Forum Moderators: open
There is no evidence that Google currently uses any "page quality" metrics, but it is very possible that at some time they will do so. Popups and dead links would be two possible targets of any "page quality" metric.
Assuming that Google did start evaluating page quality, what features do you think they would take into account?
- Multiple (over three mabye?) applets / embeds / objects.
Jordan
- javascript (the more, the less quality imho)
- css (if it isn't there, less quality imho)
- filesize (although this is kind of already done for pages over 100k I believe)
Of course, those are only probability things. It is just 'less probable' that a site would be high quality if it has lots of javascript and no css. It doesn't rule it out though. That's the nature of these things though.
I'm sure there's more, but that's all I can think of off the top of my head. It would be an interesting addition to the algorithm if you ask me.
Don't get me wrong, I'm not beeign anal about this, jsut that I've often thought that if google would penalise spelling errors, a lot of people would start spell cehcking their sites, AND ergo, top results would contain well seplled clean pages made by people who really care about their pages, rather then fire-and-forget mess-content.
SN
I use markup as content, CSS for form, and JavaScript for tasks (isn't that the way the standards are written to be used?). A quality site often needs a good amount of tasking (e.g., see Gemal's Psyched Blog in the BrowserSpy section; not my site, but a good example of the kind of thing I'm referring to). No banners or spam or anything like that, just tasking in order to let you see different information about your browser, which AFAIK could not be accomplished any other way.
For example, just a few things that could not be done (client side) without JavaScript, which almost any quality site would want to do at least some of:
Cannot hilight text on page.
Cannot dynamically evaluate / change content.
Cannot load new pages in frames or iframes except from links (not from a button for example, and technically, the "target" attribute of links is deprecated in XHTML even though it still works in the bowsers).
Cannot link buttons to any other content.
Cannot read text input values.
Cannot do math.
Cannot alert or prompt or warn users.
Cannot determine content dimensions (for whole or individual elements).
Cannot determine current value of listbox content.
Cannot get or set cookies.
Jordan
Usefulness of content is what matters, and I don't think any of the metrics mentioned are strongly correlated with usefulness of content. In fact, I don't think there is any decent metric for usefulness of content. Page Rank and related linkage analysis methods are probably the best things anyone has come up with so far, but clearly they're flawed, blunt instruments.
G has a content-centric view of search, and I doubt they'd dilute that by handing out rewards and punishments based on coding issues.
Pete
A mini survey I did in the offices around me showed that "ordinary" people thought pagerank was about a sites quality of content.
That is probably as it should be, but I cannot imagine any way that an algo could do this, this has to be done by a human reviewer.
And then we would be thinking in terms of the DMOZ problems... :) :) :)
b) pr seemingly was supposed to be about site quality, but the fact that it was also tied to serps created an exploit which many webmasters employ: mining for pr. it would seem that, in a perfect world, sites would only link to other quality sites. and, if you think about the internet as a community, as it was much more of in the mid/late 90's, then folks would only link to other sites that were either on-topic, or considered cool. unfortunately, now there are webmasters securing 30,000 inbound links, buying pr, and putting no-index, no-follow on pages with outbound links, in an attempt to mine and hoard pr. unfortunate, but true. can't blame anybody for it, although i wish that weren't the case.
when i think of a quality site, i think of clean design, no script errors (html errors who cares...like closing font tags, no alt tags in images, etc), and easy to use navigation. but, then again, i've seen some wicked cool sites that google can't even accurately crawl. i don't see how an algo could do page quality. that's especially true when you look through dmoz, and see that even human editors have to leave page quality at the doorstep, and focus instead on content quality.
it would be nice if page quality could actually be measured. then we might actually get the most desired results, instead of artificially inflated pr, spammy keyword layden sites that 'appear' to be most relevant to the algo.
se results tend to remind me of school text books, while there are tons of works-of-art on the web that don't ever get ranking. my bet is that someone will try page quality metrics, however, and i wouldn't put it by the UI happy folks over at MS. their sites are all elegantly done (in a corporate kinda way), and easy to use...even though they're huge. even though i don't use their OS for anything more than a paper weight/file server, i do think that they're the most likely to try and implement any type of page quality metrics to their serps, as a differentiating factor.
anybody entering the se market will have to come up with a competitive advantage that brings in users. msn being fully established already, they will have to spark the curiosity of surfers who don't use their engine with some sort of competitive advantage. pqm would make sense.
sorry, a little long winded.
I dissagree. I have found that most sites/pages which are via Frontpage often have good quality content. At least compared to those written by Webmaster that make the assumption everyone else on the planet has a super high speed connection just like them. You can pick these sites easily as it takes at least 10 minutes to figure out the reason for the sites exitence. That's after getting past the 10 minute Flash intro.
