Legal advice from automated testing tools?
| Cerbera wrote: |
| The only things you can be sure about is that a table without headers is not a correctly formed data table - but even that still doesn't mean it's a layout table. |
I think we've all seen plenty of those. It would be good if your tool could feed back on how accurate a certain test is, or provide a method for the author to disagree with the test result. this would provide valuable feedback on tool performance, helping to make it more accurate in the future. It would give you some idea of how accurate the tool is at any given time. A manual check of course should be able to identify each type of table easily.
My point is that a Site with headings in the wrong order will fail a SiteMorse test. A site with no headings at all will pass (at least it used to). The worry with automated ranking is that authors learn how to cheat the results to gain a higher ranking. Your tool should be as flawless as possible or otherwise try and give an idea of the accuracy of any given test.
Share: 
I think we'll have to post some examples of our new reports online for you to look at - any suggestions for a suitable website (or two) to test?
Share:
I'd like to see what it makes of the ultimate automated tool test:
Bruce Lawson's Geocities 1996
Share:
GB, that's evil of you, lol.
Here are some sites you might want to test your tool on:-
- http://www.satzansatz.de - example of very simple yet attractive design, with good accessibility and almost perfect coding.
- http://www.gtaforums.com - a large internet forum using the Invision Power Board system. Should help to test detection of layout tables, poor structure, presentational markup, decorative images, invalid code, poor use of CSS, improper ALT text for user posted images and a myriad other bad practises.
- http://www.calthorpepark.hants.sch.uk - pages generated entirely by Publisher. Has quite a lot of pages, with varied elements and useful content, much of which is only available as PDF files. Could be useful to test the reliability of your parser under extreme conditions and whether you can spider a website which has no site map and no global navigation. It does not use conventional hyperlinks in places and some links are inaccesible without Javascript, which your tool might not be able to handle.
One way to implement this whilst catering to normal clients would be to provide a notice at the top of the results, stating that the results were limited to everything contained within the specified folder (including subfolders) but no other parts of the site were tested. Providing a "run the test again for my entire site" link after the explanation would make it easy to get a full site test if the user didn't mean to request a limited test.
Maybe something worth looking into? Btw, good to see Silktide are continuing their peer review and staying involved in the developer community, as well as continuing to improve their tool. Oh, any ideas when the scoring system will be made public?
Share:
The new version allows you to specify a list of urls to spider (you can for example now merge multiple domains - www.microsoft.com for instance annoyingly puts most of it's urls through a tracking g.msn.com domain). You can also specify urls to exclude.
At the moment we're considering adding language filters (i.e. "only include Chinese pages") because the testing conditions for languages are so different and they're often hard to split up by path alone.
Sitescore Enterprise is currently in Beta, we have selected people using it already. We're hoping for a public Beta in January, with the free cut-down version Feb-March 2006.
So here's another one - what tests would you like to see? So far we cover:
* W3C / WAI / 508 compliance
* Quality of URLs ('dirty', too long, human readable, bad query parameters etc)
* Broken links
* Use of CSS (for layout vs tables, external stylesheets, valid stylesheets)
* Alternative text (ALT / longdesc etc) - whether present and quality of
* Size of pages, download speed graphs for specified target audience(s)
* Distribution of file sizes and types
* Readibility - various metrics for quality/reading ease of text
* Meta tags / quality of
* Amount of content / distribution of
* Incoming links (from other sites), quality of
* Outgoing links (to other sites)
* Search engine placement for specified terms / engines vs competitors
* Search engine optimisation (use of various core techniques)
* Cookies (detect, compare with policies)
* Legal / brand compliance (custom rules to ensure websites meet internal guidelines)
* Popularity (Alexa, maybe tie into Google Analytics one day)
* Spelling (with custom dictionary)
* Security (check forms for bad inputs, common security vulnerabilities etc)
That's from memory so I've doubtlessly forgotten something!
So any more suggestions?
Share:
oh, is it really SILKTIDE, them selves making comments? 
I apologize for butting in here, but love your free testing service. I do have a few points to point out, if I may.
When logged in and it says test up to 25 pages, I would like it to test as I specified 25, not 15. I also have had the oppourtunity to use the site via several browsers, none work the same. My preference is FireFox, as all the stuff lines up in what I assume is the right manner.
