Has Microsoft really opened up?


Photo by Mag3737

I've been closely following Microsoft's Open Specification initiative. A lot of people were sceptical, based on MS's long history of using its closed platform to block competitors and the fact that the move was driven by EU legal action. I had hope, mostly because the company has superb third-party developer support in its DNA. Since July I've been actively using the Exchange and Outlook documentation, and it has lived up to its promise. There's been errors and oddities, but no big gaps or censorship that I could detect.

I've only been scratching at the surface though, in comparison the folks on Samba have been battling the Beast of Redmond for secret API information for over a decade. That makes Andrew Bartlett's description of their team's week-long visit to the Microsoft campus all the more astonishing:

We were very surprised by the extraordinary degree of effort that
Microsoft put into this single vendor plug fest. We were given direct
access to the Active Directory product team in Microsoft, plus we had
a team of 6 Microsoft engineers working with us full time for the
whole week

They knew the protocols in great detail, and were extremely
helpful in working through test cases (both in their own test suites
and ours) and suggesting testing and debugging strategies.

Some of the Microsoft engineers have been actively working with the
Samba code base, particularly our test tools, and had prepared for the
plug fest by doing extensive testing of Microsoft tools against
development versions of Samba4

Microsoft's motivations may not be pure, but this is still massive news. They're offering serious help to any third-party developers who need to interact with their server products. This makes a whole set of products suddenly move from engineering fantasy land to a real possibility. If you've ever had an idea that could benefit from working closely Exchange or Active Directory, dig it out from the back of the drawer and re-examine it. You might find it's more feasible than you think.

Why VCs don’t trust you


Photo by Shazz Mack

I'm a trusting person at heart. That's meant I've had to work hard on the second part of 'trust but verify' to make sure I don't get ripped off. It's also made me (possibly too) sensitive to anything that seems like sharp practice. I ripped into Rick Segal when he joked about a trick question he'd throw at entrepreneurs. I'd overreacted, apologized, and noted that VCs have a tough job dealing with hundreds of people a year wanting money and figuring out who's genuine.

It didn't sink in quite how tough until I saw Rick's last post. He received an unsolicited email with a deck attached. The email was an obvious spammed form letter, but what was really mind-blowing was the extra slides at the end of the powerpoint presentation. Clearly intended to be used only for live talks, they were entirely different versions of the business plan for different audiences, with completely different facts about the team, market sizes and objectives. Dealing with a constant barrage of similar folks queuing up to deceive you would be enough to make Gandi turn cynical.

I often feel like a country cousin when I wander around the Valley, not really understanding how the game should be played. There was a Gaping Void I can't find that had a VC and founder; "What's your business model?" – "You are, rich boy". I like money as much as the next guy, but a startup seems a really hard route to riches unless you're strangely obsessed with building a company. If there's one silver lining to the current funding crunch, I'm hoping it will at least cut down the number of hucksters like these sucking up the investment oxygen.

How to connect to Exchange from Linux


Photo by Bellah

The OpenChange project has produced libmapi, a framework to allow Linux developers to connect to Exchange using the same MAPI/RPC protocol that Outlook uses. This potentially gives you access to the full range of data held on an Exchange server, so I've been experimenting with it for Mailana.

It's very much still in development, so you'll likely have to build it yourself from source, along with Samba which it heavily relies on. That Samba dependence also means any code you link libmapi against also has to be released under the full GPL. In practice, you'll probably want to build a simple command-line tool or some other thin interface to hide the MAPI nastiness anyway, so the rest of your system doesn't have to go GNU.

Initially I downloaded the April 08 'phaser' stable release of the code, but subsequently moved to the top-of-tree to get some of the calls that have been added recently. A new stable release is due within a few days, the code is frozen and waiting for the official release of the Samba version it relies on.

The best place to get started is to browse a few of the example utilities included. If you're interested in mail messages, openchangeclient and exchange2mbox are good demonstrations. For appointments and meetings, check out exchange2ical. There's definitely still parts remaining to be implemented, but in my tests it worked flawlessly grabbing the basic information for all messages on an Exchange server. There were some bugs I had to work around, like the recipient's email addresses and display names showing up in the wrong members of a structure, but overall it was very impressive.

I'm going to keep hacking on libmapi, putting together some issue reports and hopefully finding some possible fixes. Julien Kerihuel and the other developers deserve a lot of credit for tackling such a painful API, I'm looking forward to seeing a lot of interesting projects taking advantage of their work.

How OneRiot could be the foundation of social search


Photo by Danny Hammontree

I'm convinced that the next wave of innovation will be about breaking open data silos that are currently off-limits, and applying some of the lessons we've learnt from the web to build new tools. My focus is on email, but another area full of promise is users' browsing history. We all have the equivalent of a local neighborhood of websites that we frequently visit, and our visits demonstrate that we trust them. Having the option to weight those sites more highly when we're searching could give results personalized to our interests. If you trust your friends judgment too, and use visits from your social network in the weighting, then that adds another level of filtering.

