Alignment as an option? Yes, please. Alignment as an option, that if entirely cut out, does not break a class/require lots of houseruling to make it work/negates one or more meaningful class features? Hell yes.
Interesting direction to take the detect and protection from series of spells. At the least it sounds like that DM's will not have to rely on contrived spells and magic items--such as undetectable alignment and a
ring of nondetection--to prevent the characters from just randomly
locating the villain.
Even the paladin, its early incarnations famous for adherenece to both the Lawful Good alignment and a code of conduct, is going to be built to focus on what alignment represents. The example of a Chaotic Good paladin translating into a paladin of freedom reminds me of 3rd Edition's variant paladins from Unearthed Arcana (the paladins of freedom, slaughter, and tyranny). As before when Mearls mentioned paladins having alignment-based class features, this sounds like a solid, interesting way to differentiate them from clerics.
Finally, Lawful monks will likely end up being an option that DM's can invoke. In Mearls's own words, alignment will be in the default rules, but it will not be the rule.
Their campaign to keep spellcasters from easily taking down an enemy involves--so far at the least--two things. The first is requiring a spellcaster to maintain concentration, preventing a single spellcaster from just piling on debuffs (as well as buffs, meaning that we likely will not see spellcasters transforming into better fighters than an actual fighter).
The second part requires making damage spells actually useful--and maintaining this usefulness--and having hit point-ignoring spells allow multiple saves before they take effect. I have mentioned direct-damage spells multiple times, and that they rapidly become useless without some form of scaling. Thankfully, according to Mearls, this is being addressed.
As for save-or-dies, I really dig the idea of a wizard having to actually work to maintain a spell in order for it to have its full effect. 4th Edition was an excellent start in diminishing the anti-climax of save-or-die spells, but once they kicked in generally a wizard could keep heaping on other spells while the monster tried to shrug it off (occasionally having to funnel Minor actions to keep it going).
Having to concentrate and keep weaving magic on a round-by-round basis sounds awesome, and if you couple this with the bad guys being able to disrupt it could make for some tense situations. Of course, I may be biased because this is similar to a spellcasting mechanic I thought of quite awhile ago, where a wizard would have to take time gathering energy to cast more powerful spells.
As with wizards being able to cast more like clerics, this sounds like another big step in the right direction. The only bad part about this article is where Mearls closes with the statement that DMs can just opt to not use save-or-die monsters, and that they should be confident that players will build characters without making things wildly unbalanced.
Personally I think it would be a small feat to include variant attacks that allow multiple saves, and balance character options--ie, spells--to the point where it would be very hard to make unbalanced characters.