Short to the point decriptive text and Short to the point decriptive links is number 1 IMO
Dave
Also, broken links using c://mypc/mysites/iuseddreamweaver1/iamadummy.html should be banned for life.
Actually, I just think Google should boost sites with good clean html and navigation. Oh, they do already?! OK, how about sending to everlasting Internet pergatory all those affiliate sites that deem to put garbage like www.google.com&search=mydomain.com, www.whitehouse.gov&search=braincells in their html?
Bit hard as it all comes down to personal taste. Besides, most search the net for info, not good colour schemes :o)
Dave
If doing things properly (2 minutes more per page) would get you higher up, more would do so, and Google could have a far reaching, deep influence over th equality of the web.
I have plenty of old spaghetti sites, and I'd be glad if they get borried in the serps until I canfix them, they're an embarressment. ;)
SN
I do realize, though, that Google can't avoid influencing the web in some ways--all the linking and not-linking that goes on to try to game the system, but still, to me, there's a difference between creating a system that will have unintended but unavoidable consequences and intentionally trying to impose one's idea of what the web should be.
Natural said:
my bet is that someone will try page quality metrics, however, and i wouldn't put it by the UI happy folks over at MS. their sites are all elegantly done (in a corporate kinda way), and easy to use...even though they're huge. even though i don't use their OS for anything more than a paper weight/file server, i do think that they're the most likely to try and implement any type of page quality metrics to their serps, as a differentiating factor.
That's an extremely interesting and extremely frightening thought.
>> I disagree. <<
There is an awful lot of junk code out there, much of it very easy to tidy up.
As an example of something seen recently, changing from this style of code:
<br></font><br><b><font face="Times New Roman, Times, serif" size="2">COUNTYSIDE <br>
<br></font></b><font face="Times New Roman, Times, serif" size="2">North
Countyside Widget Centre<br></font><font face="Times New Roman, Times, serif" size="2">
Someplace, Nr. Town, Countyshire CC00 0AA<br></font><font face="Times New Roman, Times, serif" size="2">
Tel: 01000 000000<br></font><font face="Times New Roman, Times, serif" size="2"><b></b>
Widget Starter : approx £80<br></font><font face="Times New Roman, Times, serif" size="2">
Widget Foobar: approx £80<br></font><b></b><font face="Times New Roman, Times, serif" size="2">
Widget Weekend: approx £100<br></font><font face="Times New Roman, Times, serif" size="2">
Contact: Some Person</font><font face="Times New Roman, Times, serif" size="2">
<br></font> <br><b><font face="Times New Roman, Times, serif" size="2">COUNTY</font></b>
<b><font face="Times New Roman, Times, serif" size="2"><br><b></b></font><br></b>
<font face="Times New Roman, Times, serif" size="2">Someplace
Widget Centre<br></font><font face="Times New Roman, Times, serif" size="2">
Someplace Road, Someplace TT00 0AA<b><br></b>
Tel: 01000 000000</font><font face="Times New Roman, Times, serif" size="2"><br>
Contact: Mr Someone</font></p>
<p><font face="Times New Roman, Times, serif" size="2">Big Manor Widget
Centre <br>Some Lane, Little Village<br></font><font face="Times New Roman, Times, serif" size="2">
Someplace, County, MM00 0AA<br></font><font face="Times New Roman, Times, serif" size="2">
Tel: 01000 000000</font><font face="Times New Roman, Times, serif" size="2"><br>
Email: info@widget.co.uk<br></font><font face="Times New Roman, Times, serif" size="2">
Web site: www.widget.co.uk <br></font><font face="Times New Roman, Times, serif" size="2">
Contact: Jimmy Foobar<br></font><font face="Times New Roman, Times, serif" size="2"> <br>
Actually the above code isn't as bad as some examples I could include. I really hate those with thousands of <o:p> tags dotted throughout, and embedded inline CSS styles on every tag.
.
Changed to this style of code, as basic HTML. It would have been even better to dump the <font> tags completely and do it with CSS instead, but that is for later:
<h3 align="left">COUNTYSIDE</h3>
<h4 align="left">North Countyside Widget Centre</h4>
<p align="left"><font face="Times New Roman, Times, serif" size="2">
Someplace, Nr. Town,<br>
Countyshire CC00 0AA<br>
Tel: 01000 000000<br>
Widget Starter: approx £80<br>
Widget Foobar: approx £80<br>
Widget Weekend: approx £100<br>
Contact: Some Person</font></p>
<h3 align="left">COUNTY</h3>
<h4 align="left">Someplace Widget Centre</h4>
<p align="left"><font face="Times New Roman, Times, serif" size="2">
Someplace Road,<br>
Someplace TT00 0AA<br>
Tel: 01000 000000<br>
Contact: Mr Someone</font></p>
<h4 align="left">Big Manor Widget Centre</h4>
<p align="left"><font face="Times New Roman, Times, serif" size="2">
Some Lane, Little Village,<br>
Someplace, County, MM00 0AA<br>
Tel: 01000 000000<br>
Email: <a href="mailto:info@widget.co.uk"
title="email Jimmy at Big Manor">info@widget.co.uk</a><br>
Website: <a href="http://www.widget.co.uk/"
title="Big Manor Widgets"
target="_blank">www.widget.co.uk</a><br>
Contact: Jimmy Foobar</font></p>
That was a basic start. I realise that there are a lot of other things that can be done, but the main one was to get rid of bloat (knocking 25% to 35% off most files), eliminating nesting errors, adding structure (<hx> tags) to the data. Each page is now about 25K. Next step might be to subdivide the content a little more.
Cannot hilight text on page.
Cannot dynamically evaluate / change content.
Cannot load new pages in frames or iframes except from links (not from a button for example, and technically, the "target" attribute of links is deprecated in XHTML even though it still works in the bowsers).
Cannot link buttons to any other content.
Cannot read text input values.
Cannot do math.
Cannot alert or prompt or warn users.
Cannot determine content dimensions (for whole or individual elements).
Cannot determine current value of listbox content.
Cannot get or set cookies.
Don't get me wrong, I'm not saying that sites that ALL sites that use JS are bad, I'm just saying that the probability of a good site having lots of JS is small. Same as the fact that sites with no incoming links aren't necessary bad, just a higher probability.
pages w/ large flash intros, and, generally pages using flash extensively should pe penalised.
ehem, isn't that already a self-inflicted penalty? Flash sites rarely rank well, because they are using flash and as a result have very little content in the serps.
- spelling errorsDon't get me wrong, I'm not beeign anal about this, jsut that I've often thought that if google would penalise spelling errors, a lot of people would start spell cehcking their sites, AND ergo, top results would contain well seplled clean pages made by people who really care about their pages, rather then fire-and-forget mess-content.
Did I just find four spelling mistakes in there :-) I like the idea of checking for spelling, but considering the large amount of user submitted content out there, I don't think it would work. I feel that similar to the flash thing, there is kind of a penalty for that already anyway. After all, if there are typos, those sites are missing out on coming up higher in the serps for the term spelt correctly.
That part was not meant to be serious. If you like, I can do this <joke>in future</joke>!
this is why page quality metrics will never work on a se designed for the lowest common denominator. too many people have no sense of style.
yawn.
Erm yes. Most would at least use a smilie if they ARE joking.
E: There is an awful lot of junk code out there, much of it very easy to tidy up.
Yes it is easy to tidy up.......if you know HTML. Most that use Frontpage etc know little or nothing about HTML. I find these sites stick to good quality content without all the bells and whistles. True they are larger than need be, but as it's not Flash etc the extra time to download, in most cases, is not noticable.
Quite frankly, I wish some of these 'so called' Web designers would go back to basics and stop adding all the fruit they can lay their hands on simply because they know how to.
Dave
i hope that everybody that sees graphic rollovers and flash and other eye-candy type features are unnecessary, or should be penalized, aren't as boring irl. a whole closet full of white, shortsleeve, button-down shirts, and poly-blend slacks in different muted dark colors, and a matching pocket protector for every day of the week?
lol. fair point. eye-candy has its place on the web, certainly.
Unfortunately, many of those sites are
a) slow to load
b) badly done because they are done by people who decided they should just use every gadget available rather than the ones best suited to the job.
I love eye-candy and do a lot of work in flash - I limit the distribution of that pretty much to CD-rom though, because I can bundle it as a projector and won't be inflicting bandwidth congestion on anyone. I do a lot of animation for tv as well and will make that as complex and as enjoyable to watch as possible. No restrictions there.
It's all about knowing the medium you are working in. Certainly there are times when you can use the web to dazzle people and I have seen flash and javascript used well for that purpose. Excess use of both of those though (in my experience) has never been a good thing and will rarely be what I am after in the search engines.
