Skip to: Navigation | Content | Sidebar | Footer


Full Archives

2004 Stats

December 24

Google's 2004 Zeitgeist showed up this week, so in that vein I present the 2004 mezzoblue Logfile Blitzkrieg.

In a few hours I jump on a California-bound plane for my first Christmas without snow. Despite living in a city that only gets one or two mild dustings each winter, if that, I've somehow managed to find it every year. No more. Somehow I think I'll live.

Having cleared up all other loose ends before departing, I've spent a bit of time going through the aggregated stats for this past year. They're painting an interesting picture, though perhaps only so for those of us with a close personal attachment to this site... like, uh, me.

Totals

First off, the aggregate graphs for the year. What you're viewing is generated by Urchin and covers the span from 1 Jan 2004 to 23 Dec 2004 (with a few weeks gap when Urchin went on the blink).

Sessions Graph

Above is the user sessions graph. The only real use of a graph like this is analysing the trends, rather than working with the specific numbers. Although getting accurate server stats has always been a challenge, the highly-trafficked nature of RSS feeds skews them even more, so the overall trend can only be an approximation of what's happening.

What's happening in this case is that traffic is obviously continuing to grow, which is absolutely amazing considering the frequency of new content on this site lately. (Although when the book finally ships to the printer, which will happen soon, I'd hope to see a little change there.) The oscillating peaks (five days), and troughs (two days) show that weekends are always lower traffic days. During the summer there's an obvious lull, which is far more a plateau than a dip, which is kind of weird, but I'll take it.

Bytes Graph

Next up we have the bandwidth utilisation graph, which looks quite a bit different than you might have expected after the previous one. A sharp upward spike is curtailed in early April, which doesn't make a lot of sense. I figured it was due to the lower image hit of the redesign (later replaced in August by the current design) but that launched in late May—the downturn was already underway by then. Perhaps gzip compression kicked in late in April, otherwise I'm not sure quite what happened there. The continued lower bandwidth is most likely a function of the redesign, at any rate.

I'm not going to translate the sessions into numbers, but I will tell you that averaged over the entire year, this site transfers just over 900MB per day. That's rather a lot, but a far cry from the Zen Garden rest assured. (That one's just about up to 200GB per month; thank god for Dreamfire)

Breakdown

Now for some of the specifics. We'll start with the most obvious, the browser stats. Over the past year, these are the top ten user agents visiting this site and their numbers. There are two tables here that paint very different pictures. The former appears to be biased in favour of repeat visitors and those who access it only for new articles, and the latter seems to be biased in favour of those coming in off search engines, browsing the site, and looking for specific information. The real numbers must be somewhere in the middle.

Top 10 browsers by Session
User AgentPercentage
Mozilla18.04%
NetNewsWire16.12%
Internet Explorer13.75%
SharpReader6.86%
FeedDemon4.70%
Safari3.42%
Opera3.20%
NewsGator3.13%
Netscape3.05%
Mozilla Compatible Agent2.68%
Top 10 browsers by Hits
User AgentPercentage
Internet Explorer57.40%
Mozilla21.55%
Safari6.33%
Opera2.77%
NetNewsWire2.01%
Netscape1.60%
SharpReader0.78%
Mozilla Compatible Agent0.64%
Feedreader0.50%
NewsGator0.49%

A far more interesting story occurs after filtering out the RSS readers, and charting the percentages month by month. So here are the top 5 non-RSS user agents over the course of the year. I'm going to chart this based on the Hits statistics from above, since they seem to be a little more realistic. IE includes Windows and Mac versions, Mozilla includes Firefox et. al, and I think the numbers are being influenced more than a little by the growth of RSS readers, especially since these started out the year as the top 5 user agents but ended up scattered over the top 10:

Top 5 Browsers by Month
IEMozillaSafariOperaNetscape
January71.32%13.63%5.71%2.44%1.56%
February76.89%11.02%3.77%1.84%1.40%
March78.35%10.81%3.80%1.73%1.41%
April66.01%16.78%5.85%2.44%1.56%
May48.89%26.20%8.62%3.82%2.65%
June44.58%27.62%8.63%3.64%2.10%
July41.09%29.59%8.38%3.48%1.87%
August38.32%31.55%9.27%3.70%1.55%
September36.11%32.92%8.21%3.57%1.32%
October35.19%32.66%7.83%3.51%1.31%
November34.41%34.30%6.52%3.02%0.92%
December31.71%37.54%7.31%2.92%1.02%
Browsers by Month

I'd expect that sort of trend amongst the readers of this site, but it's great to see it happening, and so quickly. Interestingly, while shuffling through the above I noticed that a not-insignificant chunk of bandwidth was consumed by bots and automated apps this year, well over 2 or 3%.

