For starters: This isn't going to be a scientific comparison review. If you want to follow the debate from a technical standpoint (whether or not scaling down the 24mp files produces the same/better results, the true impact of the loss of light from the translucent mirror, etc), then you'd be better off looking elsewhere. This is simply my real-world experience with both cameras, with non-scientific photo comparisons galore.
When I first decided to switch out of the Pentax family early this year, my main focus was on Sony. I had been an early adopter of the NEX-5, though I couldn't justify keeping it when finances got tight. The NEX-7 had eluded the markets in Fall 2011 due to the floods in Thailand, but rumors persisted it would be released soon. Still, I wasn't sure I could use the NEX-7 as a full-time camera for professional shoots, as its size and layout is much more suited for a travel/street/journalism shooter than a studio/portraits/weddings sort of shooter. As I did more research on the matter, I realized that the A77 offered the same sensor in an SLT configuration, of which many Sony-proponents assured the translucent mirror did not degrade the image quality more than 1/3rd of a stop. Better yet, you could get the same sensor in the A65, which mainly got rid of the A77's weather sealing and battery grip capabilities (which I figured I could live without). Long story short: I took the plunge in a masterfully-conceived plan to incrementally buy/sell my gear over the course of a month without impacting any scheduled shoots, and to stay within the confines of my measly credit limit. And almost immediately, I began to regret my choice.
Above: Photographs showing before/after shot from Pentax K-5 (using Sony's 16mp sensor) at ISO 1600, extreme adjustments made to exposure/shadows/blacks. Shadows retain detail without becoming too muddy, even with noise reduction.
Don't get me wrong--I thoroughly enjoy both the Alpha and NEX lines, and I plan on sticking with Sony for quite some time. But the A65's capabilities were simply not up to par, especially when compared to the Pentax K-5 I was switching from. To keep non-Pentaxians in the loop: The K-5 has the same(ish) 16mp Sony-made sensor as the Nikon D7000, Sony NEX-5n/c3, A55/57, A580, so on and so forth. I say "ish" because each of these sensors are calibrated slightly differently, yielding different performance scores (with the 5n/57 possibly being a newer generation...debates go back and forth on this topic all the time). When I bought the K-5 at launch, it was basically the best-performing APS-C sensor camera on the market, and it certainly didn't let me down in that regards. So why did I switch out of the Pentax family? One word: backfocusing.
The K-5's phase-detect autofocus sensor was (and still is) infamous for front/backfocusing under different lighting conditions. If I calibrated a lens to focus perfectly while shooting outside, I could get constant and considerable backfocusing while shooting in the studio later that night, or at a bar the next day, or an event the next. It got so bad that, for the last few events I shot on the K-5, I actually used the live-view function 80% of the time, simply to take advantage of the spot-on contrast-detect autofocus that it offered. That why, when switching out, I decided that focus performance was my top priority. And in this regard, the Sony A65 certainly did not let me down. Though naysayers mock Sony's SLT technology (and sometimes rightfully so), the EVF on the A65 combined with a fantastic focus mechanism led to spot-on focusing and snappy performance in nearly every situation. Beyond that, however, the A65 is a very flawed camera in many ways which are often overlooked in most reviews.
I purchased the A65 in February 2012, and in the course of two months, I shot a fair amount of paid work on the camera. While it worked great under some circumstances, my experiences shooting ISO 1600+ at a friend's wedding a week or so ago ended up being the final straw. It wasn't the first time I'd shot that high under low-light settings, nor did I have any particularly important use for the photographs (since I was just there shooting for fun). But my frustration at trying to get a usable image from a RAW file that theoretically would be easy as pie to edit (had it come from my K-5) was simply infuriating. Beyond that, I had gotten pretty fed up with the lag that the A65 produces when trying to change simple setting like aperture and exposure. An update in March did help speed things up a bit, but it still felt like you were trying to operate the camera in a zero-gravity environment or something. The interface was ever-so-slightly sluggish enough to make everything feel bogged down and slow, as opposed to the instantaneous dial feedback I was accustomed to from every single DSLR I'd owned up to that point. A couple days later, I found a listing on KEH for an A580 kit in like-new condition for $695 (on a free-shipping weekend too, to make things even better). I was sold.
I've only had the A580 for roughly a day, but within an hour or so, I knew it would be a keeper. Physically speaking, it's actually somewhat larger than the K65 (and much larger than the K-5). It's body is still made out of plastic, though it feels slightly heftier and more substantial in your hand than the A65. The buttons on it are snappier, with some producing more of a sharp "click" sensation than others (as opposed to the A65's array of mostly "muted" button). And the flip-out screen, though it can't be flipped in on itself nor flipped around for self-portraits, feels much sturdier (with fewer frustrations when mounted on a tripod). The OVF on the A580 is actually much smaller than the EVF on the A65, which is comparatively large (and always bright) thanks to its inherent nature. In all seriousness, the photo to the left doesn't do it justice. Still, I prefer the layout of having the old-school status/indicators along the bottom of the OVF on the A580, than having to look around the screen on the EVF to reference what settings I'm on (which can get overwhelming, though it's somewhat adjustable). I also got accustomed to being able to review my images in the EVF on the A65, though playback on the A580 via its LCD screen seems to be much more responsive than the A65 in general. In terms of "professional" features, it's also worth mentioning that the battery life on the A580 is rated to be MUCH higher than that of the A65, with 1050 shots compared to 510 (both figures from Sony, measured in "viewfinder mode"). The A580 also supports adding a battery grip, which would effectively double those figures (I have a cheap 3rd party grip on its way, I'll review it once it gets here).