My purpose in joining this forum just the other day, was in fact my silktide accessibility score. A rather nice score as it is, but have been wondering how to improve it. Why, cause I can, or I think I can, maybe not.
Wow, I think I am going to like it here in this forum.
opps, yes me again. Had another thought. Have you ever mis typed your url in the test area? I have done that, and silktide has offered me results test scores for a url that does not exist. How can that be?
Share:
Well there are known problems / limitations with the existing free service, but as said earlier we're working on a replacement. We never expected it to be quite so popular!
| Quote: |
| When logged in and it says test up to 25 pages, I would like it to test as I specified 25, not 15 |
It's trying to find 25, but only received 15 it could test and ignored the rest.
| Quote: |
| I also have had the oppourtunity to use the site via several browsers, none work the same |
There are some incompatibilities yes, especially with the progress bar (code we wrote back in the days of IE 5).
| Quote: |
| My purpose in joining this forum just the other day, was in fact my silktide accessibility score. |
Not too hard - make your site XHTML W3C compliant and most of the work is done for you! Of course with Sitescore 2 we're a lot more discriminating...
Share:
Thanks! It is nice that you responded to my note. I didn't mean to sound picky although that is exactly how it sounded.
Fairly new to css, validation and so on, I don't know if I am ready to convert to xhtml yet.
I look forward to the new service, although your mention of it has slightly set me askew. lol
Share:
hmmmmmmmmmm, this is not the thread really to post this, but since it was suggested here, I thought I would respond here. Feel free to move if you must.
I did convert my site to xhtml, it does validate, no errors and no warnings.
My silktide score has gone up and down. It makes me crazy.
It is finding a page with no css, well it sees something I don't, and I know every page accessible to bots. I do have old examples inside a test folder that is supposed to be protected by a robots.txt file.
It finds a page that it says is almost devoid of text, then when you follow the little number 1, it shows a mix of my 404 page and my sitemap.html page? Wow I can't figure that out.
Accessiblity score has gone up, but my two #10's are gone.
Marketing still is horrible, but then I am not marketing. Links to could use some help, but heck that comes with time.
Nurnan, sorry, it seems I am beginning to be a pest.
Share:
| Cerbera wrote: |
| GB, that's evil of you, lol. |
Indeed
but if you really want to find out the limitations of your tool, that's the way to do it.
| silktied wrote: |
| So here's another one - what tests would you like to see? So far we cover:
* W3C / WAI / 508 compliance |
Hmm, you are only testing partial compliance (I guess 50-60%). I understand the need for clear marketing, but don't go to the dark side. At least have a disclaimer. Even SiteMorse had that.
Do you have a list of criteria for testing so we can see what you are checking for?
Last edited by Grant Broome on 20 Dec 2005 02:01 pm; edited 1 time in total
Share:
| Quote: |
| Hmm, you are only testing partial compliance |
Sorry do you mean at the moment (with our existing Sitescore tool) or "W3C / WAI / 508 compliance" is only partial compliance?
Share:
| silktide wrote: |
| Sorry do you mean at the moment (with our existing Sitescore tool) or "W3C / WAI / 508 compliance" is only partial compliance? |
I mean that you can't fully test for W3C / WAI / 508 compliance with an automated tool. I'd go as far as to say that you can't fully test a single priority 1 checkpoint using automated tools:
http://www.gawds.org/show.php?contentid=147
That's not to say that they aren't useful, but customers like to think that if a computer test tells them that their site is accessible, then it must be, and being the zealot that I am, I think it's important to let the customer know that manual testing is a very important component of accessibilty auditing.
Share:
Just tested my site against the silktide tool and discovered I'd improved to an 8.4. Nice.
However, I was puzzled by this:
| Silktide Sitescore wrote: |
| The following 7 features were specifically identified: ...Search facility, Discussion forums, Contact details, News, Downloads, Moderate profanity, Website map. Generally, our analysis detected a very positive selection of text and features |
I wasn't particularly aware profanity was considered a 'feature'. I'll have to offer it more often
Jack Pickard The Pickards Information Services| Blog | Twit
Share:
| JackP wrote: |
| I wasn't particularly aware profanity was considered a 'feature'. I'll have to offer it more often |
I wonder how the script of Pulp Fiction would score
Accessibility != Bobby
Share:
| Cerbera wrote: |
| ...Here are some sites you might want to test your tool on:... |
Did Silktide ever come back to anyone with those results?
Share: