Actually it's the bone-dry carpet that will generate static. A simple way to negate that risk is to lightly mist the carpet before you start. When I lived in the desert (where "dry" means "spit and it doesn't hit the ground") I had to use a weak solution of Downy to get rid of carpet static; this treatment was good for about 6 months.
That said, most components are not as sensitive as we've all been told. The worst are the pinless CPUs -- fer ghu's sakes,
don't touch the contacts! Conversely, you wouldn't believe the abuse RAM puts up with. -- I've been building the beasts for 20 years now and have never zapped a component, despite that I don't take any special precautions other than to work on a wooden surface (and I often have a naked motherboard lying on the table, hooked up and running... in fact one machine lived that way for quite some time before I got around to putting it into a case!) I carry a chunk of cardboard in my house-call kit for the same purpose. Plastic/Formica can be amazingly staticky. Incidentally, so can some cardboard, especially the shiny-surfaced stuff-- check it by doing the carpet shuffle and see if you zap yourself when you touch it. If you're paranoid, touch something conductive just before you handle any component.
As to PSU weight vs quality, sorry, this still holds. This article is old but look at the photo of heavy vs light innards (the one on the left is an Enermax; the one on the right is probably a Powerman) and the reason becomes obvious:
http://www.directron.com/psu.html
http://www.directron.com/table1.html
If you want stable power handling, capacitor size matters. It's no coincidence that cheap PSU with dinky little capacitors correlates very well with premature motherboard death (which otherwise almost never happens).
As to efficiency, last time I saw a thorough comparison (about 3 years ago), the range was around 75% to 87%, and Enermax were still on top. And I've never seen a dead Enermax PSU. I've seen lots of dead lightweights. (I always strip deaders for fans and screws, so I've noticed what all is inside 'em.)
The other point with PSU is look for lots of connectors and heavy-duty wires. That tells you what it will actually support, vs the wattage on the label. (And this will go right along with weight.)
Three years ago I had to replace a TOPower PSU (the old one had been running at max capacity 24/7/365 for ~15 years, finally died of a nearby lightning strike that fried right through two layers of surge protection -- but didn't get past the PSU) and I bought another TOPower because it was the only AT type PSU I could find (yes, it's an old machine!) ... expecting that quality had sadly declined as have so many things... nope; it's
identical to the old one. But they sell mostly to the server market, and that will always be better quality than stuff made for the consumer market. (This is also why they still make AT PSUs -- there are still some very old irreplaceable servers out there.)
I would not go with an I3 in a new desktop (unless it's replacing something as old as my AT beast, then it's an upgrade
) However I'm not sure the difference between I5 and I7 is significant unless you're a high-end gamer or do video rendering and the like.
As someone says above, overclocking is a lot of the issue some folks have... but that said, technically there is no such thing as an "overclocked" CPU:
CPUs are made in one speed and type at a time. But when you stuff umpty=zillion transistors in a part the size of your fingernail, not all of them turn out "good", and the more of 'em don't work, the slower the individual CPU's functional speed. So CPUs are batch tested (typically about 0.1% of each manufacturing batch) and whatever is the lowest speed it batch-tests at is what the entire batch is labeled. But since
some of each batch will be more functional than others, those CPUs can be "overclocked" by setting them to whatever is the
actual performance ability of that individual unit. (I've also seen a few that worked fine when set slower than they were labeled, but wouldn't run at all at their supposed speed... luck of the draw in being worse than the batch test indicated.)
Intel has sometimes locked the entire batch at the tested speed, which means those CPUs cannot be overclocked. (This probably saves a lot of grief in the long run.) But AMD made hay with having particularly unreliable manufacturing, in that speeds per batch were all over the place, so a lot of their CPUs could be spectacularly "overclocked"... and the more juice you put through a given CPU, the hotter it runs.
AMD also has historically had an errata sheet (bugs and defects found after manufacturing) 3 times as long as Intel's. If you haunt "Windows problems" forums that require posting of system specs, you'll discover that AMD users with stability issues typically outnumber Intel users 10 to 1, despite AMD having a fraction as much of the market. AMD has not been good about owning up to and replacing CPUs for fatal bugs, either. (Intel generally will replace such CPUs, even out of warranty.)
