Bots having trouble reading nested tables?

[LD]

"Bots and crawlers have trouble going more than 4 nested tables, so it depends on how complicated your build is."

This point above was taken from a topic on another forum regarding the practicallity of using tables vs. CSS in site design. One of the posts indicted that SE bots have trouble reading nested tables and that the assessment of a site is affected when reading content from tables if they aren't crawled in a proper order. Any thoughts?

[Jean-Luc]

I cannot believe that reading nested tables could be difficult for bots.

On the other hand, it is clear that using CSS gives you more freedom in organizing your source code, i.e. texts at the top of your source code can appear at the bottom of your pages and things at the end of the source code can come up first on the screen. This is a tolerated way to show things differently to bots and humans.

Jean-Luc

[LD]

Originally Posted by Jean-Luc

I cannot believe that reading nested tables could be difficult for bots.

On the other hand, it is clear that using CSS gives you more freedom in organizing your source code, i.e. texts at the top of your source code can appear at the bottom of your pages and things at the end of the source code can come up first on the screen. This is a tolerated way to show things differently to bots and humans.

Jean-Luc

I don't see how there would be difficulty in bots crawling the code, even if it's nested in the 5th layer table - code is code. However, how the underlying code/text content is validated and indexed in sites with multi layered tables, I guess, is the question. And I think it also was inferred that the assimilation and indexing of the "content" might not be done efficiently for those web pages. I'm not suggesting this, but I believe that was the inference.

[Dubbya]

It's not that difficult for anyone, including a silly bot to read table code, it's just that using CSS can save a ton of bandwidth, is more easily ported to additional formats (think mobile browsers or printed output).

Here's a simple comparison between what a bot would see when parsing the code in a table based site vs. parsing the code in a CSS based site.

<table width="500" cellspacing="2" border="1">
<tr>
<td align="left"><h3>This is a subject Heading</h3>
<table width="120" cellspacing="2" cellpadding="2" align="left" border="1">
<tr>
<td>Here is your first bit of text.</td>
</tr>
</table>
<table width="380" cellspacing="2" cellpadding="2" align="right" border="1">
<tr>
<td>Here is your second bit of text.</td>
</tr>
</table>
</td>
</tr>
</table>

Seems simple enough, right?

Now check out the same information presented using CSS.

<div id="content">
  <h3><b>This is a subject Heading</h3>
  <p>Here is your first bit of text.</p>
  <p>Here is your second bit of text.</p>
</div>

With CSS formatting linked to externally (at least it should be), the bot can get straight to the meat and potatoes. You reduce the bandwidth required to navigate your site, make indexing your pages less complicated for the spiders, make pages easier to edit and don't have to change the content to port it over to mobile browsers, screen readers or printed output.

While bots might index your entire page regardless of the layout method used, I'd ask you what you'd rather feed them. Code or content?

CSS... It's the right thing to do.

[LD]

Originally Posted by Dubbya

While bots might index your entire page regardless of the layout method used, I'd ask you what you'd rather feed them. Code or content?

CSS... It's the right thing to do.

I understand. It was just the inference from the poster that indicated the following:

"Search engines can't read everything in tables and it becomes even worse if you have many nested tables, where most table layouts will become nested too deeply."

I was looking for clarity on this, which is being reiterated in the replies here. However I do appreciate the CSS info as well.

[opalcomp]

Originally Posted by Dubbya

It's not that difficult for anyone, including a silly bot to read table code, it's just that using CSS can save a ton of bandwidth, is more easily ported to additional formats (think mobile browsers or printed output).

Here's a simple comparison between what a bot would see when parsing the code in a table based site vs. parsing the code in a CSS based site.

<table width="500" cellspacing="2" border="1">
<tr>
<td align="left"><h3>This is a subject Heading</h3>
<table width="120" cellspacing="2" cellpadding="2" align="left" border="1">
<tr>
<td>Here is your first bit of text.</td>
</tr>
</table>
<table width="380" cellspacing="2" cellpadding="2" align="right" border="1">
<tr>
<td>Here is your second bit of text.</td>
</tr>
</table>
</td>
</tr>
</table>

Seems simple enough, right?

Now check out the same information presented using CSS.

