Diary of a schwag hag

I managed to accumulate two main things at StackOverflow DevDays Cambridge yesterday:

  1. An almighty hangover, courtesy of Red Gate‘s endless supply of free booze in the evening.
  2. A huge bag of quality schwag garnered from various corners of the conference, including a free copy of the Aardvark’d DVD, some fish toys that Red Gate had brought for some reason and various stickers and stationery etc.

And inbetween hunting and gathering, Carsonified put on a conference. Talking of which:

Joel Spolsky – Opening Keynote

So I only saw half of this, because even though I had left the house at 5:10am, I didn’t get to Cambridge until 9:10am. For the record: TweetDeck and Byline had my iPhone battery at 50% before even stepping off the train. Despite arriving late, I had no worries, because the pieces of Joel’s keynote that I missed seemed just to be what was on the last StackOverflow podcast. What I did catch, however, was typical Joel stuff. He was riffing on JWZ’s “How will this software get my users laid” and managed fairly well to mix threads of low-grade erotic fiction and software usability into a quality rant. Good stuff, although I couldn’t stop picturing how much Joel would look like Stanley Tucci if he had less hair. I don’t know if that affected anyone else’s enjoyment of the talk. Probably not.

Ryan Carson – Carsonified

While Ryan wasn’t a speaker at the conference, we got to see plenty of him, and he entertained with his now trademark “three minutes left”, “two minutes left” and incessant polling. Nah, just kidding, Ryan’s alright.

Christian Heilmann – Yahoo! Developer Tools

This was a pretty good, made all the more enjoyable because hearing a big ginger German guy swearing is good, clean fun. Christian‘s was one of the talks I was looking forward to, having heard the buzz about YQL from attendees of the previous London event. As you’ll read from everyone else’s write-up, Yahoo have wrapped dozens of data sources of their own, and made it possible for others to expose theirs, allowing queries in an SQL-like syntax over HTTP. You can imagine the rest, but here are some specifics from the talk and a brief chat afterwards that might not have been mentioned elsewhere:

  • It’s throttled to 10,000 requests an hour, 100,000 a day.
  • It runs on Hadoop over 50 machines at Yahoo, although Christian’s not on the team and wasn’t sure of the exact architecture (he mentioned pixies though).
  • They intend to offer a version that you can host yourself on Google App Engine.
  • He regretted not mentioning in the talk that you can write transformations for YQL results in JavaScript. This uses Rhino, presumably running in the above-mentioned Hadoop cluster.

The other thrust of his talk was about writing apps/widgets for the Yahoo home page. Yahoo have made their stuff Caja compliant to allow sandboxing of client-side apps, and while only large partners are able to publish their widgets, soon anyone will be able to offer useless crap to put on their Yahoo home page (and reach the 330 million users thereof). Good opportunity when it happens, although if you planned on monetizing your app somehow, it sounds like you’ll suffer the same problems getting noticed as you would on the iPhone app store – there’ll be featured apps and categories, but I think it’ll just end up as a sea of under-appreciated apps drowning each other out. I can’t help but think the Facebook model of apps spreading virally is the way to go here – you install an app because someone has attacked you with some sort of flying monkey, and when you click to fling a monkey back, you install the app as a side-effect. You never knew you wanted the app – you’re hooked in before you knew it existed. That said, Facebook apps are all shite and I don’t use them, but I don’t use Yahoo either so I guess I’m not the target market.

Like most Facebook apps, you’ll need your own server to serve up the content of your app, and while I asked if there were any plans to offer hosting for pure HTML/JavaScript apps (perhaps targeting just YQL data), Christian said they didn’t want people to have to rely on them to keep their stuff alive, which perhaps in the light of GeoCities’ retirement is fair enough.

Frank Stajano

Frank‘s a security researcher at Cambridge, and presented his seven principles for systems security. These principles are basically seven easy exploits in human psychology, upon which all sorts of scams, both online and off, are based. The majority of the talk was just extracts from The Real Hustle which he says is required viewing for his PhD students, but he didn’t really make a big enough leap from there to how this affected computer systems day to day. A few of us chatted to him afterwards at lunch asking about what’s to be done to remedy some of the problems, especially as he mentioned that often we compound the issues by teaching bad behaviour – online banking sites on weird, random URLs for example. Unfortunately, his research seems purely related to uncovering the exploits, he could only offer the guidance that we just take the inverse of his seven principles. So, our vulnerability to ‘deception’ being one of them, we… er… don’t lie to our users. I’m not sure if that’s what he meant. Anyway, I also asked him what he thought of things like OpenID and 3D Secure because I think these are two hugely flawed systems that for many users just teach the behaviour that if you see the right logo, you type what’s asked of you. His reponse to that is that if we’re going to use those systems in our apps, we should be the ones that take responsibility for our users or their data being exploited. Which is funny, because the removal of that burden is pretty much exactly why 3D Secure exists.

