strawberries, and planning for the next round would take even more. 

That's still a worthwhile use of time if the funding is going towards rocket boosters to strap onto a killer idea. I don't think it's lack of confidence that's telling me none of the approaches I'm working on has crossed the threshold where it's worth it though. There's funding available, I just don't think the strings attached make up for the help. Better to live frugally, consult when I have to, and focus on learning. The more progress I can make, the more I can learn before I need to buckle in to the rocket sled, the better. I'm looking forward to honing my ideas until I can build the real business I dream of, whether that's fully boot-strapped or with the help of investors.

Gadhafi’s Speech Transcript

Word cloud

I was surprised to find that there were no transcripts available of Gadhafi's long speech on Tuesday (February 22nd), so after looking into transcription services I decided to bite the bullet and go through the whole 70 minutes myself. It was a deeply unpleasant experience, but it seems worthwhile to capture his words in an accessible form, since the world is facing so many tough choices in how to deal with him. I don't want to say he's crazy, that seems too easy, but the benevolent mask he obviously tries for keeps slipping as he nakedly threatens his 'people'.


It's a Google document that anyone can edit, so I'm hoping people who know the area much better than me will correct the many mistakes, especially with the place and people names. I also hope it does some good.

I generated a word cloud from the speech, to help me understand what the key obsessions were. People was a key verbal tic, as were 'youth', 'revolution' and 'rats'. Benghazi is obviously preying on his mind too, and he kept coming back to security. I can't claim to make sense of his speech at all, but it feels like the only avenue to understanding what's going on in his mind.

The Portraits of Three Novels

The books I love have their own unique souls, but I struggle to communicate what they are. I can describe the plot, or talk about the themes and the language, but there's an essence that's hard to get across. I find word visualizations a fresh way of capturing that flavor, but I've never been able to get the results I can picture in my minds eye. That drove me to build Wordlings, an experimental service for creating graphical word clouds. Here's what I came up with when I applied it to some of my favorite novels.

Kidnapped, by Robert Louis Stevenson

Carefully disguised as an adventure story, this is actually a beautifully-drawn portrait of friendship. The psychology of the protagonists' relationship veers between suspicion and reliance, anger and affection, in a way it's rare to read. Here's Stevenson's dissection of the tangled reaction to a dependent friend:

The thought of a separation ran always the stronger in my mind; and the more I approved of it, the more ashamed I grew of my approval. It would be a fine, handsome, generous thing, indeed, for Alan to turn round and say to me: "Go, I am in the most danger, and my company only increases yours." But for me to turn to the friend who certainly loved me, and say to him: "You are in great danger, I am in but little; your friendship is a burden; go, take your risks and bear your hardships alone—-" no,that was impossible; and even to think of it privily to myself, made my cheeks to burn.
And yet Alan had behaved like a child, and (what is worse) a treacherous child. Wheedling my money from me while I lay half-conscious was scarce better than theft; and yet here he was trudging by my side, without a penny to his name, and by what I could see, quite blithe to sponge upon the money he had driven me to beg. True, I was ready to share it with him; but it made me rage to see him count upon my readiness.

What jumps out from the word cloud is the centrality of Alan to the book. With his accidental companion David as the narrator, you can see how the interplay between them comes out in the frequent invocation of his name. You can also see how plain Stevenson's language is; "Back", "Down", "Man", "Like". It's why the book flows so easily, despite the depth of ideas it wrestles with. That made the ship that Alan and David were thrown together on feel like the right outline for the cloud, gliding through the water apparently effortlessly, while in fact relying on an exquisitely well-organized web of rigging. 

Pride and Prejudice, by Jane Austen

I'm a late-comer to Austen, only discovering this book when I was a teenager. I would read a few pages to my girlfriend every night, and the book's world became part of our private language. I'd always liked to think I was Darcy, I can certainly match his pig-headedness, but in my darker moments I feared the unwitting ridiculousness of Collins was a better match. The older I get, the more happy I'd be to just achieve Bingley's quiet decency and contentment. There's a lot to be said for nice, but dim.

It's fitting that names dominate the cloud, since the novel is like a pool shark's trick shot, as actions and reactions cascade in an intricate yet believable way. Austen's genius is the way she sketches characters who behave like actual people, but still contrives to arrange events to produce a satisfying outcome for them all. Her dark wit and cutting sarcasm make the happy endings palatable to a cheerful pessimist like me.

