That is always the entertaining part in talking to some service guys from navy and air force, how much of the most advanced equipment is still run by some computers built around 20 years ago.

 

Probably because that generation of computers was more reliable!

 

The most reliable system is still the Eyeball Mk I.

 

Funny as it sounds, an old Bofors 40mm with visual guidance is not going to fail you on air defense, even if the computer and radar completely fail. So in some ways, older ships are more dependable, if less survivable.

Yep, if you're going to go down, you may as well have something to do while you wait.

In the same way it's nice to have some (unreliable?) radars, computers, SAMs, rotary canons etc to have something to watch as well.

As to the reason for keeping old computers for fire control and the like, probably a pretty easy job for a computer and the old ones do it fine. I think original fire control computation was done by mechanical computers a bit like large adding machines (a small room full of gears and shafts), I think it was one of these units I saw shoot down a towed target with a 4.5" gun back in the early 70's.

Don

But in the case of computer systems available to the public, the old stuff was more reliable. The recent generations have evolved in such a highly-competitive enviroment that robustness often has been sacrificed to the gods of complexity, "capability", and sale-ability. If it can be touted as having better features, higher technology, and/or higher speed, or simply being more in vogue, than the competing brand, then it's unleased upon the market - despite often being ill-tested, bug-ridden, improperly integrated, and/or having a really marginal design that's prone to failures, especially when "stressed".

 

I (somewhat fondly) recall the days when the only headaches one had with computers were the result of one's own faulty programming logic. Nowdays, even perfect programming seldom guarantees proper performance - because of faults hidden elsewhere in all the not-so-robust layers of complexity.

 

Heh, hogwash. The old stuff wasn't more reliable at the same point in their lifecycle...

 

I wasn't speaking to military hardware in general, nor even to militarized computers...

 

If you are talking about plain jane commercial hardware, it is even less true. If you want to pay the same *adjusted* price for hardware and software, you can get better reliability and performance. ...

 

I was speaking specifically about the reliability of commercially-available computer systems for industrial, business, and personal use. And I'm not disputing raw performance - although performance means squat if the system is down due to reliability issues - or even worse, if it is "working", but is merrily doing something that is totally bogus or even dangerous!

 

 

... But I don't see many people spending megabucks to buy a super-solid computer any more... And it doesn't necessarily make much sense to do so. Tradeoffs are a fact of life...

 

In the not-so-distant past, it wasn't necessary to buy any high-priced or "hardened" computer in order to have ultra-reliable operations out of an "ordinary" computer system; but today, it is virtually impossible to buy a robust modern system - at any price. It's really surprizing that we don't have more "industrial accidents" or financial melt-downs than we do - considering the high level of reliance on computerized control systems throughout the manufacturing and business sectors. Of course, as the frequency of failures or accidents increases, the "tolerance threshold" goes up as well - so it now takes a more major incident to become "worthy" of notice, while all the "little" screw-ups and minor "accidents" are just seen as the norm - and as "acceptable trade-offs" for "higher performance". This sort of trade-off isn't a fact of life - but it can be (and sometimes is) a matter of life and death. Is human life now being sacrificed on the altar of high-tech and high-performance? Seems like it is...

I wasn't speaking to military hardware in general, nor even to militarized computers...

 

If you are talking about plain jane commercial hardware, it is even less true. If you want to pay the same *adjusted* price for hardware and software, you can get better reliability and performance. ...

 

I was speaking specifically about the reliability of commercially-available computer systems for industrial, business, and personal use. And I'm not disputing raw performance - although performance means squat if the system is down due to reliability issues - or even worse, if it is "working", but is merrily doing something that is totally bogus or even dangerous!

 

 

... But I don't see many people spending megabucks to buy a super-solid computer any more... And it doesn't necessarily make much sense to do so. Tradeoffs are a fact of life...

