That is not my experience at all. Average Joe does not back up their data, some do most do not.
I know this because I have built/serviced/repaired computers for quite some time now and the big area that stands out is "not backup up data"
There is no debate about data recovery from SSD v HDD's because everyone knows you have a next to no chance for SSD recovery, and a fairly decent one for HDD's.

We all love the idea of SSD, but it's not ready for the big time just yet, cheaper, more reliable, bigger capacity and we might be in business.

What do you mean by "the idea of SSD"? Are you waiting for them to push out HDDs completely? That will never happen. SSDs will take over gradually as the technology matures, pushing HDDs into niche uses. This is already happening in the case of PCs. In my case I first replaced my OS HDD for an SSD. And when the SSDs got bigger I could do without a local HDD altogether.
But I'm not suggesting that stuffing your NAS with SSDs rather than HDDs is a good idea.

Dismissing SSDs with the argument that a HDD is more reliable is a little bit misleading since a HDD isn't that reliable either. For the non-power user a single HDD system might make more sense than a single SSD system, but due to cost imo, not reliability. If I help a friend to set up a system I make sure I let them know about the importance of backing up important data. But most people don't care about data security at all.

I'm not waiting for SSD's to replace normal HDD's, I am waiting for them to be
More reliable (sorry my own experience is not very good in this regard)
More cost effecitve, I think they are overly expensive right now
And higher capacity

Everyone is entitled to an opinion on this. Quite happy for you to disagree with me, my opinion for now remains the same.
I can comment on reliability since I have been building PC's for some time, and I have on request fitted SSD's so I have first hand experience on them.
In the same way I comment on specific makers whom I have had issues with or have had problems. I try to take a faulty product in the context of other makers and how many I find cause problems over a longer period of time.
I find the failure rate on SSD's to be quite a lot higher than normal HDD's, that is my own experience which may not match up with other people's

I've never personally observed an SSD failure, but the technology is still new and while I build system for friends now and then I've only seen perhaps ten SSDs total. I realize this is anecdotal. Is the reliability of SSDs really that bad? Seems to me that in laptops at least they must be more reliable than HDDs. I've seen quite a few dead 2.5" HDDs from laptops. The last one was dead-dead, wouldn't even spin up.

Every HDD will fail. Every SSD will fail. If we accept this to be true for a second, then the only solution is to backup.

If we don't want to backup, we should keep our data engraved in heavy stone plates, since i know some stone inscriptions still legible after several hundred years.

I don't consider my own personal experience to be a representative sample, however we do have some useful info on return rates from a major etailer. HDD return rate averages 2.09% and SSD averages 2.06%. The data is confounded by user returns for reasons other than failure and returns direct to the manufacturer, but assuming that this is a constant factor amongst drives, it does suggest that SSDs aren't quite as ruinously untrustworthy as your personal experience indicates. The return rate for the Intel SSDs in particular were lower than any model of hard drive.

As for SSD longevity, synthetic benchmarks taken at xtremesystems on write durability suggest it isn't a problem for most ssds, (assuming that's the most likely mode of failure). For example, the 830 has had over 3.7PB written to it. In 6 months, I haven't written 3TB on mine (and I have the pagefile on it), so that would be over, what, 600 years use? The solder on the pcb will have corroded by then.

I agree with everything else in your post, but I have to pull you up on this one.

When you get a "bad sector" beware: there is a very good chance you will get more, if that is the case then statistics show that your drive will be a brick within 6-months. Where do "sectors" reside, on the platters of course (AKA "media"), you can re-install, you can do a full erase (as well as many other things to "fix" the problem) and re-install, but if you are getting more bad sectors you will keep on having to do this due to data corruption, I don't need to exaggerate when I say that I have seen this 100+ times in the last decade.

Also the "HDD correcting itself" relies on using "spare sectors" that are not user addressable, these sectors are replaced automatically when the drive detects and automatically re-maps the bad sector to one of the spare sectors via SMART. SSD's also do this for the same basic reasons, although I know far more about HDD's and reliability than SSD's (due to the obvious fact that SSD's have not been around for very long and there are a fraction of the amount in the wild), I have seen many HDD's that have replaced bad sectors that survive with no further issues for years, but I have also seen many that have sectors failing faster than the drive can replace them which usually results in the drive running out of spare sectors meaning that the drive is unusable - personally I put a drive "on-watch" when it hits 20-30 faulty sectors, once a drive hits 50 - its replacement time.

A fully working boot drive and a full data backup (of all drives) is the ideal position, I have that, most people don't.


Andy

Or just buy this.

Might find some interesting data here....

What I meant was that the disk itself doesn't fail, but things like the head hitting the disk, where as in the case of flash, the media (transistor) fails. I don't meant the media doesn't fail, but most media failures aren't directly the media, but something else failing, and screwing it up.

Just to point out that following some recent price changes the Samsung 830 series and Intel 330 series are now amongst the cheapest SSDs around, certainly in the UK at least. There isn't really much need to look any further in my opinion. We used the Samsung drives as OS drives in our small form factor servers without any issues, we've recently switched to Intel based more on the marketing value of their perceived superior quality and reputation than any dissatisfaction with Samsung. I wouldn't use OCZ or some of the more consumer-orientated brands in a serious application, although to be fair the four OCZ Vertex drives in my gaming PC haven't put a foot wrong in three years or so (running in RAID0!)

Those are old Q1/2011 statistics. Here is some new info (04/05/2012) from the same source. Crucial also makes most reliable RAM modules.

http://www.hardware.fr/articles/862-7/ssd.html
Crucial 0,82%, Intel 1,73%, Corsair 2,93%, OCZ 7,03%

A few observations:

First, it's not like normal hard disks don't fail. So the choice between "unreliable SSD" and "reliable rotational disk" is a false choice. However, the numbers I've seen indicate SSDs have already surpassed their rotational counter-parts in terms of reliability.

