PAGE: PREV 1 2 3 4 5 6 7 8 9 10 Archives NEXT

Early release of expired domains is rare

Monday, January 18, 2010, 1:06 PM
Business by John (Article #261)

The last weekend I started compiling data on how expired domain names are handled. Here are the highlights:

1. While a lot of registrars do hold onto higher quality domains, the percentage of total expired domains that scored well in my analysis (respectable in-bound links, domain uses real English words, etc) that the registrars actually held onto themselves was surprisingly low.

2. More domains made it out of the auction period than I would have expected. In fact, of those expired domains with a PageRank of three and at least one inbound link from a ranked website, about two-thirds coasted through auction without a big.

3. Very, very few domains were released early. Of the approximately 2000 expired domains included in the analysis, only three were released within two days of their expiration.

Now this is all very preliminary. So, don't get married to any of this not-really information. But, I thought it was interesting and worth sharing.

[NOTE: Added on January 19.]

Having tweak my analysis tool quite a bit, I've determined that the quick drop rate is somewhere below five percent.

The one thing I have to say is that the people at various stages of handling the death of a thousand cuts that is an expiring domain are good at what they do. The whole system has that slightly too mysterious quality that any good racket requires. It's hard to determine when a name will drop. Notification of a genuine delete -- the last stage where the name becomes immediately available to be registered anywhere -- held onto until as late as possible.

Based on the domains that are thrown overboard early, the domain registrars know exactly how well any domain performs while in expired status. If a domain drops early, you should take that as evidence that it has nothing -- no link juice, zero visitors, not enough value to even waste the electricity to keep a "this domain may be for sale" pitch up and running on the URL.

What I'm very interested in now is seeing how the better domains (any domain with a PR greater than 3, a single inbound link, or any real traffic besides domainers) move through the pipeline and become honestly deleted.

At some point the damned things do have to drop into the public pool of domains. And even hunting for domains manually, I've found some winners. So, its not so much a cut-throat business as it is a slightly dreadful business amid a tsunami of data.

We'll see.

Mail article to a friend

The PageRank experiment first results

Wednesday, January 6, 2010, 11:16 AM
Website Design by John (Article #260)

The PageRank experiment I had mentioned in December has already yielded some interesting results.

Let me just say I don't understand the first result at all.

There were three sites used. The first site was a revived site that had previously been deleted. It has a few inboud links from a PR3 website. The second site is a continuously registered domain with links from many PR2 and PR3 sites. The third has a single link from a PR5 website.

Google began updating PageRanks right after the New Year. Apparently the first wave of updates pertains to new websites, because those are the ones that seem to be popping up with new PRs all over the place.

The third site, with the single inbound PR5, has a PR0. It turns out, I forgot to move this site over to a new server when I canceled the one it was on. Oops. But easy to understand.

The second site went from PR0 to PR2. Roughly what I expected. I thought it was on the border of PR2 and PR3 and it fell a bit short. Again, very understandable.

The first site is the one that perplexes me. It has several inbound links, but only from a single PR3 website. It went from no PR to PR3. The inbound link site links to five other websites. My understanding of PR indicates this should have resulted in a PR2 website. And this makes me wonder if the inbound link site has some special juice as a trusted website. Hmm.

PageRank followers believe that Google will begin updating more often in 2010. In previous years, Google updated PRs four times a year, on a very rough schedule of every three months. The current prevailing view is that Google is now updating every two months.

So, the experiment continues.

Mail article to a friend

Fixing the FH_DATE_PAST_20XX bug

Monday, January 4, 2010, 2:05 PM
Website Design by John (Article #259)

Like many folks using a standard LAMP stack system, I've gotten the joy of waking up this Monday to support calls caused by the FH_DATE_PAST_20XX bug.

FH_DATE_PAST_20XX is a bug in SpamAssassin, that scores any sent email from January 1, 2010 or later as spam (approx a 3.0 - 3.8 score, usually on a 5.0 scale to be declared spam).

Here's how you fix it (note that almost all of this requires root access).

1. Find the prefs file where the scoring is done.

On my GoDaddy virtual server, this file was /usr/share/spamassassin. The file is 50_scores.cf.

On most other systems it is going to be user_pref.cf. Or, if you're lucky, you can use the local.cf file in /etc/mail/. However, a lot of systems use different paths. Be warned.

2. Make a copy of the CF file.

I made a file named 50_scores-bak.cf.

3. Using the contents of the CF file, create a new one.

Of course, my new one was titled 50_scores.cf.