 

In the not-so-distant past, it wasn't necessary to buy any high-priced or "hardened" computer in order to have ultra-reliable operations out of an "ordinary" computer system; but today, it is virtually impossible to buy a robust modern system - at any price. It's really surprizing that we don't have more "industrial accidents" or financial melt-downs than we do - considering the high level of reliance on computerized control systems throughout the manufacturing and business sectors. Of course, as the frequency of failures or accidents increases, the "tolerance threshold" goes up as well - so it now takes a more major incident to become "worthy" of notice, while all the "little" screw-ups and minor "accidents" are just seen as the norm - and as "acceptable trade-offs" for "higher performance". This sort of trade-off isn't a fact of life - but it can be (and sometimes is) a matter of life and death. Is human life now being sacrificed on the altar of high-tech and high-performance? Seems like it is...

 

In the not-so-distant past, you spend a premium for a computer and didn't get much in the way of reliability. The common wisdom (before the price fall) was that $2000 would buy you a "good" computer for any given period. You can buy a "good" computer today for just about 20% of that today, get more reliability and more power. Of course, if you want to go out and buy a computer worth $2000 in 1990 dollars, you'd get one that would blow away the reliability of even military/space hardware of the "good ole days". Or you can buy a nice gaming rig. And that is where the money is going these days... Getting capabilities you couldn't get back then.

 

If you want reliability, you can buy consumer hardware these days that detects memory errors caused by cosmic rays (Dude, you can buy a full gigabyte of ECC DDR2 RAM for $19.99!), has multiple CPUs for crosschecking results, redundant mass storage (with error checking on top of that and terabyte (yes, about 50 thousand times larger than my first drive) drives are running $75 on sale regularly), etc etc etc. But most people don't want it because their machines have all the reliability they want to buy...

 

I'll not shadowbox with scary pronouncements about how dangerous this brave new world of computer control is... I'll let the steadily increasing life expectancies and steadily decreasing accident rates speak for themselves.

In the not-so-distant past, you spend a premium for a computer and didn't get much in the way of reliability. The common wisdom (before the price fall) was that $2000 would buy you a "good" computer for any given period. You can buy a "good" computer today for just about 20% of that today, get more reliability and more power. Of course, if you want to go out and buy a computer worth $2000 in 1990 dollars, you'd get one that would blow away the reliability of even military/space hardware of the "good ole days". Or you can buy a nice gaming rig. And that is where the money is going these days... Getting capabilities you couldn't get back then.

 

As I mentioned, I'm not debating capabilities... but I am most certainly challenging the idea that reliability is improving. More importantly, in the situations that I'm thinking of, few of these "advanced capabilities" are needed, yet robustness is a requirement... so, it's not a good trade-off to buy unneeded capability at the cost of sacrificing urgently needed reliability.

 

 

If you want reliability, you can buy consumer hardware these days that detects memory errors caused by cosmic rays (Dude, you can buy a full gigabyte of ECC DDR2 RAM for $19.99!), has multiple CPUs for crosschecking results, redundant mass storage (with error checking on top of that and terabyte (yes, about 50 thousand times larger than my first drive) drives are running $75 on sale regularly), etc etc etc. But most people don't want it because their machines have all the reliability they want to buy...

 

Even if it were entirely true that hardware robustness has improved (in some ways, it has; in other ways, it has gotten worse), hardware is only part of the issue - and probably the smaller part as compared to software issues. (Or, don't you believe those Mac/PC commercials?? )

 

 

I'll not shadowbox with scary pronouncements about how dangerous this brave new world of computer control is... I'll let the steadily increasing life expectancies and steadily decreasing accident rates speak for themselves.

 

You must have a different statistician than I do... or perhaps we're speaking of different situations. As I mentioned, if accidents now need to be much more serious in order to be considered noteworthy, then perhaps the recorded accident rate is decreasing. But, in my industry, we've seen a lot more systems failures in recent years than in the "old days", that were traceable to operational failures in the control computers or even in microcontrollers (which is the really scarey situation). It's not that the "brave new world" of computer control is inherently bad, but rather that reliance on inherently unreliable computer systems is dangerous. And many -if not most- of today's mainstream commercial and industrial computers/controllers are inherently less robust than their predecessors - due in no small part to their "enhanced capabilities".

