Sorry, can't leave this one alone.
Sorry, I disagree (again).
If you were to do the math, you would see that choosing the right caps and applying the proper aging functions based on the ratio of operating temperature to design temperature it would be obvious (to somebody who actually understands how this works) that the base lifespan gets multiplied by the reduced aging loss coefficient. Even an "average" cap can last 10k hours under properly applied condition. A better than average cap will in fact likely last 20k hours with properly applied aging factors. That's (obviously) 1000 hours a year for 20 years, or about 3 hours EVERY DAY.
Not sure if I've been impressed with intelligence or baffled with B.S. here

. Of course you'd have to know the engineering specs on the caps that are in your amp, and you'd have to know that the caps were actually meeting their specifications, which without a professional parts qualification team, you don't. All you have is faith in the team that designed and built your amp and the reputation of the parts manufacturers. Figures don't lie, but liars figure. Even in the case you state, your assuming 3 hours a day? Well, maybe for a practice amp, but my Linux workstation here is on 24x7, so should I expect the caps to die in 2 1/2 years? That's still about the most unreliable part in there, except maybe the fans. Anything with fans is a failure waiting to happen. They don't last.
In the case of our B20, there is very little information on the caps. Apparently no batch number, no date code, even the name on them seems to lead back to a distributor that buys whatever cheap parts they can get. They have a number, CD110, who knows what that is, and a temperature range, -40 to +85C, but without a spec, or certification from a reputable lab, that is worth the cost of the ink it's printed with and that's about it. Unfortunately, this is probably typical of the quality of the parts that go into gear that is built to compete on price above all.
Replacing caps every few years is simply absurd, a waste of time, and is likely to damage PCBs within the first couple of attempts. There is no evidence whatsoever that this is beneficial in any meaningful way.
Did I say I would replace the caps every few years? Did anybody else? I believe I stated that if an amp was a few years old, I was in there anyway, and the caps that were in there were not top quality parts, I'd change them. I stand by that. After replacing the failure prone crap with good stuff, I wouldn't expect to be back in there for another 10 years if I only used the amp occasionally. Our experience with the B20 makes a pretty good case for this. Generally, I let reality be my guide.
If someone's soldering skills are so bad that they can't replace a part once or twice without trashing the circuit board, either they are a newbie that didn't heed my advice to practice up on junk first, or they are a klutz that should be forever banned from handling tools. Soldering really is an art, a matter of pride for those who master it, any good tech. A competent tech with a quality soldering iron can replace a part a number of times without trashing the PCB. I don't think I have lifted a trace since about, maybe, 1973?
Increasing capacitance might be ok but it might not be as well. You are suggesting that folks who have no idea what they are doing second guess the engineer that has both the education and knowledge to make this decision on the product. Many protection systems are designed around the amount of energy stored in the power supply. Increasing this energy can cause the power supply to fail under a fault condition, increasing the risk of damage and possible fire, which would invalidate the safety agency listing. The liability would fall squarely on the person who made the unauthorized modification, as it rightly should.
Wow. A couple comments.
I find it intensely amusing that all this caution is coming from an employee of Mesa/Boogie. According to a piece in the book "The Tube Amp Book" by Aspen Pittman (good book by the way), Mesa/Boogie came about when Randall Smith, a technician, stripped the guts out of a Fender Princeton, and stuffed in the circuitry of a Fender Bassman, and a JBL D120F speaker. My kinda guy

. Sounds like something i would have done, if I could have afforded a D120F. Somehow, I doubt that the project was blessed by a real honest to god holier-than-though engineer, or that it passed a battery of regulatory compliance tests (which our B20 didn't either). This kind of innovation generally comes from free thinking individuals who are not getting all caught up in regulations and "what-if"s". Sure he might have gotten sued if he sold it, some dolt mis-used it, and hired a lawyer that thought Randall had enough money to be worth trying to screw out of it. There is a certain security in not being rich enough to be worth a lawyers efforts

.
I think the odds of doubling the size of the filter caps becoming a hazard are about the same as the odds of being hit by lightning. The upside is that the amp will exhibit less noise, and last significantly longer before the caps age to the point of being unable to do the job of keeping the hum level down. Odds are the engineer, if he/she were a music person, was forced to chose values lower than he/she would have preferred to begin with, for cost reasons. The engineers don't have final say in these matters, the bean counters do. Such is the reality of consumer electronics. "Build something as good as you can for $X." Thats how it goes. Granted if the power supply is extremely marginal, the additional start up current of the bigger caps could be an issue, but it's pretty unlikely. Personally, I'll take that risk, or improve the power supply if necessary.
I can't help but take issue with the engineer worship going on here. Yeah I became a Senior Member of Technical Staff at Tellabs, an engineering title not awarded to many engineers. But for a number of years of my career, I was jumping on planes and going into critical outage situations that were more often than not the result of an error on the part of engineering (granted the usual hardware failures and application errors were typically solved before I got involved). Engineers are people, and people make mistakes. People well trained in theory without a whole lot of real world experience don't always get it right. Our product got a whole lot better after I created a test group, staffed by a couple degreed engineers, and a couple techs with significant real world field experience, to beat the heck out of the product before we sent it out. That is what it took to achieve the 99.999% reliability that we eventually achieved.
Regarding DIY repairs, I have unfortunately seen the results of DIY attempts and while some may be ok, the majority I have seen are abysmal. The results are damage that often renders the product unrepairable (or not economically repairable) when the original real problem was a simple, cost effective repair when the tech correctly identifies the fault before hand.
In general, if you don't know what you are doing, leave things alone.
When you say leave things alone, you are suggesting what, tossing the amp in the trash without even trying, or paying someone more than the unit is worth to repair it? Unfortunately, that is often the reality.
Again, the majority of DIY repairs that you have seen. Most likely, you would not have seen the successes. Your sample group is biased.
What we are talking about here is an amplifier that on the whole was built pretty well, but got caught by sourcing cheap caps that were part of the plague of Asian caps made with bad electrolyte. Doo-doo can happen when the price of the product will not support top notch components and quality control. I really feel for the guys that had to take that gamble, and lost. I might have taken the same gamble, given the choice they had. It was a gamble trying to deliver a price competitive product, they lost this time.
Now when someone comes across one of these amps, they have a choice:
1. They can take it to a shop, which probably needs to charge $50 or $60 an hour to pay the staff, the rent, the heat, and keep the lights on. That is the reality of the situation. Keeping a repair shop operating costs money.
2. They can scrap it, the speaker isn't bad, I intend to test mine and model the Thiele-Small characteristics to see what else it may be good for, before I put this thing back together.
3. They can take a look inside and see if their amp looks like so many other B20's, the cheap caps failed, are bulging on the top or have blown open, started drawing excessive current, and blew the fuse. Ironically, the fact that so many of them fail in exactly the same manor is actually an indicator of good quality. They all only failed because of the same failed part. If the cap supplier hadn't failed them, most likely they'd all still be up and running. The goal in quality is consistency, even in failure.
For the price of a decent six pack they can replace the caps with a better brand, and the fuse, if they know how to solder, or are willing to learn.
I came up through the school of hard knocks. I learned to troubleshoot when my stuff didn't work. I got good when I wanted stuff that didn't exist or I couldn't afford to buy it. So I built it. It served me well as a profession, and as a bass player.
I'm near the end. My time is nearly up. I want to encourage another generation to blow off the naysayers and take on the challenge of first fixing what's in front of them, and then going a step beyond that, creating something that is new, that advances the art of communicating emotion through music. Somebody needs to do it

.
Marty