<div id="content">
  <h3><b>This is a subject Heading</h3>
  <p>Here is your first bit of text.</p>
  <p>Here is your second bit of text.</p>
</div>

With CSS formatting linked to externally (at least it should be), the bot can get straight to the meat and potatoes. You reduce the bandwidth required to navigate your site, make indexing your pages less complicated for the spiders, make pages easier to edit and don't have to change the content to port it over to mobile browsers, screen readers or printed output.

While bots might index your entire page regardless of the layout method used, I'd ask you what you'd rather feed them. Code or content?

CSS... It's the right thing to do.

That is pretty silly. The table text would be slightly larger, but not bloated the way you wrote it. Here is my version:

<table>
<tr><td><h3>This is a subject Heading</h3></td></tr>
<tr><td>Here is your first bit of text.</td></tr>
<tr><td>Here is your second bit of text.</td></tr>
</table>

Sometimes CSS users sound like they are preaching a religion. CSS is better in a number of ways, but tables are not always so bad, and should be used in data presentation, at the least.

[outlawtorn]

I agree with OpalComp, having been a 'table' users for years and now experiencing my developers working with CSS and DIV tags I can defintely say that 'appear' to be more hassle than they are worth, and yes, they do love to preach it from the rooftops, however I face a constant battle of getting the same results in different browsers through CSS and DIV tags. But this is heading off topic.

I can't see any reason why bots would battle to read deep into nested tables, are there any bot developers here that can enlighten us?

[Terry Van Horne]

Most info you read about any source code affecting rank is at best a guess and at worst complete speculation. Tables add a little extra code and ensure your presentation will be the same for everybody. The CSS devotees take for granted everone is running the latest greatest browser and CSS degrades terribly on old browsers.

[bluemi]

I can't put this into absolute numbers but would think it to be logical that the bots like Googlebot have a limit in time that they want to spend on your site and amount of data they want to "eat". If you give them 100 pages with 50 kb each they will need much longer to go through the whole site than with one that has 100 pages with 15 kb each. If then your server is slow the bot might decide it's not worth to come back soon - take a look at your crawl frequency in the Google Webmaster Tools.

I have recently taken over a site which was done with an old Dreamweaver version by a web"master" who had no clue about html. The pages had on average 60 kb, some up to 200 kb, and up to 10 (!) levels of nested tables. I have completely redone the site with css, cutting down file sizes to one third or one quarter of what they were before. After a while Google started indexing and updating the site much more often. I would therefore say that using css clearly gives you an advantage over nested tables.

[Webnauts]

Originally Posted by Terry Van Horne

The CSS devotees take for granted everone is running the latest greatest browser and CSS degrades terribly on old browsers.

That is pure BS Terry. Sorry.

I made a screenshot of my homepage with Interner Explorer 3, but splitted in 3 images. You better look at that before you go on spreading such ridiculous claims:

TOP: http://www.seoworkers.com/media/ie3-part1.gif

and

MIDDLE: http://www.seoworkers.com/media/ie3-part2.gif

and

BOTTOM: http://www.seoworkers.com/media/ie3-part3.gif


Now, which site with tables can we check with IE 3 next? If designers do not know to markup web sites, that is not a problem of the markup itself and its compatibility.

Also look at the text version of my homepage in Google cache: http://209.85.129.132/search?q=cache...&hl=en&strip=1

Which site with tables should we look at its text version in Google's cache?

[tenkiya]

People can't seriously be discussing tables used for layout in 2008.

[nullvariable]

Originally Posted by tenkiya

People can't seriously be discussing tables used for layout in 2008.

heck its practically 2009! I agree though, I thought tableless was pretty much standard as of 3-4 years ago now... as Webnauts says, if you're following hierarchical, semantical structure (which you should be) then you shouldn't use tables for layout or design. If the information you're presenting is best presented in a table (like a spreadsheet) that is something different entirely...

[Webnauts]

Originally Posted by tenkiya

People can't seriously be discussing tables used for layout in 2008.

That is not 100% correct. If someone knows at least HTML, he should already that tables were never there for layout. If you would check the W3C HTML specification, it clearly states that tables are for the presentation of tabular data. Is a web page a presentation of tabular data?

[linnetwoods]

Originally Posted by tenkiya

People can't seriously be discussing tables used for layout in 2008.

Well excuse me! Some of us don't feel the need to rubbish anything and everything that is more than twenty minutes old and none of us should feel the need to rubbish anyone who is doing things differently to ourselves, especially when most of us do what we do to the best of our ability which must, by the law of averages, vary.

[ami_iss]

1- There is no telling what bots will do and the only way to be sure is to run a googlebot simulator or look at results in Google analytics to see if there is any strange problem like no text scanned.

2- I don't think they should have trouble with tables more than anything else. They aren't that interested in the logic of the html. The things that stop a spider are inherently unpredictable because they are _bugs_ which means the designers didn't envision them either.

3- CSS is certainly useful for taming <H1> <H2> etc. and for making paragraph spacing be the height that you want.

4- The example given about is not one in which tables or CSS are needed particularly

5- I am certain that all Webnaut's and tenkiya's css pages show up perfectly in all versions of IE, in Firefox, Safari Opera and all other browsers. However, with respect, we all know about the numerous Web sites in the real world where the sidebar is beneath the text instead of at the side in IE (A common fault of blogger.com templates of the old kind) or vv, th center text is below the sidebars in Firefox, and we also know about designs that get messed up for odd reasons when something is changed. It has to work for dopes like us in the real world, not just for geniuses.

6- Schlock CSS can waste as much code as tables - look at blogger.com code or the code produced by certain unnamed wysiwyg editors - the ones that make a new style for every 3 words of text.

<please add your link to your signature>

[Webnauts]

Originally Posted by linnetwoods

Well excuse me! Some of us don't feel the need to rubbish anything and everything that is more than twenty minutes old and none of us should feel the need to rubbish anyone who is doing things differently to ourselves, especially when most of us do what we do to the best of our ability which must, by the law of averages, vary.

I can understand your point, but if I did not misunderstand what you met with that "20 minutes old", HTML 4.01 is celebrating its 9th birthday on the 24th of December next month.

P.S.: http://www.w3.org/TR/REC-html40/struct/tables.html

[linnetwoods]

Originally Posted by Webnauts

I can understand your point, but if I did not misunderstand what you met with that "20 minutes old", HTML 4.01 is celebrating its 9th birthday on the 24th of December next month.

P.S.: Tables in HTML documents

Hell, what seemed like 9 years 50 years ago feels like 20 minutes now... you wait and see

[Webnauts]

Another question to the thread. Redundant code increases the loading time of pages. Do you use AdWords? If yes, can happen that you will get into trouble: Inside AdWords: Landing page load time now affects keywords' Quality Scores

Just a bloody example.

[Webnauts]

Originally Posted by linnetwoods

Hell, what seemed like 9 years 50 years ago feels like 20 minutes now... you wait and see

I can't wait! That is why I upgraded my site's markup to XHTML+RDFa, which is official since the 14th of October last month:
More: RDFa Primer - RDFa which provides a set of XHTML attributes to augment visual data with machine-readable hints. Doesn't that make sense for bots?

So it looks like happend some milliseconds ago. Correct?

[weegillis]

Originally Posted by Webnauts

...Is a web page a presentation of tabular data?

Isn't the document tree an array?

For bots, as with screen readers, how much sense a page makes is partly derived from its linearity. Does one idea or point flow logically into another? Is the relevant link adjacent to associated text, and so on? Crashing into a td closetag is about as good a way to interrupt the flow as there is. Where is the associated 'next' item in the chain supposed to be? Right? Left? Above? Below? Data tables can be given th tags with scope to aid in this process, something that's impractical to do in a layout table.

On the strength of layout consistency, tables have earned their place, but few developers have spent time and energy giving their layouts the linearity that would make their flow comparable to a CSS only layout. This is, someone pointed out earlier, a flaw on the part of the devs, not the browsers or specifications. The spec recommends, it doesn't decree.

All hail the mighty CSS 3!

[Orion]

Originally Posted by linnetwoods

Well excuse me! Some of us don't feel the need to rubbish anything and everything that is more than twenty minutes old and none of us should feel the need to rubbish anyone who is doing things differently to ourselves, especially when most of us do what we do to the best of our ability which must, by the law of averages, vary.