Top 5 bots by Bytes Transferred
User AgentPercentage
GoogleBot3.43GB
wget1.42GB
msnbot1.17GB
Bobby554.92MB
Jigsaw355.20MB

Miscellaneous

And finally, I leave you with a few assorted lists of this and that. There's probably not much of general interest in these tables, but here they are anyway.

Top 10 Search Terms in 2004
Term
css zen garden (and variants, like 'garden design' etc.)
color schemes (and variants, like 'color+scheme')
mezzoblue
times new roman
dave shea
rss+atom
26 (?)
texture
how to conceive (!!!)
browser stats
Top 5 Search Engines in 2004
Search Engine
Google (and its many regional variants.)
Yahoo
MSN
DevilFinder
AOL
Top 10 Referring Sites in 2004
SiteNumber of Sessions
Bloglines23,450
A List Apart17,900
blo.gs11,150
Stopdesign10,850
SimpleBits10,500
Cameron Moll7,450
MeyerWeb5,600
Simon Willison4,650
CSS Vault2,800
Fairvue Central > Bloggies2,400
Top 5 Entry Pages in 2004
Page
/rss/index.xml
/
/rss/2.0/
/zengarden/alldesigns/
/rss/2.0/dailies/

Alright, now that I've got the stuff that doesn't matter out of the way, on to that which does. A week of sun, here I come.

Permalink › | no comments

The Quality Premium

December 20

This is what happens when the most popular product on the market is actually the best product available: Any would-be usurpers come across as shoddy knock-offs.

Apple's iPod next to Creative Labs' Zen Micro

On the left, the reigning champ supreme, Apple's iPod. On the right, Creative Labs' Zen Micro (no relation).

What we see in the user interface buttons of the Zen Micro isn't much of a user interface. It's more like what goes underneath the interface—the touch-sensitive pads you'd expect to find if you peeled off the metal faceplate, or the messy internals you'd be exposed to if you dropped your cell phone on the floor and shattered the front. If they've thought about those buttons any further than "Back on the left, Forward on the right" I'm having a hard time finding evidence of it.

A few minor iPod usability quirks aside (which mainly revolve around difficulty operating the song/playlist selection through a jacket, without taking it out, which is mainly in turn a combination of exceptionally high expectations and just out-and-out laziness) Apple has the portable music browsing interface perfected.

No one is going to topple the iPod on ease of use. Which leaves features, or price. When the consumer has the assurance of quality, combined with the lure of popularity, paying a premium isn't that hard to stomach.

Permalink › | 44 comments

e-commerce Hall of Shame

December 10

I ordered a product for download from the Adobe Store last week, for the first and last time.

I've made it a policy of never ordering downloadable software when physical media can be had. I've paid for various programs that are only available as a download, but when shrink-wrap is available, shrink-wrap is what I opt for.

Backup has long been the reason for my bias—at least with physical media, there's always a safeguard. I'm not responsible for keeping the software protected from data failure, aside from storing the CD properly. Never did it occur to me ordering downloadable software might have even more consequences than data loss.

We're currently in the 'first pass' phase of editing, which means chapters have been submitted, edited, and dumped into the book template. This is more or less the final review, unless enough needs to be changed to require second (and heaven forbid, third) passes.

Since the content is now visually mocked up, obviously Word isn't going to cut it anymore. We've been passing back and forth PDF files which are best edited and annotated in Adobe Acrobat, a product I've never had reason to own.

