Fifteen Classic PC Design Mistakes 806
Harry writes "Once upon a time, it wasn't a given that PC owners should be able to format their own floppy disks. Or that ports should be standard, not proprietary. Or that it was a lousy idea to hardwire a PC's AC adapter, or to put the power supply in the printer so that a printer failure rendered the PC unusable, too. Over at Technologizer, Benj Edwards has taken a look at some of the worst design decisions from personal computing's early years — including ones involving famous flops such as the PCJr, obscure failures such as Mattel's Aquarius, and machines that succeeded despite flaws, like the first Mac. In most instances — but not all — their bad decisions taught the rest of the industry not to make the same errors again."
Big ISA bus flaw (Score:5, Interesting)
IOCHRDY signal is active high instead of active low. Causes no end of problems.
worst: sharp unfinished inside edges in cheap case (Score:5, Interesting)
My personal list...
- 15 to 10 years ago, you had to be careful when installing drives, or RAM. You could almost slice your hand on a cheap case that had unfinished and sharp edges.
- Beige Only. You can pick any color, as long as it is beige. Why did it take so bloody long to offer any other color then beige? Critical mass?
- LOUD systems. Have to thank George for showing me just how nice a quiet system is.
- Power hunger systems. 2 molex connections for a GPU ?!
- Crap 3D Video cards in laptops, and almost no benchmarks from the "classic" hardware review sites so you know how bad it sucks compared to a "real" GPU. (Thankfully the S3 Virge is gone from desktops, but laptops are still stuck with poor performance unless you pay an arm and a leg.)
--
"World of Warcraft (TM) is the McDonalds (TM) of MMOs."
-- Michaelangel007
Re:Keyboard layout... (Score:5, Interesting)
Re:The 15 problems (Score:5, Interesting)
Sony VAIO desktop problem... (Score:5, Interesting)
Re:#1 failure... (Score:4, Interesting)
Why? What other processor(s) should have been used, and what would have been the benefits? No, not trolling. Just interested in what you said and would like more information.
The Amiga (Score:4, Interesting)
It amazes me how advanced this system* was for it's time and that it didn't catch on better than it did. The graphics and sound (just for starters) was many years ahead of it's time; x86 was still in EGA and speaker beeps at the time.
[*] - http://en.wikipedia.org/wiki/Amiga#Graphics [wikipedia.org]
Apple Lisa (Score:5, Interesting)
Does anybody know what the "unique document management metaphor that has yet to be replicated in a mainstream OS" is, and why it might have set a new standard in computing? It sounds terribly intriguing. Might this be something that could/should be added to Linux?
So, the one-button mouse didn't make the list? (Score:4, Interesting)
That was one of the most serious design mistakes of the last thirty years, but it's only really interesting because it's symptomatic of Apple's design philosophy, which is: "Do as I wilt".
The one-button mouse spanned multiple generations of Apple computers and underscored Apple's stubborn unwillingness to produce computers that do what their users want, and not what Jobs or Apple's HID team think they should do.
Really. Apple refuses to correct the annoyances of the UI that should not exist. Why doesn't OSX have a maximize window button? Why does clicking on "one hour before event" for an ical event reset the clock to one hour before the time you click the button, and not one hour before the event? Why doesn't finder support afp connections over ssh?
None of those things seem to be complex, every one of them is a failure of the UI, and yet none of them have been corrected.
PCjr (Score:5, Interesting)
The biggest single problem with the PCjr was that it was late. In 1984 it was supposed to be on the shelf in the fall - October is the usual month when things are supposed to be shipped so they are stocked and on the shelf in November.
Didn't happen. Macy's had received $50,000 to hold shelf space for the PCjr and they left them empty.
The PCjr came out in February. A little late for Christmas. Everyone had created products for Christmas 84 specifically for the PCjr, but there wasn't anything to run them on. January 1985 CES was pretty dead - lots of PCJr games that nobody cared about. Parker Brothers closed down their electronic games division, as did lots of other companies right about then. It was a year or so later that the Nintendo finally started making inroads into the home game market but between the PCjr and Nintendo things were very, very dead.
You can say all you want about a poor design of the keyboard and limitations of the hardware. But it is even more difficult to use when it doesn't exist and cannot be purchased. Not having it in time killed it, not any stupid design decisions.
Re:Worst Mistake That Still Needs Fixing (Score:4, Interesting)
Why doesn't Microsoft just forget software and go into hardware?
