Links: Underline vs Colour change

Here are some  areas to consider:

Issues with using colour only to denote a link would be 1 accessibility (cognition that a link is in fact a link, and also colour blindness or macular disorders require sufficient visual difference between a link and non-link) as well as 2 deviating from the commonly understood pragmatics of the underline denoting a link.

For hoverstates, consider tablets and mobiles. How would users recognise a link with no hoverstate?

Who is / are the target users and what is the site for? E.g. A banking site Vs a music site coul call for different treatments to create an approproate experience. 

If you use colour to denote a 'mechanical' link function, how will you create mood / reflect emotional brand values with typography?

To unpack more, it's worth reiterating that colour blindness (or colour definiency) sits as one of some combined small issues which, In ensemble, can create an experience of some abiguity. This points back to user base which can affect level of accessibility you decide to go for. As a case study, this site doesn't, in all likelihood, have a user base who are cognitively impaired or who have poor online literacy so the interface language can be subtle.

You can download plugins to simulate the effects of colour blindness - protanope, deuteranope and tritanope. As Mike points out, tritanope is severe so you get greys, blues yellows and pinks. The others are less severe in terms of colour shift, but in between there are different gradations of severity. Colour blindness is mainly a man thing (recessive chromosome) and fairly uncommon.

Cognitive impairment and low reading age are some if the forgotten accessibility issues, as is interface literacy. So how you use colour / underlines is a context / audience thing again .

Lots of blah on accessibility!

Brand, as I metioned before can affect what you might choose to do. This site uses two colours, so using a colour to differentiate links from copy (and using pictos/icons to enhance meaning) is a simple task. If the site used a load of colours to push / follow brand identity requirements, differentiating links from non-links is trickier.

I've been involved in usability testing sites for years.  Going back to the initial question, I've seen issues with links that are both too bold and too subtle.

We're now at the point where a traditiional blue underlined link, while technically proper, looks old.  I've had some test subjects call me out on that.  On the other side, I've also seen regular-sighted users have problems identifying links because the color choices and/or rollovers were too subtle.  

This is somewhat anecdotal, but these are my four key takeaways I've had from seeing users react to links in so many tests:

• A change in shade will go a long way, and avoids issues of detecting a change in color.   My trick is to squint my eyes and try to detect any change on rollover or click.  If I can see the change, most color-blind users can as well.
• Try and keep some blue in your links.   Regardless of shade or underline, I think we all look for links by scanning for some form of blue.
• Make a link too big, or too graphically treated, and users will scan right past it.   They're generally expecting something roughly the same size as the text around it.
• Use link styling as an opportunity to colaborate with your art director.  There are a lot of solid creative solutions such as the above.

I agree with Chassy's comment, "Make a link too big, or too graphically treated, and users will scan right past it.   They're generally expecting something roughly the same size as the text around it."

The main reason people tend to ignore links that are too graphically treated (large call to actions), is because when the user scans the page they often asscoaite large graphical links as an advert, and therefore ignore it.