So, time being an issue, off to the Adobe Store I went to pay for and download my copy. (Side note: as I'm sure is the case with most outside of the US, prices are magically increased by a margin I've come to refer to as the "non-American tax". Acrobat Standard: $450 CDN in stores [pre-tax], $299 USD online [no tax]. Given the currently weak American dollar, it's a no-brainer.)

Now here's the process I expected, as my incentive for buying the downloadable version was to have it immediately so I could begin my notes:

  1. Add to cart.
  2. Input personal information, credit card, etc.
  3. Checkout.
  4. Download.
  5. Install.
  6. Begin editing.

I guess I have more faith in technology than I should. Murphy's Law and a decade and a half of computing experience has taught me that nothing ever goes that smoothly, but ever the optimist, I still expected to have the product in my hands shortly after paying for it. You know by now, don't you, that this wasn't to be?

I made it as far as step 4, Download. Adobe has a custom download application that allows you to resume broken downloads, a great idea... with a horrible execution. Instead of ensuring I get the file, the download manager itself refused to start. Double click, wait, crash. Try again: double click, wait, crash.

Why not re-download the download manager? I couldn't, the download had already been purged; as far as Adobe was concerned, the first few steps had been fulfilled and I now had everything I needed to complete the next steps. Except, of course, I didn't. After a last-ditch effort involving a reboot and some finger-crossing, it was clear that I could do nothing but call the Adobe support line and have them help me out.

Here's the process that ensued:

  1. Hit up Adobe.com for tech support phone number.
  2. Call support line, get told that hours are 6am to 8pm, and they're closed for the evening.
  3. Wait a night.
  4. Call support line, summarize the problem.
  5. Provide name, phone number, order confirmation number.
  6. Listen and write down the non-toll-free number of the Acrobat support line.
  7. Call non-toll-free Acrobat support line, wait on hold for 10 minutes.
  8. Connect with Acrobat support technician, summarize the problem.
  9. Confirm that yes, I'm running Panther. Further explain that Panther translates to OS X version 10.3.6, more than enough to meet the 10.2 minimum requirements. Wonder if perhaps support technician ought to know that Panther is better than 10.2.
  10. Ask which file I was supposed to delete in order to try again? Confirm that I didn't have that file, as I hadn't even made it that far in the process. Suggest that download wasn't the issue, whereas software quality might be.
  11. Listen and write down the toll-free number of the Online Store.
  12. Call toll-free online store support, wait on hold an additional 8 minutes.
  13. Summarize problem, explain support technician recommended a download reset on the online store.
  14. Receive confirmation of reset, hang up.
  15. Re-download download manager. Install. Watch it crash.
  16. Call back toll-free online store support, greet support line attendant with a growing sense of familiarity.
  17. Summarize problem, explain again that it really doesn't seem like the download manager wants to work on my Mac, ask what further options are available.
  18. Confirm that no, I don't have "Netscape" on my Mac, but okay fine, I do have Internet Explorer 5.02.
  19. Re-download download manager before hanging up; confirm that nothing has been fixed, suggest that perhaps the browser is not the issue since the crash happens only upon install, and not during download.
  20. Confirm that yes, okay, I do have a Windows computer.
  21. Boot Windows computer. Ask if I must really load up Internet Explorer to do this. Confirm that yes, I suppose I do have "Netscape" on this computer. Use.
  22. Login to online store. Download download manager, install, and run.
  23. Thank attendant for help.

Adobe has always been an extremely Mac-oriented company, by necessity. Have they lost touch with a large part of their core customer base? Needless to say, shrink-wrap is all that I will buy from now on, if at all possible.

Permalink › | 54 comments

Early Presents

December 9

A couple of early presents for you, available in this site's new Downloads area.

After writing about my custom editing controls last week, I've had a few requests to share. So I built a ZIP archive and supporting page for the code and images involved, now available as the rather lofty-sounding v1.0. A bit of knowledge is required to implement them, since documentation is a thankless task. Still, I've included a readme file which should hopefully answer any questions in advance.

And vector line-art being what it is, taking the winter header image I'm running during this month and converting it into something a little more consumable was a pretty brainless task, so along with my postcard submission to Jeff Croft's contest, I've uploaded a few wallpaper-sized variations for your enjoyment.

Both available at the new Downloads page, which may or may not grow as time marches on.

