[Gloucs] Re: Web site building and ranking

[Guy Edwards]

I know it's normally rude to post a private mail to a list but it's
relevant and I've removed anything personal.

On Mon, 2003-01-06 at 11:03, bjh wrote:
> This is starting to get to be an interesting exchange and the 
> implications for building web sites for blind persons to see 
> are normally totally ignored by most site builders...

I think the reading devices used by the blind are either voice
synthesizers or braille screen readers, I don't know for sure. Either
way I thought they used a basic input from a text based browser, or
variation upon that. I know there was an article in LinuxFormat a long
time ago saying that SUSE could now be installed and run by a blind
person as they had added braille reader support to the SUSE install
right at the start.

If no-one minds me asking, is there anyone on the list using a braille
or voice device to surf?

> A possibly useful site to look at for this purpose is 
> http://vision.unco.edu/edtech/primer/ which examples effects of 
> a number of factors we have been talking about - may be worth a look 
> for you...

The link is good. Here's a snippet of something I hadn't thought of...

"White-on-white is a technique that provides extra help for users who
may require additional information. While sighted people are easily
distracted by a lot of extra words on a page, blind people often need
some extra information to help in navigation or to understand what is
happening. We put that extra information in a font that is the same
color as the background. Since most of our web pages have a white
background, we call this technique white-on-white. It would work just as
well in other colors." 

[ snipped ]

You've been coding 16x longer than I have :-) I've just remembered who
you are too. Is the date on http://www.gloucs.lug.org.uk/about.php
correct? E.g. Was the LUG founded in June 2000? I meant to ask many
months ago. 

I understand html has changed a lot but most of the massive early
changes were due to the browser wars, whereas now, with xml creeping in
and browsers priding themselves more in being bug free and standards
compliant, I really think it's getting better.

> Putting it into context - a site can be technically perfect in terms of 
> minimal html coding etc., but if it is never found, is a total waste of 
> precious time and resources (time is money) taken to put it together.

Again, we have different audiences. With an Open Source project I would
rather have a technically perfect site full of content, as it will be
the program users seeking information, help, or discussion. I would
quite happily sacrifice search engine rankings for that.

With Open Source you aren't trying to attract customers - that's a by
product of good effort. Linus was just having fun when he made his modem
terminal, and everyone else joined in. He wasn't thinking about stealing
market share from Microsoft or replacing Unix in in corporations.

I wouldn't worry that an Open Source project isn't attracting new users
through its search engine ratings. I would worry that it didn't have
ease of use, reliable documentation or stable code. If you make
something half decent and put the effort into your project, people
appear from nowhere. I use mailing lists, newsgroups, websites and lots
of 2 way emailing (eg conversations with users who have asked for help).

> Technical computer books are not necessarily written by persons qualified by the 
> practical experience of the reality of using the facility, in fact many books 
> unfortunately spread complete nonsense, because the writer will use other 
> previously written material and regurgitate the information as FACT without 
> extensive TESTING on the net of the subject matter to see if it really
> does work.

I agree, I have a "teach yourself" book on html and xhtml but the xhtml
additions are a complete afterthought and most of the code will never
get through a validator. This is more a problem of tracking down the
right book (or other resource) I feel though. that and knowing which
bits to igore and which bits to remember.

This next bit might be of use to people so:

> Suggestion - based on experience - made to help our reader(s) (smile), 
> and save them months of wasted time, who may be thinking of building a 
> site or page for the first time - is:
> 
> Sit down with a piece of paper and pencil, decide on every possible key 
> word for the intended site which you think an internet browser may just 
> put into a search engine to find the subject matter...
[snip]
> Then go to the computer and submit each of those words (and phrases 
> or small group of words) into a number of different main target search engines...
> 
> Following each separate search, click on the number one site, and with 
> the site open on the screen, click on View on the browser header bar, 
> and then source, this opens all the html coding on screen used to build 
> the page... 
> you are now looking for each (if any) meta tagged keyword and 
> the actual order the keywords are listed - then click back on to normal 
> screen and check to see where your own selected search words were actually 
> visible on the page as text... - RECORD THE SEARCH ENGINE and INFORMATION 
> - YOU NOW KNOW THE FACTUAL INFORMATION TO BUILD YOUR SITE for that given 
> search engine and order of search words to achieve a high ranking 
> position... (ignore all references to graphics when doing this exercise!). 
> Only at this point can you even think about how a site is going to be built 
> to make it as visually attractive as possible to appeal to the viewer when 
> he/she has found it... (Perhaps with very careful use of colour alone and 
> the MINIMAL amount of text on each page to create the interest from the 
> reader without boring and also to achieve the search engine ranking 
> required - remember, each additional NON keyword will normally dilute 
> the effect of the keywords for ranking purposes - it is about achieving 
> balanced copy without trying to put a book on the page (like this 
> email is starting to look like - smile)...