All in all, some very interesting content from the theoretical standpoint, and certainly some useful stuff if you are into threat modelling and such, but it would be nice if Frank gave us some pointers to at least someone else’s work on the flip side of preventing these problems – it was an audience largely made up of implementors, after all.

Joel Spolsky – FogBugz

This would have been too much of a sales pitch if we weren’t sort of in the market for a (bug/issue/requirement/case) management system at the moment. From what we saw, FogBugz towers above Mingle in both functionality and especially ease-of-use. We also got a quick demo of the brand-new Kiln. While it looked useful (hosted source control and code-review), the fact that you’d have to host your code on someone else’s servers would be a deal-breaker for every company I’ve ever worked with. Bringing that up with Internet-famous Aardvarkster Benjamin Pollack over lunch, he assured me that the current number one priority was making the Kiln plugin available for the licensed (self-hosted) version of FogBugz. It’ll come to Windows first, Unix afterwards and six months sounded like a reasonable time-frame. To that end, I put my name down for the beta, and as I think I qualify for the free startup licensing that Fog Creek offers, I’ll definitely be spending some time with both tools.

Red Gate – Lunchtime Presentation

I arrived a little late for this, and never caught the name of the product being discussed (although I’m pretty sure Reflector was mentioned at one point – more on that in a bit). The talk was given by a designer at Red Gate whose name I also missed (Update: Brian Harris of Red Gate tells me he was Stephen Chambers, and very good he was too), but it was actually a really revealing presentation. Red Gate iterate early and often, and record regular design reviews. We shared the presenter’s pain as a user on the recording tried desperately to find the code he was interested in, with no help at all from the tool he was testing. This feeling was intensified for me because I’d come in late, and assumed we were actually witnessing some disastrous live conference-call-slash-demo as part of the presentation. However, the presenter’s pain was only remembered, and the talk went on to show how the early embarrassments helped them focus on areas for improvement (although some of the bugs didn’t seem like they should ever have left a programmer’s machine) and track their progress honestly and openly. Recordings of later, successful demos revealed an atmosphere at Red Gate that must have driven at least a few people to speak the recruitment team manning their booth in the dinner hall.

Talking of which, I learned a couple of interesting things. First, I was never really clear on the nature of their Springbroad programme. For some reason I thought it was pretty much just office space and some free coffee, but they’re actually funding the startups they brought in over the Summer, and unlike YCombinator, not taking any equity in return. Now, I suspect the quality of leads generated by the latter is higher, but you can’t argue that it’s a good deal, and while details weren’t clear on whether there was to be a Winter intake to the programme this year, the people I spoke to said they were always open to good ideas and would consider any applications that came through any time. Interesting stuff, although I’ve not followed any of startups involved, so have no idea how successful it’s all been.

More importantly, I managed to speak to Bart Read, head of Red Gate’s .NET developer division. I’d been slightly worried that nothing much had seemed to happen with the awesome Reflector since they’d taken it over, but he pointed out that there was currently a beta available the next version. That’ll bring Visual Studio integration, and cool stuff (for .NET developers at least) like stepping through other people’s source code. It’ll be a paid product (current functionality will always remain free) and sounded like it might land in February. I intend try the beta over the next few weeks, although it sounded like VS 2010 beta 2 might have broken things.

Steven Sanderson – ASP.NET MVC

I think most people that care about ASP.NET MVC know about it now, so I won’t go over the content too much here. I did have a lively conversation at tea-time comparing it to Tapestry, but it’s safe to say that it joins the vanishingly small list of web frameworks that I would describe as ‘not bad’.

The thing I’d like to focus on is actually Steve himself. As others will have mentioned, he was by far the quickest of the presenters on the day, and got through a fair amount of real code interspersed with enough dry humour to keep it engaging. What I think was brilliant, though, was just the slickness of the presentation. He had complete command of his content and his tools, and sailed smoothly from step to step without any ceremony. I came away really impressed that while we watched two presenters on Macs beachballing all day, Windows 7 and Visual Studio 2010 just seemed to get on with the job. Needless to say, this has not always been my experience with Vista and VS 2008.

Anyway, the talk picked out a few of the nice features of MVC 2.0 that you of course already know about, but it was nice, as I’d only bought his book the week before. In true schwag hag style, I asked Steve to sign it, and while he was a bit confused by the whole experience, he very kindly obliged.

Remi Sharp – jQuery

It must have been tough to know where to pitch this talk, as there was a real variety of levels of experience with jQuery in the room. As it turned out, Remi opted to launch an almost encyclopaedic attack on the jQuery API, and given my 4:45am start, I found myself drifting a little. Right at the death, he started on a practical example, and there was palpable tension in the room as the clock ticked down. It might have been nicer to start with the example and expound upon the features and syntax it demonstrated, but as it was time ran out and Remi had to finish it in the break. I didn’t stay around to see the finished article, although various tweets said it was well worth the wait.

Michael Foord – Python

It was a huge surprise that the Knuth-packing beardo in the front row turned out to be a Windows developer, demoing on IronPython. Michael was excellent, and after demoing some IronPython/WinForms interop for shits and giggles, was working with the now hugely over-analysed spelling corrector by Peter Norvig. I suspect his talk was better than Mark Harrison’s at San Francisco – although I only have a snippet of podcast to go on for the latter – if only because Michael sounded the more opinionated. There’s nothing hugely new to say here about either spelling corrections or Python, but Michael’s day job at Resolver Systems making highly programmable spreadsheets sounded interesting.

Jeff Atwood – StackOverflow

Comedy slides and an ironically short, subdued talk about passion closed out the day. Again, I’d heard most of this on the podcast and honestly was just thinking about the free beer, so can’t do Jeff justice. The most satisfying takeaway from the talk was really just ‘mission accomplished’. StackOverflow probably isn’t perfect, but they executed on their vision and it worked out. While the launch of StackOverflow careers and Stack Exchange have obviously taken up the two teams’ time recently, it wasn’t clear what else remained on the roadmap, or if Jeff and Joel had any plans outside the Q&A arena.

After Party

And then there were beers at Henry’s. Met a very interesting guy called James Nicholson, who develops the mobile clients at Taptu, a mobile search engine. James stayed in Cambridge after uni and it was fascinating to hear just how many startups, and how much VC money, has come to Cambridge and its science park. It sounds like the perfect place to be if you’re starting up in the UK, and that the whole area is very easy on the eye is just a bonus.

I also bumped into my brother in law. Random.

With a dead iPhone battery, I crawled home on the last train to add my schwag to the mountainous hoard I keep in the dungeon below my castle. Sometimes I sit on it and wait for men with swords to come. They never do.

This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

5 Responses to Diary of a schwag hag

  1. bri says:

    The name of the designer from Red Gate who gave the lunchtime talk is Stephen Chambers.

  2. Thom says:

    Cheers Brian, I’ve now added his name. Just out of interest – any clues as to what products we were seeing?

  3. Stephen says:

    Hi Thom,

    I demonstrated usability sessions from 3 different products.

    The first 2 videos demonstrated sessions from an ANTS Memory Profiler 5 early access build being user tested and a follow-up video showing user feedback after lots of changes had been made to improve usability etc…

    The 3rd video was from a usability session with the upcoming .NET Reflector Pro involving picking assemblies.

    The last video was from a usability session we conducted with ANTS Performance Profiler 4.

    As for letting things out with bugs – we know they have bugs but that’s half the point really with the (very) early testing. We have crashes or things go wrong during sessions etc… but if we kept fixing lots of bugs then we vastly reduce the time we could be using to get some much more valuable feedback. Users are also extremely understanding of early access builds – they don’t seem to mind at all and we often joke about how rough a build is. :)


  4. Ed Bowler says:

    so … you’re suggesting that banks should provide a page that you can customise yourself that they will show you to confirm their identity before asking you for any info, with whatever content you put on there … like any text and pictures that you like … … hang on – that’s a saleable idea … ;)

  5. Pingback: I Met Jeff Atwood And All I Got Was This Sticker « Analysisuk’s Blog

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>