I don't think it's an issue of doing your best etc. It's more kind of shocking as tableless layout has been around since xhtml / css came out 9 years ago.. tables were only really used in the first 3 or 4 years (96 to 2000) of the Internet.

We were teaching tableless design in web design courses in 2001.

That AND the fact that designing with tables is a LOT more work.

[Orion]

Originally Posted by Webnauts

I can understand your point, but if I did not misunderstand what you met with that "20 minutes old", HTML 4.01 is celebrating its 9th birthday on the 24th of December next month.

Um html 4.01 was released in 1998 lol... xhtml came out in 2000

[Webnauts]

Originally Posted by Orion

Um html 4.01 was released in 1998 lol

Not on the 24th of December 1999? HTML 4.01 Specification

[Webnauts]

Originally Posted by weegillis

Isn't the document tree an array?

For bots, as with screen readers, how much sense a page makes is partly derived from its linearity. Does one idea or point flow logically into another? Is the relevant link adjacent to associated text, and so on? Crashing into a td closetag is about as good a way to interrupt the flow as there is. Where is the associated 'next' item in the chain supposed to be? Right? Left? Above? Below? Data tables can be given th tags with scope to aid in this process, something that's impractical to do in a layout table.

On the strength of layout consistency, tables have earned their place, but few developers have spent time and energy giving their layouts the linearity that would make their flow comparable to a CSS only layout. This is, someone pointed out earlier, a flaw on the part of the devs, not the browsers or specifications. The spec recommends, it doesn't decree.

All hail the mighty CSS 3!

My site is already CSS3. I use elements that are browser compatible, even if CSS 3 is still not officially recommended by the W3C yet.

[Orion]

ah yep.. right.. xhtml still came out back to back with it though...

[morestar]

they can't have probems when crawling tables nested 40 deep, what does it do? throw a fatal error?
it can read it as deep as it wants to...

with that said, most search engines intend on providing relevant content to the searcher and if they were to maybe 'neglect' certain pages/sites because they had content in nested tables, they could be missing out on good, friendly and useful content...

and we all know good, friendly and useful content can be found in any level of nesting...

maybe the se's like or would prefer us all to use divs but that might boil down to resources...

[Dubbya]

Well morestar, you're certainly welcome to use nested tables 40 deep if you want to.

In fact, please do. Be sure to let us know how that works out for you in the long run.

[Webnauts]

Originally Posted by Dubbya

Well morestar, you're certainly welcome to use nested tables 40 deep if you want to.

In fact, please do. Be sure to let us know how that works out for you in the long run.

I am very sure that Google couple years ago did not crawl nested tables deeper that 3. And I did not ever hear if that changed or not. Just FYI.

I think a test would be worth. But I would not do that, since I would only use tables for the semantical presentation of tabular data, so I can't go wrong.

[morestar]

well i'm going to do just that, test it...

i'll have to use a pretty un-important and non-competitive keyword but...i'll rank number 1.

i just can't see how google would miss out on listing relavent content just because it is 40 deep and because the webmaster doesn't know any better...

the user doesn't see it's 40 cells deep, so why would google?


i will get back to you with my results...

[Terry Van Horne]

Originally Posted by morestar

well i'm going to do just that, test it....

... and you'll find they do! The discussion isn't will it place... or index the 40th table... but whether Google indexes past 3 deep! Just quote something in the fourth table. I know they do because I've done it and do it all the time because I build all pages in included files to separate design from programming. It bloats the code a little but makes content management easier so.. I'll give up a little of something I know is weighted low or perhaps not at all.

[Dubbya]

Originally Posted by morestar

i just can't see how google would miss out on listing relavent content just because it is 40 deep and because the webmaster doesn't know any better...

the user doesn't see it's 40 cells deep, so why would google?.

Yeah, just make sure you close all the tags or nobody will see anything.

good luck with that.

[Webnauts]

Originally Posted by Terry Van Horne

... and you'll find they do! The discussion isn't will it place... or index the 40th table... but whether Google indexes past 3 deep!

To avoid any misunderstandings, I mentioned above:

"I am very sure that Google couple years ago did not crawl nested tables deeper that 3."

I repeat that is was so a couple years ago and I am not aware of any changes if any.
The problem was not that Google could not crawl them, but they did not do so.