I thought the bot did more than just grep (not grep, but you know what I
mean) the page for words. There's some kind of html parsing going on.

> You then proceed and carry out the lengthy exercise for each of the target 
> search engines and slowly a picture will develop for you to move to the 
> final stage of actually building your webpage for that given set of circumstances 
> - it may be that in order to get certain search engines to rank your site that 
> you will have to build different pages with basically the same information, 
> but structured in a totally different way so that the site will be ranked 
> highly by different search engines - if you have to do this, you then submit 
> the different pages to the specific search engine and you should achieve 
> utopia over a period of time - I have noticed that some search engines 
> will suddenly rank one of my sites anything up to two years after submission 
> for no particular reasons...

I'd say because it's ranking against a lot of other sites which rise and
fall as they're developed or removed. But there's more to it of course
(different passes in the first and second month are done by different
types of googlebot for instance)

> The fastest search engine to rank is probably at present Lycos, and although 
> they push paid for submissions, they will rank a free submission site after 
> about four to six weeks!  http://www.lycos.co.uk/service/addasite.html  , 
> forget what they say on the site, they do rank FREE submissions quickly - 
> they need the information you have on your site to build their data base 
> and attract new customers! 
> *****************************************************************************************

> > Do a Google for Linux cad viewer, or Linux dwg, and the sites at the top
> > of the search. Or do a search for Gloucester Linux , top again.
> 
> *** Sorry, but that is because you are being very specific - that is Gloucester Linux will
> only be found by someone who puts "Gloucester Linux" in the search engine at that time...

My argument would be, if you're looking for Linux in Gloucester, it's
the two words you type in. It's a very specific site. How do you Google
for Linux in Gloucester without using those two words? If they're after
"Linux Support" there's better sites out there to help them, why bring
them to this site? They aren't going to give me money and I'm not
providing a better resource that the bigger sites.

> linux, help, support, information, assistance, meeting, meetings, regular, expertise, 
> looking, for, programmes, programs, of, message, board, glos, gloucestershire, cotswolds, 
> uk, up, dates, updates, need, reviews, projects, code, codes, presentations, talks, experts, 
> begginner, begginers, on, distros, method, methods, booting, dual, windows, with, install, 
> installing, installation, dual, and, easy, I, use, mandrake, suse, Suse, SuSe, red, hat, 
> redhat, derbian, in, member, members, group, to, (or whatever)...

Right, for a start that capitalization doesn't matter (suse*3) as far as
I know, and second, "on, and, I, up, for, with, need" ..etc are junk. I
would refuse to put them in any site.

>From the present GLUG site:
 <meta content="Linux Gloucestershire GlosLUG Linux User Group
gloucs.lug.org.uk Linux support Cheltenham Stroud Gloucester Cotswolds
Forest open source mandrake redhat suse" name="keywords" />
                
We just slapped some words in, (Just noticed Linux is in 3 times - doh!)
I think we've got the main meat of it already. My future idea would be
to go through each of the GLUG sites PHP pages and add a set of tags
specific for each page. Same for Lx-Viewer (I'll probably do this
tonight).

[snip]

> The combination now starts to get large and the potential cachment factor huge 
> - use mis spelt words as meta tagged key words to get all the incorrect keyboard input!!!

This is where I decide I don't want junk in my html. If you misspell a
search, I'd expect the user to notice. Google will tell you, or you'll
end up with an obviously depleted or weird search result.

I'm not going to penalize everyone with a bucket load of html to try and
drag in misspellings. Just a choice I can afford to make. If I was being
paid wads of cash, I might change my mind.

That and don't you think it's possible that the search engines may
oneday penalize overloading of search tags? I've no idea what the limit
might be. You point below is quite valid.

> *** Some of us are starting to think that the Google bots will pick up 
> "booked mark" sites as links, the Google guys are very reluctant to 
> let people know how there bots really work - if they made the 
> information truly known we would all be using the info to fix our 
> listings!

Guy