Re:#1 failure... (Score:3, Interesting)
I think you're talking about the 8088. The 8086 was a true 16bit chip, the 8088 had an 8-bit bus. The chief reason was, as I understand it, that 16bit hardware was extremely expensive at the time, so IBM went with it to keep the price of the unit lower, and to make it less expensive for expansion hardware to be built.
And that's the real secret here of the success of the PCs and PC clones. They were never as good as a number of competitors; Apple had the better GUI, Amigas had the better graphics, the various *nix workstations beat it hands down, but none of them were as open or as easy to build hardware and peripherals for. The PC was, for all its flaws, a highly implementable open standard. That's why PCs still dominate, by a wide margin, the industry, and why a number of machines that were superior got left by the wayside.
Re:Worst Mistake That Still Needs Fixing (Score:3, Interesting)
My new Mini is actually my first Apple ever. So far, I have not been impressed.
TI Sidecars (Score:3, Interesting)
Re:Worst Mistake That Still Needs Fixing (Score:4, Interesting)
Your "oversized" enter keys are standard for non-US keyboards. I don't like the US style ones, because I am used to a larger enter key..
Re:#1 failure... (Score:3, Interesting)
680x0 was, IIRC, around at the time, and had a much more elegant, though still CISC, instruction set. Plus it was 32bit internally, though the 68000/10 only had 24 external address lines.
I seem to recall that writing (GUI) apps in assembler for the (68000-based) Amiga was, although time consuming, perfectly possible. I'd have hated to do it on the register-starved 80x86.
BS! Re:I don't agree (Score:5, Interesting)
I was programming in x86 assembler (by necessity - not choice) at the time and the X86 instruction set sucks big time. The 68000 was far easier. No programmer worth his salt would choose X86.
The X86 still used 32 bits for the address but they overlapped the two 16 bit pieces so there were many ways to form the same address. It was INSANE!
IBM missed the boat, created a major competitor in the process and short themselves in the foot many times as a result. About all that saved IBM's PC bacon back then was that they had a lot of feet to shoot at.
IMHO when I read the article - its great. It shows how the rush to market can put a company out of business real quick.
BTW, I looked at the Lisa. I didn't buy it. I looked at a lot of the other computers in the list. I didn't buy them. Apple has not EVER sold me a computer. Funny. IBM has not EVER sold me a computer.
I have been running clones since 1986.
I'll predict that Microsoft's days are numbered as well. I think the number might be large however given their cash reserves. However I am hearing people tell me they are sick and tired of the shoddy windows code and the problem with malware. I think a lot of this problem stems from the X86 days and windows 3.11
The way I see it... the general population in many ways is like a school of fish. They tend to clump together for safety reasons. However, few have much in the way of any enduring investment and just like a school of fish they can all change direction rather quickly. If/when this happens then we may see the fortunes of a company like Microsoft turn sour about as fast as we saw the fortunes of GM and Chrysler turn sour.
If this happens then people will not go back. These paths tend to be traveled but once.
Not 'classic', but still... (Score:4, Interesting)
It isn't really a 'classic' mistake, but the biggest PC design problem today from where I'm standing is over-reliance on fans. High volume fans will result in fuzzy lint growing on the devices which can least afford a layer of fuzzy lint.
In the past year, I've revived dozens of computers, and nearly every failure can be directly attributed to lint induced by fans.
A few of my favourite things - from the workshop (Score:5, Interesting)
I could go on...!
Re:What , you mean like the Mac , Amiga, Atari ST? (Score:3, Interesting)
I honestly don't think the PCs success had much to do with marketing at all. I remember that Commodore, Apple and even Radio Shack poured a lot of money into marketing, and even in the mid-1980s, PCs were still very much considered an unsexy "business" machine, with inferior graphics. And yet, IBM won because it was simply more easy to expand than Amigas or Macs. Sure the basic graphics sucked, but you could always go out and buy a CGA or Hercules video card, and later on EGA and VGA and so forth, and it wouldn't cost you an arm and a leg. The reality is that IBM and the clonemakers were smart enough to realize that if you wanted to make a popular general use computer that could be used in most environments (corporate, small business, home and education), you opened things up.
I know that by the time I got out of high school in the late 1980s, the old guard were in their death throes. Commodore and Atari were in trouble, Radio Shack had abandoned its old 8-bit lines and was producing PCs, Apple had dropped its own 8-bit products and was put all its energy into the Mac, but the control-freakish nature meant it was simply cheaper for hardware manufacturers to build for PCs. And that's the trade-off the industry made. Yes, an open architecture meant incompatibilities could creep into systems, and sometimes your video card would play havoc with the serial UART and your modem would suddenly croak in the middle of a 9600kbs transfer, and it was a big pain, but that was the trade-off, and that's why Macs ended up a niche.
