Mark has a good idea about having your main subcategories work... not all sites do this, but for example our e-commerce sites do (if it's something like domain.com/toners/p/some-particular-toner, you can get to a general list of all toners at domain.com/toners/ as well). I think what he might have meant about removing subcategories was if you have more than a handful of them, or have pages that are kind of individual and don't necessarily fit under any particular category.
Dave said try to make them as short as possible: of course this is a good thing, but it's just something to generally strive for if you can. Of course shorter would be easier for someone to remember, but we know from user testing that if it's somewhat longer, yet composed of human-readable phrases, that it's still easier to remember than either a long or short URL composed of garbage.
lavluda.com/2007/07/15/how-to-enable-mod_rewrite-in-apache22-debian/ is an example of an admittedly long URL, very typical of wordpress. His subdirectories work as well, so if I only remembered his domain name and the year (2007) I can type that in, get a list of everything he's published in 2007, and scan until I see the one about Apache2.2 and Debian. I don't think it's a bad URL at all. Would it be even easier to remember if it were shorter? Sure. But shorter doesn't work with how his WP setup does URLs (turns titles into unique page links). shrug
You asked, "So how long is too long?"
How many camels does it take to drink a lake? The answer is very subjective. Sometimes our e-commerce URLs are kind of deep because we have some product categories with 2 or 3 levels of subcategories, because that's simply how the databases from the clients are set up. Eventually the URLs are rather too long to expect someone to remember them (but with URLs there's also a benefit in recognition which most humans do much easier than recollection), and that's unfortunate, but we live with it. In fact at some point we got some clients to agree to remove some levels and make a "special" category for what really was miscellaneous stuff that simply each didn't need some subcategory. Now those are domain.com/special/ for the special things page and domain.com/special/p/product-name for each product. Still pretty okay.
Mark did think having 3 levels of subcategories is a bit too long. It's kind of long, but if users can use those category names as end-links like Lavluda's "2007" does, it's not necessarily bad.
You also aren't required in any way to mimic your back-end folder system. It might be easier, or not, but don't twist yourself up too much about rules. Also, while your subdirectories make some sense, the longer your URL gets, the less sense they'll make to more people. The links you've posted in this thread are maximum two categories deep and some are just one. That's why they look fine to me and the length of the actual page part is a semi-readable sentence, which makes it look long but is still doable for many people. At worst someone will vaguely remember your page name and type it into a search engine, where it will probably come up pretty close to the top.
Also, some of your subcategories might be redundant: for example if you have a subcategory called "accounting", having that listed as under "finance" might be kind of useless. However if some of your accounting articles are about using software tools, while others are about legal implications, then domain.com/legal/accounting/name-of-page and domain.com/tools(or software)/accounting/name-of-page might be useful (because users who navigate by category, which I do but don't know how often most normal people do, can get to all legal or all tools/software based on that URL). Meanwhile general accounting would be better off in just domain.com/accounting/name-of-article.
My above paragraph makes it start to get complicated, so don't let it bug you-- the last thing you want is a complicated URL system. It'll be a pain in your butt and a pain to maintain. But if you've got lots of chained subcategories now and are looking to shrink them, something like the above might help you figure out where to cut.
I might agree with Dave also that your physical site menu and breadcrumb structure are probably more important to average users. I mean, i use URLs to navigate sites, especially with structures like a typical WP site has... but I'm also a web developer, so I'm probably weird. I also know what the SysReq key does, while I doubt most of the visitors to our e-commerce pages have any idea. I personally find chopping off ends of URLs to find parent pages really really nice and convenient. I don't know if 99% of your visitors will though. People do however use breadcrumbs and do benefit from stuff seeming to be structured sanely when they visit the site, but they totally don't care about how your backend is structured. You care about that, because you're maintaining it. But they don't.
As Dave says in that other thread:
This sounds nice to me. I think showing a level is good, and maybe because I'm a developer I personally prefer something there instead of all pages coming directly off the domain.com part. But indeed a few of the examples in the other thread started to get real weird real fast (though you also had them as examples of what you wanted to avoid... good). Again, what you posted above, here in this thread, are awesome in my point of view. They're long in number of characters, but they feel pretty simple and they look doable to me.
You might want to do some user-testing on friends and grandmas or with your site's own analytics to see how people are dealing with them. Your sever logs will tell you how a hit got to your site (no referrer means either they typed it in by hand, had it bookmarked in their browser, or possibly are so paranoid that they've turned off all referrer headers from their browser). That will be useful information, as well as, are there many hits to domain.com/subcategory by themselves. If so, it would mean many people like going to the subcategory pages, and again if those didn't have a referrer (as in, they didn't get there by clicking a link on your site or in your menu) then that might suggest people are using your subcategories, and you might get a better idea if they're as useful as you think.