About the new Amazon Light

Without making significant changes to the look and feel of the original Amazon Light (in large part because users have asked for it to remain the same), I've made the page weight much smaller (by about 15% cutting from 4k to 10k per page). I've also migrated from Amazon Web Services (AWS) v. 3.0 to Amazon's Ecommerce Service (ECS) 4.0, which is lighter weight on the backend and overall faster.

Amazon has moved away from their old AWS 3.0 and is now fully entrenched in their new version of this called Ecommerce Service (ECS) 4.0. What's nice about this is that I can ask Amazon to send me just the data I need and no more, so server-time is much faster. Also, many of the quirks and problems with AWS3 seem to have vanished with ECS4. Plus there are features in ECS4 that are too good to pass up, like "Availability Filtering" and "Super Saver Shipping Eligibility" among others.

One of the most troubling (and embarrassing) things about the original Amazon Light was that it was so poorly coded, in my opinion. Yes, it worked, but only just barely. One peek beneath the covers (server-side or client-side) would reveal horrible kludges, broken HTML, error-ridden (and unnecessary) javascript, etc. It's another example of "well if I had known this would get so popular, I'd have done it right from the beginning." - and a few other valid excuses, including a mad-dash rewrite of the UI under pressure from Google's Legal department, who felt the original violated their copyrighted "look and feel".

Over time I've made various improvements to the back-end code, but largely ignored the front-end (HTML/CSS/Javascript). Why? To be honest, because I could. The code was terribly invalid - but it "worked", meaning my target audience of modern graphical brower-users never saw any penalty for my bad code. The browsers (Netscape, Internet Explorer, Firefox, etc.) "healed" my poor code in a generally predictable way, so everything looked hunky-dory.

Some of the biggest complaints I've been hearing lately about Amazon.com is the poor way things seem to go when third-party sellers become involved. The quality of service can be spotty, it's a bit confusing figuring out who you are buying something from (and if you should trust them as much as Amazon), but most often it's because of the nice features Amazon offers on the items only Amazon sells - primarily Super-Saver Shipping and the Amazon Prime program.

So Amazon Light is now 95% Amazon-only. Third-party sellers are excluded to the best of my ability (a few squeak through, thus the "95%", and not "100%"). If you see an item listed on this site, you can be pretty sure that it's being sold and fulfilled by Amazon.com.

In some cases, even those items still don't qualify for Super-Saver Shipping - usually weight-based, for example HDTVs tend to not qualify. When an item is not eligible for Super-Saver Shipping (or Amazon Prime), it will sport this label clearly on the page:

One of my personal complaints about Amazon for years has been their habit of surfacing unavailable items in search results. If The Item Is Not Available, Why Show It To Me? Of course I know (or can make a good guess about) the answer - it beefs up search results, puts more product in front of people, and holds out the possibility of cross-selling and/or customer-initiated substitution.

Now, with a new feature of ECS4, I can finally filter out unavailable items - so everything on this site is actually available for purchase (yay!) The one oddity is that pre-orders seem to qualify as "available"... Hmm, guess I'm okay with that.

Since I was going to be re-working the guts of the site from the ground up, and was shooting for the most modern up-to-date standards-compliant model I could reasonably use, why settle for HTML 4.01 Strict and not go with XHTML 1.0 Transitional or Strict? Isn't HTML 4.01 old news? Isn't it passe? Isn't moving to XHTML the right thing to do?

In short, no. For my purposes, and for what I need to do here, HTML 4.01 is perfect. I get zero benefit from having this site be XHTML Transitional or Strict. None. Let alone the hassle of trying to decide if serving XHTML (supposedly XML) with a mimetype of "text/html" (not xml) is useful or serving it as "text/xml" is worth the hassle (poor browser compliance - IE mostly).

And before I get a load of email trying to "educate" me. I have done my homework on this, and am NOT saying that old invalid HTML is better than Standards-Based methods (different argument). I'm just saying that once you've got your code whipped into top standards-based shape, there isn't much reason to make it XHTML, as far as I can tell. If you take the HTML of this new version of Amazon Light, and change just two or three characters, it is a fully valid XHTML 1.0 Transitional document - yet there's no reason for me to do so, that I can see. So HTML 4.01 is the way to go at the moment.

And yes, there are a few pages in here that may not be fully-compliant, but It's damn close, and my semantics are a long ways off, but that's next on my list, and micro-formats would be useful in many ways, and... well, I digress, and feature creep gets the best of me.

-Alan Taylor 2/13/06