Bracing against the wind  
www.documentroot.com  

Saturday, March 06, 2010

How to Fix Blogger's FTP Bug

Google is canceling support for FTP in blogger. Fortunately, there's an easy fix and you get to keep your domain, and keep everything working the way it was.

1. Switch to hosting your blog on blahblah.blogspot.com ... but don't finish the migration (read step 2).

2. DONT LET GOOGLE FILL YOUR OLD BLOG WITH REDIRECTS. This will ruin your site's ranking and be, in general, a bad user experience.

3. Use a script to synchronize the blogspot domain pages with your real custom domain blog. Here's the script I'm using, it works well and uses ETags so is remarkably efficient. I'm running 12 blogs from a cron job now, 5 of my own.

4. Turn OFF any indexing services for the new blogspot domain (Allow search engines to find this site? NO). You don't want two sets of content out there. Many search engines, not just Google, will punish your site's ranking for having multiple versions.

5. (Rant: I don't believe, for a minute, that the engineers at Google couldn't figure out how to run FTP affordably. I do over a terabyte of FTP crap every month for free at memebot.com ... and I never even look at it.... paid for twice over with cheap ads. They're either liars or they are incompetent.)

6. ADD this script to the HEAD section of your template... so people won't use your blogspot domain:
<SCRIPT language="JavaScript">
<!--
var str=location.href;
if (str.indexOf('blogspot.com') > 0) {
var newStr = str.replace('http://YOURBLOGSPOTNAME.blogspot.com','').replace('http://www.blogger.com/blog-preview.g','');
window.location='http://www.YOURDOMAINHERE'+newStr;
}
//-->
</SCRIPT>


7. I made a form so people could sign up to have their blog sync'ed if they don't know what a cron job is. If you use rsync, I don't need your password. (Why didn't Google use rsync?). I have a bunch of dedicated servers for other reasons, so it's no problem for me for now (not too many people so far):

New Blogspot Site:
FTP Host (aka custom domain):
FTP Path (for example /blog/):
FTP Username:
FTP Pass (optional):
Use Rsync*:

* NOTE: To use rsync, you'll need click here and put this public key in your authority file (~/.ssh/authorized_keys). This way your password doesn't have to be entered anywhere.



Labels: ,


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Wednesday, March 03, 2010

Nofollow, Viagra, Google & I told You So

When someone links to a website, Google and other search engines have, in the past, considered that a "vote" of confidence for the site linked to. The higher reputation a site has, the more it's vote counts... (sort of, there's lots of other factors).

The idea behind "nofollow" is that you can link to a site, but also mark it as "I'm not voting for it". That way someone who has, for example, a list of comments on their blog, won't simply "vote" for everyone who posts a comment. Presumably that would prevent spammers from posting a million comments in the hopes of getting their discount Viagra ad's search ranking up.

It's a solid concept, but in practice it has backfired because it was coupled with a "punishment" system for linking to spam sites.

This punishment system has led sites like Wikipedia, which are, for the most part, extremely reputable, to put "nofollow" on every single link in an attempt to prevent spam. These links are, usually, extremely well-vetted votes of confidence for the site in question. Wikipedia, despite its problems, remains one of the foremost authorities on "link quality".

Other quality sites that have adopted "nofollow" on every link include Digg and Twitter.

You see where this is going. Rather than rely on top authorities with quality information, like Wikipedia, Digg, etc. Google now has to rely only on links originating from people who don't know or care about their link voting/ranking, etc. In other words, generally less knowledgeable or lower quality votes are the only votes used for ranking.

This ultimately harms Google's page ranking system. What Google needs to learn is that "open" is not the right way to go for search rankings. Heavy use of personalized and regional results is the only thing they've done to halt this trend. Here's the next step: allow someone to mark another Google user as "trusted" for search results. In other words, I should be able to mark friends of mine as "trusted" (on a scale of 1 to 5 maybe), for personalized search results.

That way personalization will dominate the results, be highly relevant, and impossible to game.

Google, the last time I emailed told you to track clicks you listened...and it helped. But you never thanked me! Get back to me when this multilevel personalized trust system is done.