Permalink › | no comments

Anti-Aliasing Tweaks

December 2

Horizontal kerning is pretty easy in Photoshop: use the Alt or Cmd + arrow key combination, adjust at will, repeat as necessary. But what about vertical kerning?

The problem is this: it's relatively easy to match kerning between words in Photoshop by using the built-in kerning controls. Adjusting anti-aliasing requires a bit more effort, but it remains completely do-able.

However, there seems to be a built-in dependency on where you set type on the canvas which I can only describe as a vertical anti-aliasing starting point. It appears that the fine adjustment of horizontal letter-spacing isn't enough control, as baselines can either fall on a precise pixel, or on a percentage point somewhere between two pixels.

Examples of normal and fuzzy type

In the first image to the left, the two lines of type have been kerned equally. The top example is crisper, whereas the bottom example is fuzzy, particularly around the bottom of each letterform.

Close-ups of the e letterform

In the second image to the left, the two "e" letterforms have been scaled up a few hundred percent, and in the bottom row a white grid has been overlaid. This close-up shows that the left "e" has a more solid letterform, especially on the horizontal strokes (the crossbar, the stroke under the bottom counter). The right "e" has much heavier horizontal strokes, anti-aliased across two pixels, and looks more blurry and wispy.

The main difference between these type examples is where they were set on the canvas with the type tool. What appears to be happening is that the starting point can either be an integer pixel value—and thus render closer to the pixel grid, as in the first example—or a fractional pixel value, which throws the entire letterform slightly off the pixel grid.

So the problem then, is how to control this? Once type is set, horizontal adjustment is possible in sub-pixel values thanks to the kerning tools. But vertical adjustment only appears to be possible with the Move tool, which operates on whole-pixel values.

The only solution I can see is to keep re-setting with the Type tool, and hope that through trial and error it'll eventually set properly. Someone tell me there's an easier way.

Addendum: And there is an easier way. The answer comes from Scott Gould:

You can move text vertically in sub-pixels, as long as you’re editing the text in question (i.e. text is highlighted or there’s a text cursor flashing away), by using the Ctrl/Apple key temporary move tool. For reasons that sorta, kinda make sense, this won’t work on “Sharp” or “Crisp” antialiasing. With “Strong” or “Smooth”, you’re all set.

Works great, thanks Scott.

Permalink › | 34 comments

Web Stock

December 1

The imagery needs of most web sites are modest; haven't stock photography providers figured this out?

Due to download times and the frequent need for dynamic content areas, the web approach to photography is often different than print. Buying one single photo for a large print run is commonplace; buying one single photo for an entire web site is limiting.

The web designer needs:
high-quality, but comparatively low-res images
re-usable imagery
a large selection of easily usable photos to fill dynamic areas, some of which may draw from pools of dozens of variations
nominal fees for volume, royalty-free photography
The average stock provider offers:
Royalty-free singles at pre-determined resolutions
Rights-managed singles that have a price calculated based on size, impressions, and usage
photo CDs for $300-$600 or more

I'm all for paying for the use of someone else's creative work, but it strikes me that there's a fundamental difference between the average web designer's needs, vs. the offerings of the likes of Veer, Getty, Corbis, and the rest. Ideally I'd be buying themed photo CDs and building up my own library, but doing so involves paying for print-resolution photography.

It could very well be that there's no business case for providing stock for the web; I'd expect lower prices for the lower resolution, but of course the content of the photo is what we're really paying for. Perhaps there's a break-even point that targetting prices for web use would fall under; perhaps not.

Increasingly I've been turning to the "dirty little secrets", the open photography providers like iStockPhoto and stock.xchange. Both allow sharing of digital photos, and re-use in commercial projects. The quality is hit and miss, but the prices are low (iStock) or free (stock.xchange). The resolution is usually a few megapixels, which is rarely, if ever, not enough.

I've started my own personal collection of stock from these sites that's starting to rival what I've had to work with in the past, with the added bonus of being made up only of shots that I like, rather than shots that happened to be included on the CD. I'm having a hard time finding a downside to this approach.

Am I alone? What's everyone else doing for stock?

Permalink › | 69 comments