Re:#1 failure... (Score:1, Interesting)
There is no basis for your theory.
One could just as easily argue that IF the 80xx assembler hadn't sucked so badly, or if the Motorola 68k everyone was so in love with doing assembler on had emerged victorious, we would not have had compilers and higher level languages that speed up development nearly as early as we did. Being stuck in assembler land is not the way forward. Thus, by sucking as badly as it did, it actually helped speeding up progress.
Re:#1 failure... (Score:3, Interesting)
And much more expensive to purchase and assemble than the Intel chipset. The Slashdot Uber Tech Society often forgets that computers are designed and priced for the end user and the mass market, not the programmer and the Uber techie.
Re:biggest mistake: PC = 8088 not M68000!!! (Score:5, Interesting)
There were other reasons for IBM to go with the 8086-family chipsets:
1) the 8086/8088's bus could easily drive the 8080-family support chips such as the 8251, 8255, 8259 etc. to build a complete system. The MC68k family support chips were even later than the release of the CPU itself (in some cases like the MMU several years late) and the MC68k bus could not be easily interfaced with the Intel family chips which were cheap and in plentiful supply.
2) the 8086 family's internal data registers and addressing modes were designed to simplify conversion of existing 8080 code to run on the new 16-bit CPUs. The 68k, although a superior CPU in all respects to the 8086 family, had no tools available to make code conversion from the 6800 or other sibling CPU family (6809, 6502 etc.) simple -- all 68k code had to be written from scratch.
3) the 68k was an expensive chip, not suprising as it was complex and required a large die, necessitating a 0.6" wide 68-pin DIL ceramic package. Motorola's target market for the chip was $10,000 workstations, not "toy" desktop computers only costing $2,000. By comparison the 8088 was cheap as chips.
No standard connectors in 1983 (Score:4, Interesting)
The PCjr's serial port, monitor port, joystick ports, keyboard port, and others used different connectors from the IBM PC. In fact they were not only non-standard connectors, but completely proprietary connectors that couldn't be found on any other computer.
People, this is 1983. All connectors were "non-standard". Nowadays we're used to a standard connector and pinout for RS-232 [lammertbies.nl] and parallel [diyha.co.uk] ports on the back of PCs. But in 1983, exactly one model of computer used them: the IBM PC. It didn't more than a couple years for people to realize that the only way to compete with the IBM PC was to be extremely compatible with it. But when the PC Jr. came out, everybody (especially IBM) used business and sales models that paid no attention to the idea that computers and their components could be commodified.
Small qualification: the use of 25-pin D-shaped connectors with specific pinouts was part of the RS-232 standard. But 25-conductor, straight-across cables cost, and you actually didn't need most of those signals for typical applications. So making cables that would connect some random computer to some random modem or serial printer was a serious black art. There was even a book [amazon.com] on the subject.
(Jerry Pournelle once wrote that he used internal modems because he could never remember the pinouts he needed to make cables. But by the time he wrote this, RS-232 pinouts had been standardized and cheap pre-made modem cables were in all the stores. Pournelle is the original know-it-all ignoramus computer pundit.)
Parallel printer cables were even worse. They all used the Centronic [pinouts.ru] de-facto standard on the printer side. But to save money, everybody used 25-pin D connectors at the computer side, and the way the 36 Centronics signals mapped to those 25 computer pins was different for every manufacturer. It took IBM to standardize the pinout, and also to standardize making the printer connector female so you didn't accidentally plug a modem into it.
Re:Keyboard layout... (Score:3, Interesting)
Clearly you're not thinking differently enough!
Re:worst: sharp unfinished inside edges in cheap c (Score:3, Interesting)
I heard that beige was chosen because it is the same color as most dust so beige computers don't look dirty.
Re:Worst Mistake That Still Needs Fixing (Score:3, Interesting)
I wish that more keyboards were available without numeric keypads. If you're right-handed and use a mouse a lot, the extra keyboard length puts the mouse in an unnatural position.
This is not a problem for left-handers, but alas for them, it's hard to find a true left-handed mouse - most mice these days are designed to be mediocre for either hand.
Re:Big ISA bus flaw (Score:4, Interesting)
Depending on the bus design, active-low signals can be asserted by any device by turning on a transistor to ground. They are allowed to "float" high via pull-up resistors, so you get a poor-man's OR gate.