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Tuesday, May 08, 2007

Firefox, Ajax, Google and IE 7.0

Internet Explorer 7.0 running Gmail is painfully slow, and slowly consumes all available RAM if left running for a couple days. I've verified this on 2 machines. Has anyone else seen this? In my tests, I've found that IE7 has major problems with AJAX apps that are "always on", and are constantly updating the screen by replacing the src attribute of blocks.

Did Microsoft deliberately cripple AJAX applications in IE 7? They have a huge financial incentive to do so. In order to continue using Gmail under Windows Vista, I now have to run Firefox. Firefox on Vista works great, and installing it greatly reduced my browser latency.

It's interesting to see how the market prices of the two are strongly negatively correlated. I'm clearly not the only one who's noticed the relationship - Google is being predicted by NASDAQ investors as the loser in this browser war.

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Sunday, February 25, 2007

Locate a Page on Google

LocatePage is tool that allows you to search google and yahoo's search results. Hmm... wait a second, isn't that the same as searching them directly? No. You can search for say "books" and then search *within the first 1000 results* for a subquery.

One of the primary uses, clearly intended by the author, is for finding out the popularity of a given site for a given search term. I think Submit-It used to do this, but they no longer exist, apparently. (Microsoft ran them into the ground, yet another perfectly good internet company ruined by an acquisition.)

For example this site (documentroot.com) is listed in the first couple pages of both Yahoo and Google for the search term "homemade deodorant". A dubious distinction at best.

Labels: , ,


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Thursday, January 18, 2007

Google's Giving Away $10

If you're going to buy something online anyway, buy it from one of these stores. You'll save $10 if you're a first time user of Google Checkout. It's scary to think of how much spare cash they have lying around in order to afford to give away money that way.

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Wednesday, November 15, 2006

Google is too dependent on DMOZ

Google uses it's venerable PageRank algorithm to rank websites included in its search engine. However, when it comes to manually reviewing websites, Google defers to DMOZ, a volunteer-run community that publishes a directory of hierarchically categorized sites.

Without DMOZ, Google would rely largely on blog entries for site reviews. Unfortunately, it is ridiculously easy to create a fake weblog which fools Google's system. In fact, tricking Google (also known as PageRank optimization) is now considered an "industry".

In other words, without DMOZ and the thousands of volunteers that work to insure its integrity, the quality of Google's search engine would suffer tremendously.

I'm writing this because DMOZ is, and has been, broken for a few weeks now. It is a project notiriously short on cash, plagued with internal feuds and corruption.

Interesting to see how a multibillion dollar company like Google could suffer from the downfall of a volunteer-run website like DMOZ.

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Thursday, November 02, 2006

Google Checkout Bug

Only yesterday, Google admitted to a bug which placed checkout badges next to all of their advertiser's ads. Today, Google's checkout.google.com (GCO) website was briefly redirecting to a page that looks like this. Which looks like a hack, but probably was just an internal error.


My personal experience with GCO, as a merchant, has been that the system is a bit confusing for customers. It fails to adequately prompt for shipping addresses; is extremely slow to approve orders; is occasionally slow or broken; and the API is more complex than any other I've worked with.


The upside is that the GCO support group has been extremely responsive and helpful on every occasion. That one fact is what sets them apart from any other merchant processing service I've used. And of course, there's the free processing. Which is the best (only) reason to use them at this point.

Labels: ,


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Sunday, October 08, 2006

Google Costumes

The new site I'm working on is Moon Costumes. It's a halloween costumes website with a pretty large selection.
new linux logo
New Linux Logo?

The hard part was converting the CD's full of images i got from the manufacturer.

I used ImageMagick to convert broken images from jpeg to png and back again, and it's apparently lossless and fixes errors in the jpeg that prevent them from being seen in IE. Also used jpegtran to make the images smaller.

It uses Google Checkout as the payment system and SMX as the UI-building language, and PERL as the backend/business-logic language. Postgres for the DB.

So far Google Checkout has proven to be exceedingly complicated and very short on basic features common to most payment systems. Here's some PERL code that was very hard to come by and works great.

Labels: ,


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Thursday, June 22, 2006

Is Google becoming evil?

It was my prediction that the moment Google went public, it had started down the road to the dark side. It's just the nature of public companies to gradually absolve themselves of all personal responsibility for their actions. Now you can use this handy tool to graph Google's descent into the abyss.


There are many critics of Google. The site, Google Watch is often quoted. But it's clear to me that the arguments and documents on that website are anything buy clear. The site Google Watch Watch which claims to refute the arguments at Google Watch, is similarly flawed in it's reasoning. And yet both are frequently cited by journalists and bloggers in articles about Google. Including myself. Why?


Google's system, without intention, does the following:


  • favors large corporate entities with big advertising budgets. Corporate press releases and public interviews generate highly-rated inbound links.

  • buries sites critical of popular companies, products and people. Popular sites fill up search result pages with existing references. People search for these sites are likely to find these references and quote them - further shoring up a defense against criticism.

  • renders new, innovative, internet sites, however valuable, nearly impossible to find. Google has no review system for incorporating new sites into its directory. New sites must rely on paid ad inclusions, or favors granted by webmsters of other popular sites, to be seen.

  • favors sites, people and topics that are negatively controversial This same argument can be squarely and confidently leveled at any centralized media distribution system. Google Watch is poorly written, but it is sensationalist. Sensationalism sells.


Over time, it's seems like the nature of having a "monolithic" search company that it will become "evil", in the sense that these types of self-reinforcing imbalances will only grow over time.

A simpler way of looking at it is that Google, by centralizing traffic, makes the Internet into TV. And TV is a well-known evil.

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Monday, April 24, 2006

Open source projects crowded out by ads

SO, I wanted to print to a PDF file on my Windows XP box. I know that there must be someone who built an GNU PDF driver, but finding it was difficult. Major search engines, like Google, just can't seem to realize that what I really want is something like PDFCreator (3rd page of my search), and not the various paid and/or spyware-installing packages, like CutePDF, DocUDesk, Win2PDF, etc.

This is a common practice on the internet now. Download some open source (in this case, ghostscript), make a trivial installer or front-end for it (even if an open source one is readily available), and then either charge for it or put spyware all over the installer. Since the installer is their source code, they aren't violating the GPL.

Why isn't the open source version more prominently listed at Google? Because commercial vendors can afford to flood the internet with advertising. Using free software to provide paid-for services, is one thing. But repackaging free software with spyware is something else.

Hopefully some people needing PDF software will find this article first, before they install.

Labels: , ,


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Wednesday, August 11, 2004

IP 132 Mystery

Yesterday we found a large oval pill on our bed. On one side, it was labeled "IP 132", on the other side, it was labeled "600". Neither of us knew where the pill came from.

Doing a Google search for 600 and "IP 132", resulted in a page Help don't know what it is.

"My sister-in-law has a had a series drug dependencies and the other day she had in her pocketbook three pills they were ovalish one side was labeled 600 and the IP 132. She claimed she got them from the dentist after her root canal i dont know what to believe"

This served only to uselessly increase my paranoia.

To solve it, I visited RxList. Their advanced search allows me to search for a pill by the "Imprint Code". Typing in "*132", gives me results that list "Ibuprofen 600mg - Urgent Care Ctr".

It just goes to show you. Google results, while great for finding out what's "popular" on the web, aren't the best when searching for specific, accurate information about a particular topic. For this sort of information, you're better off going to a specialized database.

Perhaps the general case is true as well. Is the "group mind" only good at solving problems within an organized and intentional structure?

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Monday, May 12, 2003

Bloggers Boycott Google

In reaction to this move by Google, webloggers I know are boycotting the site. My new search engine of choice, so far, is AllTheWeb, for a similar balance of size and relevancy. Google apparently thinks that my site, and sites like Slashdot are merely "noise". Apparently the new facists running Google have the power to decide what news is important, and what isn't.

Labels:


[View/Post Comments] [Digg] [Del.icio.us] [Stumble]

Home | Email me when this weblog updates: | View Archive

(C) 2002 Erik Aronesty/DocumentRoot.Com. Right to copy, without attribution, is given freely to anyone for any reason.


Listed on BlogShares | Bloghop: the best pretty good | Blogarama | Technorati | Blogwise