Since numerous search results pages are already in the index then Yes, you want to use the NoIndex tag instead of a disallow. The NoIndex tag will slowly lead to the pages being removed from the SERPs and the cache.

Google Webmaster Guidelines suggests you should "Use robots.txt to prevent crawling of search results pages or other auto-generated pages that don't add much value for users coming from search engines."

After reading the further responses here I'm wondering something...

You switched to a new site, can't 301 the old pages, and have no control over the old domain... So why are you worried about pages 404ing on an unused site you don't control anymore?

Maybe I'm missing something here or not reading it right. Who does control the old domain then? Is the old domain just completely gone? Because if so, why would it matter that Google is crawling non-existent pages on a dead site and returning 404s and 500s? Why would that necessarily affect the new site?

Or is it the same site but you switched to Java from PHP? If so, wouldn't your CMS have a way of redirecting the old pages that are technically still part of your site to the newer relevant pages on the site?

I feel like I'm missing pertinent info that might make this easier to digest and offer up help.

Have you submitted a new sitemap to Webmaster Tools? Also, you could consider 301 redirecting the pages to relevant new pages to capitalize on any link equity or ranking power they may have had before. Otherwise Google should eventually stop crawling them because they are 404. I've had a touch of success getting them to stop crawling quicker (or at least it seems quicker) by changing some 404s to 410s.

In some cases, the capitalized version and lower case version of a search won't return the EXACT same results in the SERPs (sometimes a result will shuffle a bit [though not by much from what I've seen]) but Google does understand that "Example" and "example" are the same word much in the same way that it knows a singular and plural of the same word are related, i.e. "Shoes" will show up for a search of "Shoe" and vice versa.

If you also take into account the regular shuffling of the SERPs every time Google so much as blinks, there is the possibility that when the rankings data was pulled for your terms they were done so in a fashion that shows two different numbers that were both correct at the times they were pulled. Plus, in the past week (week and a half?) there was a decent bit of volatility in the SERPs that lead people to believe there was an algorithm change happening so for all you know that may have put an odd hiccup in the numbers as well.

Worse comes to worst you could always do a bit of testing to see if maybe you stumbled on a previously unknown traffic/rankings difference. Shouldn't be but with 500+ algorithm changes a year, you never know.

I've found that people like bullet points... gives them quick information without needing to read the rest of your content. Of course the problem arises of duplicate content when you use the same bullets over a large amount of products or all you do is change a word or two between every page. If the bullets are all the same then what's the value of them to the user? Bullet points (like all content) should be relevant and fresh. They should highlight the specific advantages of that product, i.e. what makes it special/different/unique/important/worth buying?