Going back to the question of RAW performance, the "usability" of RAW files shot at ISO 1600 and above is simply atrocious in comparison to Sony's 16mp sensor. In regards to the K-5's noise performance at higher ISOs, I once read a comment stating that even though the K-5's RAW files were somewhat noisy at ISO 3200/6400, the noise patterns were incredibly "usable," and had "so much information" packed into the images that one could easily tweak a few settings and come out with a gorgeous image, even under the worst circumstances. Though I'm not sure if there's a better, less metaphysical way to put this criteria into technical terms, it's true. The images produced by the K-5 (and similarly the A580, as I've found) are extremely forgiving. Tweak a few sharpness/noise reduction settings in Lightroom, and you can easily use ISO 3200 in many commercial shoots. The K-5/A580's RAW files are also very forgiving when changing exposure/contrast/shadow levels at high ISOs, so much so that when I first tried to edit an ISO 3200 file from my A65, I was half convinced I accidentally shot it as a jpeg. If you've never had a camera with Sony's 16mp sensor in it before, and you're accustomed to a more "standard" criteria of high ISO performance, then the upgrade to an A65 might not be as much of a shock. Heck, in comparison to the 12.4mp Pentax K-x I owned before the K-5, I would have considered the A65 a slight upgrade at the least. Still, when it comes to real-world editing, the A65's high ISO performance leaves much to be desired.
Here's a quick image I took, both at the same aperture/shutter speed. White balance set to the same in post. A65 is on the left, A580 is on the right. Exposure/contrast/clarity/shadows all raised equal amounts. Noise reduction/sharpness were set independently on both, with the goal of getting them as clean as possible without becoming muddy. In retrospect, I could have actually done a lot more noise reduction on the A580's image while maintaining detail. It seems that the main difference here is the size of the noise, and color shift in the shadows. The A65 has a serious problem with purple-hued shadows, which is made worse by the muddiness of its noise. While I have no technical way of proving it, the A580's sensor simply produces "finer grain" noise, which Adobe's noise-reduction algorithms have a much easier time dealing with.
Above: Photograph shot with the A65 at ISO 800. Unlike the Pentax K-5 night shots from above (which were shot at ISO 1600), I had limited control adjusting the shadows before muddy noise took over. This was the best result I could muster.
So what IS the A65 good for? A lot of things, actually. The 24mp sensor, while noisy at higher ISOs, nevertheless produces gorgeous results in daylight/studio circumstances. With that much resolution at your fingertips, you gain incredible cropping power. There were times when I shot a photograph at landscape orientation that, once I got to post, I could easily crop into portrait and still end up with an incredible amount of detail and resolution. Additionally, for circumstances like daylight portrait and studio shooting, the focus accuracy of the A65 was hard to beat (at least in my experiences). Paired with the Tamron 17-50mm f2.8 lens, the vast majority of photos I took in any given shoot were spot-on in terms of focus, to the point where I didn't even know this lens could physically be this sharp. For whatever reason, the focus speed seems to be improved as well. My Tamron 70-200mm f2.8 beast of a lens, known to be a slow focuser, actually performs beautifully on the A65. I've even gotten remarks from fellow shooters of the focus speed and general lack of hunting. In some ways, shooting with the 70-200 on the A580 seems to be a downgrade of sorts, as I am now experiencing the gist of the "slow focus" complaints that reviews of that lens usually garner.
Above: Crop (not 100%) of a studio shot taken with the A65. While there's plenty of detail, some of it is still ended up a bit muddy after adjustments left the shadows too noisy,even at ISO 100.
So my advice is this: If the vast majority of what you shoot doesn't involve low-light circumstances, and you need the boost in video features and focus accuracy, go for the A65. Or better yet, you might want to have a look at the A57, which packs Sony's 16mp sensor into the A65's body (sans GPS and a few other features), albeit with that 30% light decrease from the SLT mirror. Otherwise, consider grabbing an A580 while you can, as it will most likely be Sony's last OVF offering.
Are my methods for comparison perfect? No. Far from it. One could easily argue that this entire process was more of a poor critique of my shooting/editing styles, or that I simply expect too much editing potential from RAW files. I don't hate the A65 by any means, and I still believe it is a wonderful camera. But it just doesn't suit my purposes, in regards to both my workflow and the editing legroom I'm accustomed to. To each his/her own, I guess. As for me, while I'm planning on keeping the A580 for work purposes, I'm still lusting over the NEX-7. Despite my complaints about its sensor (though noise performance does seem to be improved with the lack of translucent mirror), I just love the form factor it provides. Give me a NEX-7 with the new Sigma 30mm f2.8 lens, and I'll be a happy camper...at least until the next best thing comes along.