Second, "untested?" No way. Some manufacturers are on their third or fourth iteration of SSD technology, and wear leveling algorithms have been around for decades. It's still certainly young technology, but "untested" is not a word I would use to describe it.

Third, reliability becomes less of an issue if you're really, truly serious about backups. The reality is: don't have backups? It doesn't matter what drive technology you're using; you're playing with fire.

Lastly, you would pry my SSD out of my cold, dead hands. I worked on a Crucial C300 for about a year and absolutely loved it. I then went to a Mac Mini with a normal laptop drive. It. Was. Awful. Replaced that with a Samsung 830, and it was again a pleasure to use. SSDs make my computer a joy to use, moreso than nearly any other piece of hardware I can think of.

I'm a bit late to the table, but I just wanted to consent to the consensus

TLDR: All drives fail, redundancy is key, SSDs cost more, especially when weighted for longevity.

I've had wonderful experience with my Crucial C300 64GB, which I've been using as an Ubuntu drive with nearly 5,000 hours uptime. I also keep a game or two on it, and frequently swap 20GB+ of game data when my play habits change (ie, I've done what amounts to many full writes). Anecdotally, this drive has performed admirably!

All that said, I keep all of my data on a HDD RAID (yay mdadm, yay GNU/Linux). Keeping data you care about on a single drive is a recipe for disaster at least once in your lifetime. If you have proper backups in place, dependability of a drive should not even factor into your considerations except:
1) A drive which fails faster costs more per operational hour. (Think cost per GB-hour rather than just per GB)
2) A drive failure can cause frustration, especially if the error occurs at a bad time and without a fast work-around. If you have a planned work-around (an alternative boot disk?), this is hardly an issue.
3) A drive failure costs user time to replace a drive and prepare its successor. The cost of this time depends on your circumstances.

IMO, if you have a good contingency plan, it all comes down to hardware cost vs gains. Some simulations I used to run were so dependent on large 100+GB database IO, and SSDs were worth their weight in gold (2 hours vs 2 days for some operations). If you are a hardcore gamer, perhaps load times are worth giving up a few FPS on some other piece of hardware. If you are just interested in your OS boot times or noise, obviously it depends on how much money you are willing to spend on a 'wow' factor, which probably has to do with your disposable income (or lack of financial planning).

I don't care if it's your computer-illiterate grandmother's computer. If she has any data that she cares about (server-deleted POP email, photos, etc) you should set her up with some sort of multiple-disk setup. SSD, HDD, they will all break at some point, and unless the user has a much shorter life expectancy than the hardware (<1 year?), data loss is a heart-breaking risk.

Here is an interesting forum topic which deals with realworld SSD write endurance testing with random incompressible data: http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm

Here is a summary picture about some different SSD drives which has been tested till their 'death':
http://img515.imageshack.us/img515/5008/ssdwriteendurance.jpg
http://img13.imageshack.us/img13/8324/ssdwriteendurance2.jpg

MWI (marked with blue): media wearout indicator, which is a write amount counter set by the manufacturer. Ususally this counter counts backwards to reach zero after the factory specified write amount. If the SSD reach this write amount, it doesn't necessarily mean, that the drive will not work anymore, but usually the manufacturer will not provide or limits the warranty of the drive after this counter reach zero.

It's also intresting to check how the different manufacturers setup this counter, some of the drives can withstand 4-8X more write amount than the factory specified value without any user noticable problems. The role of this value will be more important nowadays with the newly released , cost effective 20nm MLC NAND cells, with way lower write endurance than the old ones. With these drives the manufacturer could decide, to go for a much stricter MWI setup to limit the warranty, and the MWI value probably will be much closer to the real world write endurance.

The price/ capacity of SSDs should only be a concern for people who, for whatever reason, insist on having a single drive for anything. With the 256gb Samsung 830 being WELL under $200 now and very reliable, any performance computer should have one of those and add 2tb spinny drives as required. Data that takes up the bulk of your drive space - music, video, and photos - doesn't need to sit on fast SSDs long-term.

If you buy a cheap OCZ drive and it tanks on you, you got what you paid for.

If you leave any data on only a single drive, you're playing Russian roulette. Don't do that. A drive failure should be a minor inconvenience. They should also be rare, but will occur.

That said, there isn't even a conflict anymore. SSDs used to need to be 32-64gb to be affordable, and it was easy to get those feeling cramped with Windows and a program that benefited from fast loads. No more. If you build even an i5, your OS shouldn't be on a spinny drive.

If this is the case a 128 GB or even a 64 GB SSD can do fine. If you want everything to appear as one drive learn how to use symlinks, junctions and compression.

Symlinks can make anything appear to be on one drive while really taking up space on another. You can put your huge MP3/Photos/Video collection on your HD while still appearing to be installed on your SSD. You can take games and apps that you don't use often an move them to your HD while still being able to run them.

Junctions are a bit more coarse than symlinks, they make a whole drive appear as a folder on another. You could have your media drive mounted in a folder on your SSD for example.

NTFS compression works really well for apps (not games) on non-Sandforce SSDs. The compression in NTFS was designed back in the 486/Pentium 1 days so it is really fast on modern computers. It can actually be faster than non-compressed file reads as less data has to be read from disk. The big problem of compression, fragmentation, doesn't bother SSDs. Compress some of your apps if you are short of space.

I got along with a 30 GB SSD for a while with some creative symlinks and compression.