I can see some argument on checking for dead links. Maybe that is one area where webmasters should be responsible for making sure they still work. This is a major time-taker for us. Nowadays we dont link to any inner page, unless we are almost 100% sure that the url wont change. So if we want to refer to newspaper news pages, we will link to the newspaper domain only, and not the actual page. That has saved a lot of time. We assume that if people really do want the exact page, they can find it by cliking on in the referred domain or doing a search (if indeed the actual news item still exists at a later date)
I cant see any justification for penalization for using js or NOT using CSS. We use CSS in several sites, but in some, it is just not needed, and the idea that all sites should naturally or eventually use CSS and are somehow lower quality if they don't is badly flawed.
There are more important things however tha Google still have to perfect, - like dropping all redirect pages whatsoever, and those with substantial amounts of duplicate content.
I agree with you for the most part, but a few points:
Use CSS - just look at the site you are on now
I was meaning like this, e.g., [squarefree.com...]
It could be done partially in pure CSS using a selector like *[hilite], but even at that you'd need JS to set the arbitrary hilite attribute on content elements according to user input.
What do you mean? Rollovers?
I was meaning actual input buttons, e.g.,
<input type="button" value="Do Something" onclick="func();"/>What's a form for?
*Looks down, shuffles feet...*
I never really learned forms...I always thought that they had to be used with server side to get information from them (i.e., action="/path/to/cgi.pl")? Looks like I've been mistaken about that?
Better to not use JS, that way EVERYONE gets the message, not just those with JS turned on. Particularly if it's anything important. Just create an intermediary page with the message on it.
Good point. I agree that for alert and warning another page is usually best. But I don't know about for prompt...I'm not sure how that would be possible.
Jordan
Basically, a lot of the things you mentioned are problems because you are relying on client side only. But why not do those things server side? It is so much more elegant to do a lot of that stuff through something like Coldfusion or PHP that won't be dependant on whether a user has JS or not. There's just too often sites that I go to that simply won't work because they are using too much JS. Most hosts offer some server side scripting at least and it really isn't that hard to learn. I find it easier than JS scripting anyway.
For prompting (I presume if someone has left a form field blank for example?) , I would make a validating page and send the user back to the page with the fields marked in red that they needed to fill out. That way, the message will be there for everyone (not just those with JS).
BTW, I do use JS on my site also. I use it for popup windows (for larger versions of photos for example) and I use it for a form that evaluates a country and populates a city select accordingly. The only other places I use it are in my admin section, where I know there won't be anyone else using it. Other than that, I have gotten rid of it. I had rollovers, but I ended up finding them unnecessary and decided to live without the extra loading time. It's the excess use of JS that I feel gives a higher chance of a less high quality site. Again, I'm not saying it's impossible, I just feel it's less probable that a site would be as good as another without the high level use of JS. For starters, the one without the JS is more likely to be more accessible to a wider range of users.
Maybe counting CSS as a bonus isn't really necessary. Although font tags aren't ideal, they don't really affect the user experience. A simple filesize check may be a better concept, and that would discourage code bloat at the same time.
These are really only my personal opinions.. probably not grounded in any factual data at all :-)
Hmmm, those are some good points, especially about accessibility. Thanks for taking the time to explain your POV, I understand your inclusion of JS in such a metrics better now. :)
BTW, I have been planing on learning PHP, but I only recently got a host running PHP4. I guess it's about time to stop procrastinating and start learning...I should really do that...mabye later. ;D
Jordan
As an example, one of the things I cannot live without now is includes. This saves going through every page of a site everytime I want to change something like the body tag properties. And that has got to be one of simplest tasks for PHP (and even easier in Coldfusion).
Still don't see why that justifies Google penalizing sites?
The point is, why would the average user care? Google isn't attempting to deliver results to the .001% of people who know html.
Spelling would be tricky. Imagine what Google would do to this forum :)
Another data point, from Brewer's (Inktomi) paper about crawling the web several years ago: some 40% of html pages had some form of actual error (not just a validation problem). Someone may want to check my memory, but there's a lot of good info out there that isn't always on perfectly written pages. :) That said, I like the lists that people are mentioning (e.g. # of dead links on a site)..
I believe that properly written html helps since it speeds up the loading (page less heavy) but I am not sure about google counting the errors for ranking.
But hey SEO is a world of speculations ;)
Still don't see why that justifies Google penalizing sites?
Junk code may carry it's own punishment, without a search engine adding on any penalty. Parts of the page may have errors that can't be handled by the error recovery routines that are used. Also, junk code can make the page size much too large in relation to the text content - I know that some algos used to look at that ratio. And I know I've boosted pages that I inherited, without changing any conent but only cleaning up the tags.
One of the common problems with junk code is misuse of HTML tags. What if an algo boosts the value of spaces because that's all that exists inside many of the paragraph tags and the spaces have <b></b> around them?
