Recently there was a test comparing the Mozilla-Google phishing protection to the MicrosoftIE phishing protection. You can see the detailed results.
One of the problems that search engines hit on is that a page is just a blob of text to them. And when they look for keywords in that text they aren't 100% sure if they are looking in the "content" of the page or in links to other pages.
This can cause serious problems!
Over at The Fon Blog (unofficial one, that is) antonde has made the claim that the new FON map and the FON counter don't match up. In his sample, less than 10% of the registered foneros showed active WiFi points on the FON map.
Activity of FON Points
In my own sample (my house) the point shows as not being active, but it very definitely is active. FON has some problems to overcome with this map and I'm sure it's just a matter of time. On the one hand, they only want to show points that they are fairly certian are actually working. The most certain test of a "working" router is some recent activity by a fonero on the wifi side of the router. That test is also likely to give "false negatives" such as in the case of my router. It's unlikely that people will use it (I use the wired ethernet side), which marks it as inactive, which makes people unlikely to use it...it's somewhat self-fulfilling. Last night I used the FON wifi network to see how long it would be to show my point as active. Now (12 hours later) it's still not showing as being "active". In my test, the system is overly pessimisting.
Mentiroso FON Counter
antonde also called for FON to 'please stop that “mentiroso” counter on your home page'. Mentiroso is Spanish for "Liar", basically saying that it over-inflated the number of Foneros. Again, in my experience, I was added to the OLD map and I believe I was counted in that number before I had my access point delivered, much less installed and running. The old map and counter were overly optimistic as they counted people as soon as they registered - not based upon actual running routers.
The "right" way to count Foneros?
My personal practice would be to be somewhat pessimistic and only count the points once they've had an active user. However, they should count a router as active if it has had any activity in the past week or month and then also if it has checked in for it's nightly "heartbeat" ping.
So, I read this post in the digg blog with much fascination.
it shows this:
digg_url = \'URLOFSTORY\';
Which is pretty cool because then it makes it easier for users to digg a story on your site even if they found the story through some other means. But there's a pretty huge problem with the widget...
The workflow for this is something like:
- write story
- wait for someone to digg story
- find URL of the digg story
- add widget to my site
Because then the workflow would be something like
1. write story
And that's clearly a better workflow.
If the answer is "we just can't do that" then I've got a surprise someone else will do a better job and you'll be forgotten.
So, back on July 14th when I started discussing FON I took a snapshot of the FON maps page. I figured hey, this thing is growing pretty quickly, wouldn't it be neat to have snapshots of this data so I know how quickly it's growing. Today is the first installment (I was a week late on the whole monthly snapshot thing, but close enough...).
For the impatient - overall monthly growth across the entire earth is just under 25%. This data isn't 100% accurate (they add you to the map before you physically install the router, for one, and I have a feeling they are slow to remove inactive nodes as well) but it's still pretty interesting.
Freedbacking to FON
And, hey FON guys, it sure would be nice if I could get a breakdown by municipality as well. Now that Spain is at 10,000 FONeros (spanish) it would be nice to know how many of those are in Barcelona or Madrid or...
Google has a long history of making special logos to go along with special days: holidays, birthdays, exploration, important stuff.
I read the Inside AdSense Blog because monetization is FUN! Today I woke up to see that they now have a search box that will give results right inside your own page which is a great step beyond their previous effort (the previous effort allowed you to do some basic theming of the results page...pretty useless). So, I got all excited and hooked it up on UAGDSE.com and if you try it out what you'll find is that the search code and Drupal get mad at each other.
A Little History
Drupal, like many web applications, originally used URL parameters to create its URLs. Things like ?q=node/3 or ?q=admin let the system know that the user is requesting the "node/3" page or the "admin" page. Over the years people figured out fancy ways to make URLs work without the ?q= but Drupal still supports this notation in case your configuration doesn't allow the rewriting magic necessary. It's also handy for when you recompile apache and forget rewrite and your site breaks - then you can still get into ?q=admin/settings and fix stuff.
The Google vs. Drupal conflict
The way this becomes a problem is because Google's search expects to be able to use the "q=" parameter and it creates URLs like http://uagdse.com/uagdse-results?q=term which then confuses Drupal into thinking that I'm asking for a page named term which doesn't exist. The funny thing is that when you do a google search for a path that actually exists and is available (like node/198) then Drupal returns that page quite nicely.
I'm not sure what the fix is. I can imagine one solution being to use a more specific url parameter than "q" like "dq" but I doubt we're going to get the millions of people using Drupal to change. I can also imagine a setting in Drupal for "no really, ignore, I only want clean URLs so please ignore any q= in the URL parameters" but that doesn't feel right. Of course, part of the reason I'm writing about this is a little lazyweb plea for someone with a better idea on how to fix it.