1,000th Flickr Photo
- Posted by
- Breyten² breyten
- When
- 8-5-2008 @ 23:36
- Security
- public
Anyway, on to the next 1,000 photos. Or something
Originally published at Breyten Ernsting. You can comment here or there.
Pi Pie
uws reminded me that it's Pi Day again. So, happy Pi Day!
Originally published at Breyten Ernsting. You can comment here or there.
I've begun using SuperGenPass to improve the quality of my passwords across the (many) different sites where I have accounts. Basically, SuperGenPass uses a master password and the domain name to generate a password that is specific for the domain. This approach, combined with a handy bookmarklet makes sure that the increase in security does not come with a high cost in efficiency. The bookmarklet simply takes the existing password in a password field and uses that password as the master password, so the login procedure for a site is extended by just a simple mouse click.
Originally published at Breyten Ernsting. You can comment here or there.
I like the idea of opening up the search results, like Yahoo is doing. It opens up a ton of new possibilities, like including the complete first paragraph of Wikipedia pages, etc. And of course, this will be an amazing opportunity to make intelligent use of microformats. Will definitely be playing with it once it is available for the masses.
Originally published at Breyten Ernsting. You can comment here or there.
I just created my first pipe using Yahoo! Pipes, which is a great product. Easy enough to use, and dragging the links around to form connection is very smart.
There's only one thing missing though: When can we create and use our own modules? Such a feature would instantly turn it into an even better killer app.
Originally published at Breyten Ernsting. You can comment here or there.
Topper is a kind of an interesting take on a search engine. Their frontpage has a blurb that is filled with all the right marketing words, but that doesn't really describe the service right. The basic idea of Topper is that it will use user input (as relevance statements) to "learn" to re-rank the results. In itself, that is an idea worth investigating. We can see on Topper however, that the problem of this approach is not in the limited amount of user input, which they seem to handle quite well. This is probably based on some boosting method.
The real problem for Topper is how to present such an approach visually. Topper shows relevance scores, which change as you provide more input. However, the result list itself is never changed, which leads to strange results that will confuse most people. It took me a while to get it.
Nevertheless, this will be a good learning experience for the team responsible for developing Topper. I'm curious to see if they can get the interface right in the future.