And depending on the circuitry you're using, "drive down, float up" may be much, much, much simpler than "drive up, float down". In the pre-CMOS days, for example, only N-channel FETs were available. So a transistor to drive a bus line low is cheap, but it's not pleasant driving something high. (Fortunately, I've forgotten most of the details.)
It's one of those places where reality and theory diverge.
It too, has a single tragic design flaw (Score:5, Interesting)
Here is an article with a picture of one. [pctechguide.com]
I'm a touch typist, took a class in it in high school. Fingers on the home keys. Left hand rests on ASDF. Right hand on JKL;.
If you move up a row from ASDF, you get QWER. My left pinky is A, move up 1 to Q. My right pointer is on F, move up 1 row to R.
Move up to the next row for numbers. ASDF becomes 1234. Now here's where we get to the mistake. We were taught that your left pointer goes up 2, and towards the middle 1 to get to 5. Likewise, your right pointer goes up 2 and over to the middle one 1 to get to 6.
Notice how the 6 is on the wrong side? When my brain thinks "6", my right pointer wants to see it right next to the 7. It's now the responsibility of my left pointer to be in charge of 456, and my right pointer is now only in charge of 7.
I can't tell you how frustrating this keyboard is to a touch typing programmer. It's as if nobody at Microsoft knows how to touch type.
Re:Worst Mistake That Still Needs Fixing (Score:5, Interesting)
I wish someone would bring it back, duplicating the TY, GH, NM keys on both the left and right side.
This. Very, very, very THIS. Please. And hurry...
Re:Big ISA bus flaw (Score:3, Interesting)
Re:The 15 problems (Score:5, Interesting)
Re: The 15 problems (Score:2, Interesting)
Do what i did... Change the setting of the power button to "Hibernate". When you go to sleep, just push the little button... Problem solved. No more flashing.
Re:Big ISA bus flaw (Score:4, Interesting)
It's been decades since I've worked on ISA bus stuff, but IIRC, IOCHRDY is essentially active-low. Any card can pull it down to add wait states to the current cycle, then they let if float back up when they're ready.
The main problem with the ISA bus is that it was never engineered in the first place. The people in the skunk-works PC project at IBM slapped it together by tacking a few TTL kludges onto off-the-shelf Intel I/O parts, probably without doing any formal timing analysis. That probably worked OK at the original 4.77 MHz, but within a few years the bus had been overclocked throughout the industry to 8MHz. (I think that Dell, then known as PC's Limited, tried pushing the ISA bus to 12MHz, but that bad idea was quickly dropped.)
One project task I had in the 1980s was to sit down and to a complete timing analysis of the IBM PC/AT bus (which added yet more kludges to the original PC bus to go from 8 to 16 bits) based on the circuit diagrams in their technical reference. Some of the timings just can't work using the worst-case specifications. The computers usually worked mainly because the odds of getting actual worst case behavior out of several chips is rather low. However, there was no shortage of incompatibilities and crashes with a lot of 3rd party ISA adapters.
Re:Apple's fascination with single button mice (Score:3, Interesting)
"And the method of right-clicking on a laptop touchpad (two-finger click) is simple and intuitive"
If by simple and intuitive, you mean "in the case where you learned the behavior repeatedly and are now intimately familiar with it", then yes. But for about two or three months after I started using a MacBook, I didn't know about it. I just used Ctrl-Click for right clicking.
And to be completely honest, I have found that I'm searching out more keyboard shortcuts because using the touchpad is an overall pain.
Re:Worst Mistake That Still Needs Fixing (Score:5, Interesting)
...or the total overkill that is the Mac pro line...
As someone who also got bit by Apple's non-user serviceable part philosophy, I agree with you 100%.
I've got a Mac Pro. I'm not an Apple fanboi, I just hate them less than other computer manufacturers. My computer works great. But I didn't get the wireless card installed when I purchased it because I didn't need it. Later on, I needed the wireless capability, so I tried to buy the Airport Extreme card from Apple. The fuckers (yes, they are fuckers) wouldn't sell it to me because "it was not a user installable part." I had to make an appointment at my "local" Apple store that is 60 miles away to let some teenage "genius" install it for me. Yeah, OK, I'll get right on that, because I really want to drive my expensive 90-lb machine 120 miles on my day off so some 13-year-old-looking smartass can paw at it.
Instead, I bought it off a third-party vendor and worked out how to install it myself, since the only instruction it came with said "This is not a user installable part, please refer to the Mac Pro service manual for installation." It worked fine and I now have wireless capability, but I found Apple's actions with that upgrade really insulting.