Off-topic:

1. http://www.hotdesign.com/seybold/index.html

2. http://www.evolt.org/article/Tables_...yout/25/21429/

3. http://www.davespicks.com/essays/notables.html

[Webnauts]

I do not always agree with stuff Rand Fishkin says, but here I do:

Keep text together - Many folks in SEO recommend using CSS rather than table layouts in order to keep the text flow of the document together and prevent the breaking up of text via coding. This can also be achieved with tables - simply make sure that text sections (content, ads, navigation, etc.) flow together inside a single table or row and don't have too many "nested" tables that make for broken sentences and paragraphs.

Source: SEOmoz | Beginner's Guide to SEO - Single Page Version

Is he wrong too?

[Webnauts]

Another thing I missed here. Google in their Webmaster guidelines say:

"Make sure your web server supports the If-Modified-Since HTTP header. This feature allows your web server to tell Google whether your content has changed since we last crawled your site. Supporting this feature saves you bandwidth and overhead."

Do you really believe the last sentence "Supporting this feature saves you bandwidth and overhead" was a tip only for us?I hope you don't believe so.
What does that have to do with the topic here?Setup two web sites, one with tables and one without. Which site pages will weight more and which less?Do you understand where I am getting to? I want to assume that you do...

[morestar]

Originally Posted by Dubbya

Yeah, just make sure you close all the tags or nobody will see anything.

good luck with that.

with all respect you're kidding me right? even if you didn't close the tags you would still get indexed...

[morestar]

Originally Posted by Webnauts

...Setup two web sites, one with tables and one without. Which site pages will weight more and which less? Do you understand where I am getting to? I want to assume that you do...

Oh i do but i've discovered that if i was google (you know what i mean) i would want to push relevant content and that's that! if i had to re-write my bots to 'be able to read 40 tables deep' in order to get that than I would!

sure div'd content will be more clear to the bots and easier to gather what content belongs with that links etc but there are still ways to table code and do the same...

so in response to your statement kind sir, we would also have to take into consideration the number of links to each of these sites. we all know google takes links into consideration and I would bet a buck or two that search engines worry more about links than they do table depth.

relevant content

[Webnauts]

Originally Posted by morestar

Oh i do but i've discovered that if i was google (you know what i mean) i would want to push relevant content and that's that! if i had to re-write my bots to 'be able to read 40 tables deep' in order to get that than I would!

sure div'd content will be more clear to the bots and easier to gather what content belongs with that links etc but there are still ways to table code and do the same...

so in response to your statement kind sir, we would also have to take into consideration the number of links to each of these sites. we all know google takes links into consideration and I would bet a buck or two that search engines worry more about links than they do table depth.

relevant content

Kind sir, are you saying here that if you have broken nested tables which can choke a crawler, could help you to overcome the problem with a certain quantity of links?

Links weight is measured upon quality and not quantity. And you are providing here information how to deceive the PageRank algorithm to overcome your onsite mess?

At last, are you probably a link builder?

[dozzn]

wow this is a long thread... and I must confess that I haven't read every reply.

But, I use CSS and tables in my webpages. the css covers most all of the fonts and link styles and some of the special effects.... but the tables have their uses as well. it's not really a competition as to which is better, css or tables, instead it just depends on what's best for your own applications...

[morestar]

well i've created a sub domain containing the two keywords that i'll be competing for...i'll get back to you with the results of my 40 deep nested table test ; )

the results i'll be looking for is at least the 3rd spot on google...after about two weeks...

[full house]

yah! rather use div . . .

[morestar]

ok forget about getting #1 but at the very least the page has been found and spidered. if you do a search (google.ca) for cow magnets im sure your'e find something from the domain morestar.ca


for some reason i can't view the (cow magnets) text within the 40 deep nested table in FF or IE but can in Safari and Crome, so use either Crome or Safari if you want to see the text...i wasn't going to spend too much time on this little experiment but i do know google crawls, ranks, visits etc pages with text inside tables forty deep...

[morestar]

actually i did a little search on google.ca for the cow magnets and apparently its number 4.

40 deep

[Terry Van Horne]

Originally Posted by Webnauts

To avoid any misunderstandings, I mentioned above:

"I am very sure that Google couple years ago did not crawl nested tables deeper that 3."