The reason nobody's implemented this approach is that the data isn't easily available. To get the full browsing history you need users to install a plugin, and I know from my own experience that's a high barrier. It's also a network problem, you need a critical mass of friends before the social aspect becomes useful. Me.dium came the closest to jumping this gap with their alpha search service, so I was excited to try out their launch of OneRiot.

They're taking browsing data from their existing 2 million user base, and applying the popularity of a site to its ranking in the results. So as news stories and articles start to spread, they'll instantly appear in the rankings. Google's trying to speed up its indexing, but users are going to find interesting pages more quickly than any robot. This topicality makes OneRiot immediately distinctive, and I can see myself using it to search for breaking news and political stories.

Under the hood, they're offering the PulseChecker plugin as a toolbar to help drive the results, but I like the focus on giving the user something they can use immediately without installation. Persuading people to use an add-in is a tough nut to crack, but it will be a lot easier if the target is already a regular user of your service.

OneRiot are tackling a mammoth problem, but they're starting to build the critical mass of browsing data they need, and I'll be eagerly following their progress.

Tacit acquired by Oracle


Photo by Johnny Vulkan

I just heard that the technology behind Tacit's expert location system has been bought by Oracle. I've heard a lot of good things from users of their product, and they were true pioneers in pulling information from large email stores.

This seems like a smart move by Oracle. Their Beehive collaboration suite integrates email, IM and document management, so it's a perfect environment for a service that relies on data mining. They're still struggling to convince the market that they're a contender, and Tacit adds a unique feature to their system. With Microsoft boasting that Sharepoint is the fastest-growing product ever in the history of the company, this is obviously a space they want a piece of.

I must admit I'm disappointed that Tacit's no longer an independent company, I was hoping to see them blazing a trail for email data mining. After this and Contact Networks' acquisition by Thompson, there's not many standalone businesses left in this field.

Stretch yourself with VolunteerMatch


Forget about the fuzzy glow of satisfaction, there's a lot of hard benefits to volunteering. You can learn new skills you'd never pick up otherwise – Thanks to trailwork I'm now pretty handy with a chainsaw. You'll meet a lot of highly-motivated people who like helping others, and are just plain nice folk – I met Liz working in the mountains. And it's a great way to stretch yourself with new responsibilities. Working as a crew boss has taught me a lot about motivating people to work together and get things, especially as they're all volunteers so I can't fall back on actually bossing them around!

If you're sold on the idea, the best place to start looking for opportunities in your area is VolunteerMatch. Enter your zip code, and you'll get dozens of nearby ideas for almost any activity you can imagine. I recently added the Trails Council to the directory, so if you're in the LA area you can get the full details of what we're up to every Saturday.

For outdoors events, REI runs a fantastic website and email newsletter. Go to their store locator, find the one nearest you, and then click on the events link to see what's happening locally. Last Saturday we had 9 first-time volunteers who heard about us through the Santa Monica store's bulletin.

How can you make a big corporation work?


Photo by Frank Maurer

Big corporations are often dismissed as dinosaurs. I'd agree, but argue that we're still in the middle of the Jurassic. There's a lot of benefits to being a massive organization, from economies of scale in manufacturing and marketing, to the depth of expertise you can build up in-house. The most important obstacle holding them back is the difficulty in getting that many people to efficiently organize themselves to achieve the business's goals. When I was at Apple I spent a large part of my time tracking down internal experts, to take advantage of the company's existing assets rather than wasting time and resources reinventing the wheel.

What I needed, and didn't have, was a complete and searchable directory of employees. I had the basic white pages, that listed a phone number, department, photo and immediate superior of any colleague, but nothing that let me track down people based on their skills or interests. That meant making connections was a time-consuming process, reliant on cold-calling and word-of-mouth. Multiply this across an organization, and you've got a massive amount of waste and inefficiency.

That's why I've been so interested in IBM's Bluebook project. They've built a company-wide system that's more like Facebook than the phone book. Each employee has a profile with their basic information, but also a wealth of user-generated content, from their resumes to blogs and things they've tagged. Searching this would let me locate helpful contacts a lot faster than more manual methods. They even encourage people who've been helped by specialists to rate them, so the helpers can get credit at their next review. This seems a crucial component of any system that aims to build horizontal connections in an organization. At the moment lip service is usually paid to the idea of helping other teams, but rewards are often based solely on your own group's achievements. This leads to fiefdoms and vertical silos within companies, as department heads fight to safeguard their resources from other groups. Only recognition of cross-team help can break that inefficient pattern.

Bluebook became the ancestor to Lotus Connections. I haven't seen much coverage of it in the mainstream tech world, but Connections seems to be getting a lot of adoption. I'm convinced that approaches like this are the key to getting more for less in big companies. Microsoft's Sharepoint is headed in the same direction with its approach to social networking using 'My Site' pages for each employee.

Part of winning the adoption battle might be to lose the social networking label, and focus on a more familiar analogy, the employee directory. It's easy to imagine how having an enhanced directory can help productivity, whereas a social directory makes it sound like you're providing a service for employees to waste time chatting on.