If I am willing to pony up $4000 for a computer, chances are I have the necessary intellect and experience to screw a wireless card to my motherboard and plug in two antennas. Or I am willing to accept the consequences of my actions if I screw up. Why would a company make it hard for a consumer to use their product?
Apple's increasingly common philosophy of non-user serviceable parts, lack of mid-range user-upgradable towers, and forcing weird connectors down our throats without including the adapters for free are annoying and I think, ultimately, holding them back in the PC market. Window's recent suckage has been working to Apple's advantage, but I feel they could have capitalized on it more effectively. Of course, I am sure that Steve and his financial analysts have determined otherwise.
Re:Worst Mistake That Still Needs Fixing (Score:5, Interesting)
It's not just Apple that charges such a large amount for better parts. Dell (whose computers you can easily upgrade on your own) has prices on upgrade parts that are much higher than retail.
For example, a base model Vostro desktop lists the Core 2 Duo E8600 as an upgrade (over the Celeron 450) for $330; the E8600 can be bought [mwave.com] for $267.99 with free shipping. Dell lists their 21.5" HD monitors for $260; I recently bought two Samsung 21.5" HD monitors for $189.99 each [newegg.com] (with free shipping, and there are rebates available). Dell will upgrade your baseline Vostro from 1GB to 4GB of 800MHz DDR2 for $112; it's not hard [mwave.com] to find 4GB kits for anywhere between $40.99 and $76.99, depending on what brand you prefer. On the same machine Dell will upgrade your 80GB hard drive to a 1TB 7200RPM hard drive for $330; Seagate 1TB drives can be had [mwave.com] for as little as $89.99.
(Those aren't affiliate links, don't worry :P)
If you were to get those upgrades, Dell's markup over retail prices is as much as $400, and they pay OEM price, not retail. (To be fair, the hard drive I linked above to is OEM, not retail.)
These days, I see very little reason to buy a desktop from Dell (or Apple or whoever) unless you're buying a laptop - and even then, you shouldn't have the vendor upgrade your RAM. I bought 4GB RAM for my laptop for $20 (after rebate), where Dell would have charged me $200. (Ironically, the RAM was marketed as "for Macs", despite being standard DDR2 SODIMM.)
As a humorous side note, if you want Dell to preconfigure RAID on a pair of 1TB drives, they'll do RAID-0 for $350 or RAID-1 for $250... same hardware, different price. Fun fun fun.
Re:Apple's fascination with single button mice (Score:3, Interesting)
The mighty mouse has a middle button too if you want it (which I do). However, just like mice with a scroll wheel, trying to get a middle click without scrolling at the same time takes some practice...
Re:#1 failure... (Score:3, Interesting)
Having experienced both the 68000 and the x86 (in both old-school segmented mode and 386/486 32-bit flat mode), I agree with the statement that the 68000 is the superior CPU.
What I want to see is for someone (Intel maybe) to invent a new PC that gets rid of all the legacy cruft.
It would be a PC without any support for (even in the CPU/chip-set/etc):
Floppy Disk Controller/Floppy Disk Drive
IDE port
PS/2 keyboard and mouse ports
Serial and Parallel ports
PCI and ISA slots
VGA port
Interrupt Controller (or is that still used even today with PCI Express?)
Internal software modems
Etc
Also, it would use Intel EFI and not the legacy BIOS. It would use all the modern technology (SATA for disks, USB for peripherals, latest Intel CPU, DVI or HDMI for display output, PCI express for add-in cards etc). In addition it would transition from real mode straight into 64 bit flat protected mode as the first few instructions executed by the BIOS code. It would be compatible with all current hardware (only possible issue there is finding PCI Express add-in cards that aren't in some way graphics related, e.g. PCI-e WiFi cards) and could work with most current OSs with minimal changes (if any are needed at all).
So its the PC minus all the bits 90% of people don't need anymore and that just take up silicon and board space. Even in 2009 and even on motherboards for the latest Core i7 speed machines from Intel, you still get a Floppy connector AND PS/2 keyboard/mouse connectors. Cant we just move to USB for keyboards and mice and forget that PS/2 (the computer and the keyboard and mouse ports) ever existed? Oh and can we also stop shipping 32 bit operating systems (Vista and 7) on machines that have x86-64 support in the CPUs please?
If you actually NEED a floppy disk or a PS/2 keyboard and mouse port or a serial port or whatever, there would be many other options to pick that DID have the legacy junk. But my idea means those who don't need the legacy junk don't have to get it anymore.