And I was absolutely sure your memory was failing you. I could believe some of the older engines like AV, IS and Excite would have these sorts of limitations because they had issues with depth of pages they crawled, oddly, it was 3 pages deep and also why a lot of people put everything in the root, absolutely... the stupidest strategy ever because it was based upon a misunderstanding of how to fix the problem and how crawlers work. It actually did more harm then good later when having those folder names in the URL was picked up by Google.

No AG Generation (After Google) crawler has any issues with things like number of tables or indexing past 3 pages down the link hierarchy. Those are myths and if someone is basing strategy on it... run don't walk, they don't have a clue about what they are takling about.

[morestar]

which brings me to my point which is that it's in google's or any search engine's best interest to crawl content even 400 tables deep, as long as they can find relevant content for the user within those nested tables...

[Webnauts]

Ensure that content is available in reading order

Having discovered and arrived at your readable content, a user needs to be able to follow the content you've put together in its logical reading order. If you are using a complex, multi-column layout for most of the content on your site, you might wish to step back and analyze how you are achieving the desired effect. For example, using deeply-nested HTML tables makes it difficult to link together related pieces of text in a logical manner.

More: Official Google Webmaster Central Blog: Webmaster tips for creating accessible, crawlable sites

[weegillis]

It comes down to linearity. Any table can be crafted to be linear, in a fashion. It's moot that a structure can be built to the depth of 40 levels and the proof is there that a spider can drill down that far. As an academic study this whole supposition is laudable. Heck, we all learned that Google will give a 4 spot to a key phrase nested 40 levels deep!

Google, Yahoo! and Live, et al, spell out what they want to see: web content that is intended for an audience. Humans are expected to make as much sense of what they see as spiders (that have hundreds of thousands of CPUs at their disposal).

Any table that is linear enough that a disabled and brain dead spider can find its way from one point to the next meets the criteria of the search engine guidelines. If the developer has enough salt to do this, it should not matter whether she chooses to go tables or no tables or hybrid. The content is making its way to the audience.

Search engines are designed to get the right information to the right audience, right? The loudest is no longer the one that's chosen anymore, and I for one couldn't be happier for it.

If I owned Google, I'd be set on stamping out the lot of keyword diatribe that is out there right now. It's sometimes hard to find a real site for some search phrases. This is plain disgusting.

I probably don't belong in this forum because I am against half of what I've seen here. When fortunes are made and lost at the whim of a search engine I cringe that folks would risk somuch on such a whim. In the prairies we have the weather to take the place of Google, in this sense. One bad year, oh oh; two, forget it.

[morestar]

Originally Posted by opalcomp

That is pretty silly. The table text would be slightly larger, but not bloated the way you wrote it. Here is my version:

<table>
<tr><td><h3>This is a subject Heading</h3></td></tr>
<tr><td>Here is your first bit of text.</td></tr>
<tr><td>Here is your second bit of text.</td></tr>
</table>

Sometimes CSS users sound like they are preaching a religion. CSS is better in a number of ways, but tables are not always so bad, and should be used in data presentation, at the least.

that's funny cause this is the way I would have wrote it - even less bloated than yours.

<table>
<tr><td>
<h3>This is a subject Heading</h3>
Here is your first bit of text.<br />
Here is your second bit of text.
</td></tr>
</table>

no big deal

[morestar]

oh, why i debloated my code was because the comparison in the code in question contained two paragraphs (<p>text</p>) within one div.

[full house]

if its pure table I think google won't like it! its not user friendly. Better use xhtml and css.

[weegillis]

Originally Posted by full house

if its pure table I think google won't like it! its not user friendly. Better use xhtml and css.

From a tweaking standpoint, you may be right. Tables are a little bit clumsy when poorly linearized. However, when indexable content is present, they (SEs) find it, nonetheless.

This whole table thing is really getting tiresome. Table based sites can be made just as accessible as flow and position based. To a spider it boils down to ease of parsing and connecting information. Table related tags are easy to strip, as are all other html tags in the pre-DOM source document. Whenever text falls in the natural parsing order after tags are removed (allowing for scope and th values) it meets the requirements for logical dissemination. Discussions ought lean toward how to accomplish this, rather than argue that CSS methods are better.