4. Do a search for 20xx and change the score.

This section will vary a lot depending on your system. On my system, it appeared as:

score FH_DATE_PAST_20XX 0.075 0.084 0.054 0.088 # n=2

On some systems, it may be formatted simply as:

score FH_DATE_PAST_20XX 0.0

As a rule of thumb, try to conform to what is going on already. If it's a more complicated series of scores, conform to that. Generally, for those using the local.cf file, "score FH_DATE_PAST_20XX 0.0" will work fine. If you have to guess, that would be the first format I'd try.

5. Restart the SpamAssassin server.

If the SpamAssassin service comes back live, you're done. This also varies greatly depending on the machine you're using.

It's also not a bad idea to restart all your other mail services. Once that's done, you should be able to send with scoring 60% of the way to spam on every send.

- - -

If there are any folks out there who need help with this, I do offer my services for profit at $40/hr. I know this can be a particularly nasty thing to fix on GoDaddy systems.

Mail article to a friend

Dear Fedex: enough notices

Thursday, December 31, 2009, 1:15 AM
Thoughts by John (Article #258)

Just a small thought: why does FedEx send a freakin email notice to registered users for every tiny outage they have? I understand the desire of any major company to project that they're on top of things. But, there is a boundary between saying, "Hey, we're here and we know when we screw up" and being overly demonstrative in that effort.

It's obnoxious, FedEx. Quit it.

Mail article to a friend

An experiment in PageRank

Tuesday, December 29, 2009, 4:25 PM
Website Design by John (Article #257)

I'm tired of the great PageRank mystery. Too much voodoo, not enough what to do.

So, I've spread out several domain name purchases with the intent of figuring out what we actually gain by certain inbound links from clearly ranked, legitimate websites.

The first sample is an expired domain, registered, with inbound links from a single legitimate PageRank 3, providing links on all pages. Based on general knowledge from the internet, I'm anticipating this will result in a PR of 2 for the revived website.

The second sample is a continuously registered domain with a previous PR of 0. It is being propped up by a ton of links from PR 2 and PR 3 websites. I also anticipate this resulting in a PR 3.

The third sample is purely propped by a single paid link from a legit PR 5 website that pushes out many links. I anticipate a PR 2 or PR 3 for this site.

I wish to keep the URLs anonymous in the hope of preventing any contamination of the experiment through further links. It's a pretty straight-forward idea: plug some stuff in and see what you get. Try to limit what you plug in so you can isolate what caused the uptick in PageRank.

I am hoping to have results around the beginning of March.

Mail article to a friend

Rescue photos in Photoshop with two simple tips

Saturday, December 26, 2009, 10:46 PM
Graphic Design by John (Article #256)

If you've ever been out taking photographs, it's a simple, but dread problem: your images lack pop. Too much shadow, not enough color. Everything gets lost.

The example on the right is a very good example of this problem. The reason the before image is so lost is because it is a city shot (specifically Two PNC Place in Pittsburgh, PA). The shadows of the surrounding building lose a lot of the light for us, and leave the picture feeling lost.

Now, here are two quick tips for fixing this problem.

#1. Make friends with the Shadows adjustment.

You'll find the shadows adjustment under Image >> Adjustments >> Shadows / Highlight ...

Check the 'Show More Options Box'.

Set the tonal range under shadows to 75%. Tinker with the radius somewhere between 25-100 px depending on the image you're fixing.

Tread carefully with the highlights settings, but play with them. The highlights settings have a tendency to blow very shadowy pictures out and make them appear unreal.

In many cases, that's all you will need to fix a photo. But, if not . . .

#2. Layers

Make three copies of the layer you just applied the shadows filter to.

These layers should be, a 'Color Dodge' layer, a 'Linear Burn Layer' and an 'Overlay Layer'. Adjust the opacity of each layer to control the effect as needed.

These two techniques give you a super fast method for cleaning up shadowy and colorless photos. They also provide a lot of knobs for fine-tuning.

If you look at the photo I've included, you'll see that the finished product (right) has a lot more life and color. Especially toward the base of the building. Before, almost every color down there was lost. Now it looks like a beautiful park. Certainly it looks a lot closer to what the architects were hoping when they designed the building.

Mail article to a friend

No USPS outages this year

Wednesday, December 23, 2009, 12:16 PM
Thoughts by John (Article #255)

I just wanted to give credit where credit is due. Last year, the US Postal service API had serious problems. So far, it's been smooth sailing.

The biggest sales website I run actually ends up with a long season, due to Groundhog Day (I live in Punxsutawney, PA -- whattaya want?!). So, for my part, I don't get to breathe a sigh of relief until about two days after February 2.

On the downside, USPS total mail volume looks to be very down. The postmaster was in the souvenir shop the other day when I went in, and he was giddy just to have someone shipping a pile of packages. Some serious Thunderdome throw down is going to be occurring this next decade among the USPS, UPS and FedEx. Thankfully, DHL and Airborne are already down for the count. Not so sure about losing the USPS.

Anyhoo . . . let's turn that frown upside down. That was way too negative.

Congrats to the USPS techies. You did your job well this year.

Mail article to a friend

Mine some knowledge from your data with Eureqa

Monday, December 14, 2009, 9:04 PM
Thoughts by John (Article #254)

Eureqa is the type of clever tool that would not have been possible, on a mass scale, a decade ago. Built by the good people at Cornell's Computation Synthesis Laboratory, Eureqa is a functional desktop solution for mining data set to discover equations.

What does that have to do with web design?

Well, if you're in the CSS crowd, nothing. The CSS crowd can stop reading at his point and resuming wasting their time reading articles about CSS on Digg.

If you're in the MySQL crowd, Eureqa can be a serious turn-on. The truth is, we as a society are outputting more data every day than all of humanity did a few decades ago. It's a lot of data, and in many cases, not a lot of answers.

It will be a while before processing power approaches the point where Eureqa can be integrated into a typical web server. It's not an FFMPEG-type game changer just yet.

But, it is a huge first step toward taking some of that data and putting it to use.

The next next big revolution (two nexts, because it's a little ways away) is on demand crunching of data. We're seeing services like Amazon cloud computing bring enterprise-class numbers crunching down to reasonable costs for good projects that demand it. It's not hard to picture in five to ten years a point where we can expect to dump a numbers crunching task worthy of Eureqa into a web server and see results within a few second.

In the near term, I am hoping to take a few of my favorite numbers sets I've always wanted to crunch and build them into a website. Look for more soon.

Mail article to a friend

MySQL error 17

Friday, December 11, 2009, 8:50 AM
Web Design by John (Article #253)

I had a MySQL database give grief while dumping out an error code 17 yesterday. What is error 17? Error 17 occurs when MySQL is trying to write a temporary file that already exists.

The first cure to attempt is simple: stop and restart the MySQL service.

If that doesn't work, or you're not comfortable enough with Linux to be able to stop and restart a service, you may also try rebooting. If you're on shared hosting, contact your administrator or hosting provider. It's an annoying error that can shutdown the entire dynamic portion of a website. But, the fix is simple.

Just one of those dumb little pieces of knowledge I would never think to share until I see it happen. I throw it out there for the benefit of anyone who runs into it.

Mail article to a friend

Blaming someone else -- twice

Friday, December 11, 2009, 8:41 AM
Business by John (Article #252)

Here's something no one will ever tell you about business: sometimes you're going to have to blame someone else. In American culture it's generally unacceptable to push blame onto others. We're hesitant to even consider blaming someone else, because frankly the tendency to blame others is one of the diagnostic criteria for being a psychopath. So, we're told that functional adults take responsibility.

The problem is that sometimes it isn't your fault. Worse, sometimes someone else really, really does need to eat shit for their mistakes. In a client relationship this is difficult to navigate. You have to be with a client for a long time before trust barriers come down enough that the assignment of blame can be done without suspicion that you're simply shirking.

I mention this because with a recent client they have had some trouble receiving emails from one of their clients. In both instances of trouble, I've had the ugly duty of blaming their client's mail system.

The first instance was because their client's mail server was misconfigured and identifying emails as having originated with an unregistered domain name. My mail server rejected the emails for looking spammy.

Then they send me an email saying they can't open attachments from their client. Well, it turns out that their client's system is giving all the attachments malformed MIME types (sending them as application/octet-stream instead of the appropriate format).

I cringed at the second chance to blame the other guy. It's not something you want to do. It's not good form in a business-to-business relation to find a third party and blame them -- let alone twice.

The new mail server configuration had its bugs. Don't get me wrong. We had to migrate a lot of stuff from their old system, a system that was somewhat hacked together almost ten years ago and was begging to be update. So, there have been things in the process that were my responsibility to fix. Not that points are accumulated for taking responsibility (they aren't, trust me), but you still like to know you've been honest about your own failures before you run off blaming others.

So, I sucked it up and sent out the second email explaining what was wrong and who was at fault. Strangely, this was a case where a workaround was easier to implement on my end than it was to demand that the third party make a full fix on their end. While blame might need to be assigned, the solution was simpler on this end.

I enjoy being direct and honest. But, there are times where it can be cringe-inducing. That said, at the end of the day if everyone wants to know what happened, blame has to be assigned.

Mail article to a friend

PAGE: PREV 1 2 3 4 5 6 7 8 9 10 Archives NEXT