The shape to choose was obvious. The 1995 BBC mini-series left Jennifer Ehle indelibly as my personal picture of Lizzy, minxing from underneath her bonnet and curls. I was so hooked, I even forced my old D&D group to watch it, and even they were sold. As one of them said afterwards, "Man, that was the Star Wars of costume dramas!".

Oliver Twist, by Charles Dickens

Despite fears that he's better-known than read, I keep coming back to Dickens. His long-winded style forces me to switch gears, and the barrage of words is like a punch in the jaw compared to Austen's sweet smile while she sticks a stiletto in your ribs. It's effective though, and like reading David Foster Wallace, once you get accept the rhythm the chaos starts to make strange sense. He's unashamedly sentimental, which I appreciate, since I enjoy those sort of illusions even if I struggle to believe them.

The book revolves around power, authority and punishment, with the final chapter taking obvious pleasure in meting out lashes and rewards, sketching the future lives of even minor characters. I wonder if part of the wild success of his books was related to the audience's hunger for this sort of settling of scores? It's the same pleasure I get reading Sherlock Holmes, the relief of believing the world can make sense and is more than just one damn thing after another.

The authority is there in the words that pop out – 'gentleman', 'boy', 'old', 'Jew', 'young', 'lady'. Roles and positions are central to the action. The silhouette seemed to capture the spirit that Dickens was fighting against, self-satisfied and in charge, though I wish I'd found one with the true beadle's hat.


I'm finding this a rewarding experiment, if only for the way it's forcing me to look at my best-loved books in a new light. I'd love to hear about any you're able to create with Wordlings, and what you learned, so do let me know.

Hello Social World

Photo by Dr Case

These days it doesn't make much sense to build a consumer site with its own private account system. If users sign in using Twitter or Facebook, you can have them easily spread the word on what they're doing in your application through tweets or wall posts. The old 'enter details'/'email confirmation'/'click link' process was a leaky funnel that lost potential users too, and was a pain to implement, since automated emails alone are an art in themselves.

I wanted to use this approach on one of my sites, but I couldn't find a good example. Especially in the Rails/Sinatra world there were plenty of building blocks available for doing the individual tasks, like authenticating through Twitter or Facebook, calling their APIs, and storing data, but nothing that showed how to use them together. So, because I needed to build it anyway and it would be a good Ruby learning exercise, I set out to create a minimal-but-complete template that handled user creation, authentication, custom data storage and external sharing. In the spirit of the classic 'Hello World' code examples, here's Hello Social World.

You can try a live demo at http://hellosocialworld.heroku.com/. It's built to do the bare minimum you need for a modern social site, letting users log in, edit some data they own on your site, and then share the results with their friends on Twitter or Facebook. The data in this case is just the user's favorite color, but imagine replacing that with whatever content they actually create on your site; blog posts, photo galleries, comments, etc. It's deliberately left with zero styling, so you can add your own view code.

I was really impressed by the elegance of Ruby and the Sinatra framework. Aside from the API keys, all the code's in a single 440 line file, and 120 of those lines are comments. This is possible because the gems handle most of the heavy lifting for me. It feels very productive to focus on my application logic instead of writing boilerplate code. There were moments of frustration when things went wrong, the black box nature of gems makes them tough to debug, but there seems to be a big enough community using them that they mostly just work.

I'll be using this template heavily myself over the next few weeks, which will hopefully give me ideas on improvements, but let me know if you have bug reports or suggestions. I'm looking forward to hearing people's thoughts on this approach to a turn-key social site too. 

Ruby on Rails first impressions

Photo by Afternoon Sunlight

Ruby on Rails has often attracted me, but I've never had a strong enough reason to dive in. This afternoon, as I unhappily contemplated writing yet another web app with Twitter authentication in PHP, I finally caved. I did think about trying Django instead, since I'm using Python heavily in the back end, but the examples I found felt a bit too bondage-and-discipline for a UI that I want to rapidly iterate on. So, five years after the rest of the world, here's what I discovered in my explorations.

Fantastic Documentation. I read through the first few chapters of the Rails Tutorial book to get a feel for how things worked. It's task-oriented, hands-on, focused on things I care about and generally nicely written. All of the code and examples Just Worked, and from the footnotes this is obviously an actively maintained and tested guide.

