Archive for the ‘technology’ Category

Snapvine, stockbrokers, and voice posting

June 6, 2008

One of the many real-life intrusions lately has been the recently-announced planned acquisition of Snapvine, which I wrote about on the WhitePages Developer Blog on Wednesday morning. We’ve been working on this for quite a while, and I’m very excited about the energy I’ve seen from both teams as we start to meet each other and think about what we’ll build together.

The entire adventure reminds me of the first startup I thought about creating, which I discussed with Victor in 1999 when we were both working on NetDocs. Streaming audio was just starting to be embedded in websites, but there was no such thing as combining audio with personal publishing (and not much personal publishing, unless you count Diaryland – and I did).

My father has been an investment advisor since I can remember, and when I was a pre-teen, he ran a large office of Prudential Bache in Penn Station. One of my strongest memories was the broadcast conference calls he would lead or listen to each day – with clients, brokers, etc. – sharing what was going on in the market, learnings, etc.

So I got excited about building a set of online tools for this kind of person – someone who wanted to publish pure audio to a set of interested parties. I thought about simple web pages and simple audio controls, designed some very basic pieces, etc.

Then Shoutcast came along, and for some reason I decided that was the end of it – someone had beaten me to it. I can’t explain why I thought they were the same thing – maybe it was just that I saw Shoutcast as a demonstration that I was already behind the curve. Anyway, Victor and I worked on our doomed corporate project and played a lot of Run & Gun, and nothing came of it.

Fast-forward almost a decade later, and now I’m working with a company that’s come as close to building that toolset as anyone else out there, and I’m going to get to work on combining those kinds of tools with the most comprehensive people search website on the planet. It’s an exciting road ahead.

Oh, and I still listen to Shoutcast on my Chumby. (And I do owe Victor a call.)


Crowdsourcing bicycle commuting routes

May 31, 2008

I met someone last night (I think it’s badhill but didn’t get a last name) who described himself as a transit hacker, and we talked for a while about bike commuting (I bike to work almost-daily).

We talked about one of the problems for new bike commuters, which I’ve seen over and over again – getting started is hard because you don’t know how to get there. Especially if you haven’t bike commuted before and so are going to be nervous, you want a route that’s safe, has no unexpected obstacles or hills, and is easy to remember.

Automated online driving directions don’t cut it, even if you can skip highways, because while you might be ok taking a chance on online directions when you’re driving, biking is a bigger commitment and mistakes loom larger (especially for new cyclists).

In Seattle, you can ask for help on the Cascade message boards, but you first have to know they exist. Of course you can talk to friends, but that’s still not perfect.

The City of Seattle distributes a beautiful Seattle bicycling guide map, but it’s job is to tell you what you can do, not what you should do. They take input from commuters, but it generally comes from public forums, which are really just an excuse for people to yell at each other.

This is a perfect opportunity for crowdsourcing, though: thousands of people bike commute every day in Puget Sound, and wherever you’re going, probably dozens of others are doing the same. What if you could just follow their routes, even if you didn’t know them?

The more fun way to do this is to give thousands of cyclists tiny GPS devices that unobtrusively tracked their routes and then reported back to a server, like Dash‘s outbound features for cyclists. This would work, could update as construction impacts routing, etc. It’s also very cool. The more practical way to do this (which Brandon suggested) is to have an online map tool where you encourage cyclists to post their commute routes.

Then you overlay all of the paths heat map-style so that people planning their own route can find the most-often-trekked streets (and directions) and routes. Ideally you could follow individual routes as well, so that you could see why people pick one route over the other based on their final destination.

There’s obviously some way to save out the coordinates that somebody uses to create a path on an online map – Gmap Pedometer creates permalinks, for example – so it certainly seems possible to build this sort of thing. And as being a transit hacker sounds fun, maybe I’ll do it sometime. Or maybe you will. I’m already behind on other projects…

Stop Twitter Spammers with this Greasemonkey script

April 28, 2008

In February I wrote about people who follow you on Twitter for no good reason. Since then I’ve probably received a few dozen more requests from entirely random people/bots/MakeMoneyFasts, and while it doesn’t really mean anything, I find it irksome. Others do too – searching tweetscan for block always produces lots of annoyed people.

Blocking someone, though, is a huge pain. You get the mail message which links to the person, then you have to go to Twitter, go to your follower list, page through it to find that person, and then block. Satisfying, but tiresome.

So today I whipped up a Greasemonkey script to help. (You can stop reading and just download it now if you want.) If you’re a Gmail or Google Apps Mail user, and you receive a mail announcing yet another spammer, you can block them in just two clicks: the first goes to a block page, then you confirm. (There’s no way I’ve found to do it in one click yet.)

Here’s what your mail will look like – note the Block!:

Gmail image with Twitter Block link

Easy to do and satisfying.

You can find the script at this link at Let me know if you have any problems with it.

Some random development notes:

  • This took me about 2.5 hrs, 80% of which was spent learning Greasemonkey, the Gmail Greasemonkey API, and how to do regex replacements in Javascript. I expect doing this for just the Twitter page would take ~30min (so you click through, see that you really don’t know this person, and then block); other mail clients would be fast too, so let me know if you want one.
  • The Gmail Greasemonkey API is a nice touch, esp. because I don’t think you can find the body of a message with XPath otherwise (maybe XPath craps out at a certain # of levels, and the body of the message is 20+ levels deep). However, once you use the API, all document.* functions (including XPath ones) seem to be unavailable: maybe unsafeWindow blocks them, or maybe I just had bad luck.
  • The Dive Into Greasemonkey tutorial might be out of date, but it was still insanely valuable (even if it’s obsessed with timers for some reason).


Adventures in Microsoft PR

April 4, 2008

or, even when he’s (almost) gone, he’s not gone…

From Yahoo News, April 4, 2008:

Microsoft Corp. co-founder Bill Gates said on Friday he expected the new version of Windows operating software, code-named Windows 7, to be released “sometime in the next year or so.”…

A company spokeswoman said Gates’ comments are in line with a development cycle that usually releases a test version of the software before its official introduction.

Give her credit for a nice try, but I do not think that word means what you think it means. Perhaps “or so” would have been a better target?

Google, by the numbers

March 19, 2008

After accidentally searching for 50 on Google and getting a link to 50 Cent, I remembered the Google Suggest single-letter test when it launched, which showed the top result Google Suggest would give when you pressed just one letter. I imagine this has changed over time, but I’m doing a different test – what are the first things to come up when you search for numbers, 1-50, and what does that tell us?In each case, I’ve included links to a few of the top links (removing similar links,  which Google sometimes gets right and sometimes doesn’t), and when the source isn’t obvious from the title (like VH-1 going to and is interesting, I’ve added source information.  Also note that Google inserts a bunch of viral videos with numbers in them from YouTube etc. as part of universal search – I’ve just skipped them.

Google, by the Numbers

1 January 1 (Wikipedia [Wp for short]), MPEG-1 (Wp), CSS Level 1 (, VH-1, Deep Space 1 (a NASA project)
2 XML Schema Part 2 & CSS Level 2 (W3), MPEG-2 (Wp), May 2 (Wp) , a Ruby on Rails Tutorial (Part 2)
3 3 (Wp), January 3 (Wp), 3Com, CSS Level 3 Draft (W3), Halo 3, NYC Subway Line 3
4 4 (Wp), MPEG-4 (Wp), NBC4 in Washington DC, NYC Subway 4, 4-H, Human Chromosome 4 (note: when I did this search earlier from my iPhone, I saw 4Chan)
5 5 (Wp), February 5 (Wp), NYC Subway 5, Babylon 5, US Internal Revenue Code 5, NBC5 Chicago, Perl5
6 May 6 (Wp), February 6 (Wp), Google Holiday Doodle (the 6 is in the URL), Motel 6, Big6 (a literacy initiative), Java SE 6, Perl 6 (three links)
7 7 (Wp), January 7 (Wp), 7-Zip, WHDH 7 Boston, ABC 7 SoCal, KIRO 7 Seattle, QuickTime 7
8 8 (Wp), January 8 (Wp), Human Chromosome 8, Channel 8 (MSDN), Super 8, 8 ½ (IMDB), 8 Mile (IMDB)
9 The 9 (Yahoo!), January 9 (Wp), Channel 9 (MSDN), May 9 (Wp), Human Chromosome 9, Form I-9
10 Channel 10 (MSDN), Powers of 10 (the website), NBC 10 Philadelphia, March 10 (Wp), ICD-10 (disease classification, Wp)
20 20 (Wp), September 20 (Wp), 20/20 on ABC, Human Chromosome 20, Top 20 2007 Security Risks, 20Q (2004 Toy of the Year), US Title 20, 20 Years of Perl
30 30 (Wp), 30 Rock, September 30, (then results for 300),  a php statistics app – no idea why it shows up here, WVIT 30 Hartford
40 40 (number) (Wp), 40 Principales (Spanish Top 40), 40 (year) (Wp), The 40 Year Old Virgin (IMDB), American Top 40, WD-40, 40 Ajax accessibility tutorials,
50 50 Cent (official site), 50 Cent (Wp), 50 (number) (Wp), 50 Cent (AOL Music), Lycos top 50, Z39.50 protocol standards

Some notes:

  • Wikipedia, unsurprisingly, dominates the results – numbers, dates, years. The summaries varied widely – sometimes the first sentence of the article, sometimes a random sentence below
  • Technology, especially web standards and Perl, are all over the place. Maybe this is a sign of the long tail of technology online – if there’s nothing else more relevant, technology rushes to fill the space.
  • For every commercial brand (WD-40), there’s a non-brand (Z39.50).