 

My whole point here is that "advancement" (i.e. - being more modern & more capable) doesn't necessarily imply "improvement" in all respects, and the decrease in computer robustness is one big example.

The LCS runs on Windows 7 doesn't it?

I wrote a much longer reply but it mostly goes over similar points as before.

 

Suffice to say, the trade off between capability and reliability is always there and is a human choice within the boundaries of the available technology. Modern technology has given us a much larger space to operate in and somebody can opt for a cheaper option with "just enough" reliability or go the gold plated route. In the end, we have a continuum from the NASA model where a large amount is invested in a robust solution to the consumer electronics end where price is king and it only needs to be "mostly" reliable. And we have plenty of examples from both ends.

 

I'm betting that you aren't using a computer from yesteryear nor spent yesteryear's value equivalents.

The LCS runs on Windows 7 doesn't it?

 

I don't remember hearing anything about that but both the USN and Royal Navy have dabbled with Windows onboard warships, IIRC.

 

No accounting for taste... But that indicts military procurement and a particular implementation rather than any particular state of technology.

I wrote a much longer reply but it mostly goes over similar points as before.

 

Suffice to say, the trade off between capability and reliability is always there and is a human choice within the boundaries of the available technology. Modern technology has given us a much larger space to operate in and somebody can opt for a cheaper option with "just enough" reliability or go the gold plated route. In the end, we have a continuum from the NASA model where a large amount is invested in a robust solution to the consumer electronics end where price is king and it only needs to be "mostly" reliable. And we have plenty of examples from both ends.

 

I'm betting that you aren't using a computer from yesteryear nor spent yesteryear's value equivalents.

 

Personally, or on the job?

 

Personally, I currently use PC's dating back to 1995 (although even they don't date back far enough to be very reliable ). I also have a couple of Atari 800's set up as controllers, one managing a thermal storage unit, and one currently "unemployed". Now, those are old enough to be highly reliable - er, well, at least so long as you don't use MS BASIC on them... Sure, they don't have all the bells and whistles nor the performance of modern PCs, but the important thing is that they do the required job without blowing anything up! We once pondered putting newer PCs in place of them, but the costs of the required add-on hardware that's needed to accomplish even the simplest I/O tasks exceeded the costs of the PC, and far exceeded the cost of the entire Atari set-up, which had all the needed I/O already built-in. In addition, the PC implementation was found to be a substantially non-robust system, and the only real benefit that it offered was the capability to be networked more easily - which was not a basic need for that particular application, anyway.

 

On the job, we were generally forced by dictum over the years to use only the "latest and greatest" PC systems, even when they were proved to be inadequately reliable. As time has gone on, and newer and newer PCs have replaced "old reliable" computers, we've experienced an increasing number of spurrious malfunctions and accidents, which the powers-that-be deemed to be "insignificant" in comparison to the "benefits" afforded by the new PCs - at least from their perspective... a perspective that I seriously doubt is shared by some of my colleagues who were moderately injured when a portion of a large machine exploded due to a flashback which occurred when the controlling PC decided to take a time out in the midst of a shutdown sequence, and thus failed to switch some valves when it was supposed to. This is only one of many similar mishaps, but apparently, it is so important for the powers-that-be to keep up with the latest Internetworking gimmicks that they "overlook" such "minor" risks... and sometimes I seriously believe they might not even consider it a big enough "problem" if the entire facility blew up! IMO, they would have been much better served to leave the "old reliables" in place, and use existing interfaces to pass data and controls to/from some newer, higher-capability PCs, rather than insisting on supplanting the reliable systems entirely... but that would undermine their "progressive" images, I suppose. And one big irony of the situation was the "justification" claim that the older systems would be "wearing out", so new PCs should be installed in order to ensure "continued reliability"!! Yeah, right! What a laugh!