Arrgghh Version Changes. With my confidence boosted by working with the tutorial, I set out to integrate this open-source Twitter OAuth module. None of the directions worked, some of them didn't even make sense based on what I'd learned. What is 'script/generate'? Is that like 'rails generate'? After digging I realized it was created for Rails 2 and I was on 3. How big a change could that be? I eventually got things working, but I had to hack so many things manually, from the option loading to the database migration, it was obviously a massive shift in the way things were done. In a way it's really refreshing to find a community so willing to radically clean up things between versions, but it does orphan valuable external projects like this.

Everything's There. It is really nice to have a single, standard way to do most tasks. Since git is the recommended source tool, deploying to Heroku becomes very easy to explain. When things work, they just work. There seems to be One True Way for everything from laying out project files to which source code control system and database to use, and that feels very Apple-esque. The un-hackerish secret I learned from Mr Jobs was that denying people choices saves a lot of confusion and explaining. This is definitely very constraining at times, but when it works it gives a great user experience. I know there's nothing here that I couldn't technically be doing with PHP, but getting it all out of the box and easily google-able (because everyone's using a similar setup) makes a big difference.

I'm pretty happy with my progress today, I have a basic Twitter authentication cycle working, and I'm definitely discovering parts of Ruby that make me smile. I don't know if RoR will become my framework of choice going forward, but it definitely seems a strong fit for a lot of my tasks.

Have you seen this bike?


Some of my friends have had their bike stolen, by thieves who broke through two locked doors and cut through a chain to get to it. The bike is pretty unique, an A2B electric 'Metro' model which I very rarely see around the city, so if you're near San Francisco and spot one of these in unusual circumstances or for sale, please get in touch. There's some distinctive orange tape on several parts, and they have a serial number to check if it's the same one.

They live near the top of a steep hill near me, in Haight, and use the bike to get to the Caltrain station for a commute to the valley, and it will cost several thousand dollars to replace, so it's a real problem for them. The thieves were pretty brazen, breaking into an occupied apartment building during the night, and didn't take the keys or charger so they'll be lugging it around.

The police are involved, and they'll be scouring for this online too, but since this is such a rotten thing to happen, any help from my readers is much appreciated.

Five short links

Photo by Chris in Plymouth

Visualizing Large Facebook Friendship Networks – There’s lots of academic work emerging using social network information. What I find really interesting are the techniques people are developing to make sense of the ‘hairball’ that results from a naive approach to plotting the raw networks, since people have so many friend connections.

What the strange persistence of rockets can teach us about innovation – A really fresh way of looking at technological progress, and a reminder that what seems inevitable now is often actually very path-dependent on the past.

Why did economists not spot the crisis? – The compelling answer is “We don’t reward or encourage people to be generalists”. Academic kudos is only available to hedgehogs who know one thing really well, not foxes who bounce around. I think the skills required to be a generalist are undervalued in the technology world too, and that causes very similar problems.

Africa Rules the World – Some commentary on a slick visualization of growth rates around the world. As he says, it’s a bit misleading because a 20% growth rate in a desperately poor country is not that much in absolute terms, but it does show the dynamism of Africa.

Hilary Mason on NPR – “Everything is interesting”. Bit.ly’s chief scientist does a great job explaining the joys and perils of data.

The American Way of Dating

Photo by Brandon Warren

With the (mostly) shared language, it's easy to for people from the UK to think that America is basically like Britain, apart from the funny accents. I had a little of that attitude when I moved here, but rapidly learned how wrong I was. With Valentine's coming up, I was reminded of one of the best examples of the alienness lurking under the surface; dating. As Kira Cochrane amusingly chronicled in The Guardian, the British standard is "go to a party, down some drinks, make eye contact with a person you fancy, proceed to kissing and often much more, wake up the next morning to find that you have magically become one half of a couple". It seems like the goal was to avoid any unambiguous declarations of interest, so that at any point either person can end the process without the other losing face.

This isn't how it usually works in the US, at least in the mainstream. The formality and rituals surrounding courtship feel like something out of a Noh play. The very idea of actually asking a near-stranger for a date, explicitly and with no particular preamble, in the full knowledge that you may be turned down, seems nothing short of revolutionary compared to the system I grew up with.

Kira ended up avoiding the rules when she was over here, but even she acknowledges there's a need they're filling. Maybe it's because American culture is so varied that the system has to be so explicit about intentions, since people growing up with radically different backgrounds will never be able to communicate using the subtle signs that the British rely on. There's also something refreshingly honest about the whole procedure. A friend was telling me about her travels in Ireland, and being romanced by a hopeful local man. She discovered he was married, with kids, so she asked if it was an open relationship? "Don't be disgusting, woman!" was the reply.

Eighteen Short Links

Photo by Laura Thorne

With my book launch, BigDataCamp and Strata, I’ve accumulated a backlog, so here’s five short links, plus 13!

Gluecon – Eric Norlin knows how to put on a great conference on emerging topics, and the world of integrating different web services, APIs and data sources is one that’s close to my heart. I’m looking forward to seeing the tribe that he gathers in Colorado, and if you’re part of it, you should think about taking up this opportunity to demo your application.

Big Data with Ken Krugler – Ken’s off-the-cuff talk on the pre-electronic US Census was one of the highlights of BigDataCamp for me. This covers a lot of the same ground, but in much more depth. O’Reilly folks, you need to pull this guy on board somehow!

Mapfluence Data Catalog – A well-chosen set of geo and demographic data sets from UrbanMapping. It’s all commercial, which I have no objection to at all, but the lack of obvious pricing means you’ll have to invest time in negotiation with them to decide whether it’s for you. An unlabeled graph doesn’t count.

pipe2py – An intriguing open-source project that takes data flows built in Yahoo Pipes, and converts them into pure Python code. There’s also a quick tutorial available describing how to run the results on Google’s AppEngine.

PeopleSearch – A simple but effective hack, using Google’s custom search APIs to find people’s profiles on major services.

$3m Heritage Health Prize – A fantastic idea, using a Netflix-style data competition at Kaggle to research better ways to predict healthcare needs. There’s some questions around how to best preserve anonymity, but this is such an important goal that it’s worth accepting some small risks on the privacy front.

The O’Reilly Stylesheet – I love reading through stylesheets from different publishers. There’s been a few rules in here I’ve struggled to follow, like referring to a company as ‘it’ rather than ‘they’.

GroundCrew – A simple but effective service for organizing volunteers using cell phones.

Walkshed – There’s a lot of promise in visualizing attributes like walkability and accessibility across cities. A lot of these attributes are really hard to understand unless you devote serious time to exploring the neighborhoods, which made it tough to chose a location when I had to move to San Francisco as an outside.

Map of Scientific Collaboration – A beautiful view of the citation networks in research papers, presented geographically. The next step is to make these interactive and explorable.

Chequered Airwaves – How the high-brow Czech language radio stations ceded the battle for minds to the less scrupulous German broadcasters in the run-up to the Second World War. This struck me as relevant when we consider the right approach to ignorant populist diatribes, in the debate I keep having with myself about how sensational to go.

Ruby Geocoder – The most recent version of the original Perl Tiger/Line US geocoder, rewritten in Ruby and able to ingest the latest shapefiles.

Hacking Lottery Scratchcards – There’s a whole world of statistical data hacking out there, revealing information that publishers never believed they could possibly be exposing.

Small Business Innovation Research Grants – There’s a massive world of US government money available to startups. The main drawbacks are the almost overwhelming barriers to getting through the initial paperwork, the pernicious influence of managing to please federal managers instead of real customers, and in this case becoming part of the military-industrial complex.

Where the Ladies At? App – I may not like it, but this is probably the future of location-based services. After all, Facebook basically started as a way to stalk fellow students at Harvard.

How the O’Reilly Animals are Chosen – I still have no idea how I got a bull for my cover, but given my childhood in a farming village I can’t complain.

Strata Interview – I talk about the Data Source Handbook on camera. I wasn’t happy with this one, I should have talked about all the cool maps people are building with OpenHeatMap instead of going off into an abstract ramble.

Europe vs the US on Privacy – There’s a strong tradition in Europe of assigning a higher value than the US to privacy relative to freedom of expression and innovation. There’s going to be an increasing clash over this as more and more data sources merge and reveal increasing amounts of personal-but-public information.