• Please review our updated Terms and Rules here

Candidate for one of the worst predictions of the future...

Whenever someone speaking about computers talks about "forever," the prediction does well if it holds up for 5 years.

1989 - 5.25" floppy the mainstay with 3.5" taking up about 10% of the floppy market and a handful of other floppy types still being sold in tiny quantities
1994 - 5.25" floppy in significant decline but still shipped with most systems
3.5" floppy disks manufacturing costs declined to about 5 cents each or roughly 30 MB per dollar. Flash drives didn't hit a matching cost per bit until about 2005.

Edit: Okay, maybe the bad prediction by Alan Shugart was not realizing that other users will follow him into the world of not using floppy disks.
 
Last edited:
Well, it's fair to say that his prediction lasted a decent amount of time. Hard disks are still used in pretty significant numbers today, but it is clear at this point that like CRTs their days are numbered.
 
geoffm3 said:
Well, it's fair to say that his prediction lasted a decent amount of time. Hard disks are still used in pretty significant numbers today, but it is clear at this point that like CRTs their days are numbered.
Click to expand...
I'll buy that hard drives are going away when SSDs come anywhere close to matching them in cost-per-bit and write-cycle life-expectancy. I know everybody is obsessed with having everything as fast as possible with as few moving parts as possible because that's more futurey, but personally I like having cheap high-capacity storage.

And I'll take a CRT over an LCD any day.
 
commodorejohn said:
I'll buy that hard drives are going away when SSDs come anywhere close to matching them in cost-per-bit and write-cycle life-expectancy. I know everybody is obsessed with having everything as fast as possible with as few moving parts as possible because that's more futurey, but personally I like having cheap high-capacity storage.

And I'll take a CRT over an LCD any day.
Click to expand...

The cost of flash memory is rapidly decreasing, and MTBF is rapidly increasing at this point. I'll bet that in 5 years time it will most likely be a don't care until you get up into some higher end server storage solutions, and even then maybe not.
 
They were mainly talking about bubble memory, and as I recall there as a lot of buzz and hype about bubble memory back then, but the tech at the time simply wasn't practical. And removable disk and hard disk sizes kept increasing, outpacing that sort of thing for a while.

And honestly, floppy disk - or more accurately removable spinning disk media probably would have lasted a lot longer if it hadn't turned in to a mess of different formats and incompatible physical forms. Meanwhile someone got the idea to attach some flash ram directly to a USB plug - a plug indented to accommodate all those random varying removable spinning disk drive systems - and bypass the entire idea of a "drive" altogether.

The idea that people would stay on 5.25" disks instead of 3.5" does seem a little silly. But in 1984, they were still a bit non-standard and uncommon. The transition was not easy, especially when copy protection was involved.
 
It's positively mind-boggling just how cheap and dense solid-state flash memory is today, especially if we consider how much we used to pay back in the day for a comparably (laughably) tiny amount of storage. In 1983 a 10MB Winchester hard drive system like the one in the IBM 5160 cost about $2,500. That's about $5,800 in 2015 money. You can pretty easily find a 128MB MicroSD memory card for $50 so for the same inflation-adjusted cost as a ST-412 you can have 116 pinky fingernail-sized flecks of plastic that add up to nearly 15TB of storage space. That's 15TB of removable storage in approximately 19cc's of volume, assuming I didn't drop a decimal point somewhere.

(Obviously we can't *quite* achieve that density in a single device yet because of wiring/cooling issues, the flash in a cheap MicroSD card isn't going to blow you to the back of the stadium with its performance, and the $5,800 price point is about ten times what two cheap 8TB desktop hard drives cost, but... still sort of feels like the writing is on the wall here.)

It's really sort of hard to blame someone from 1984 for not seeing that coming.
 
Flash memory will eventually hit a brick wall for a while where it won't get cheaper, spinning hard disks will be around for a long time. Where would be in floppy drive capacity if the technology used on fixed disk was ported over to it?
 
SomeGuy said:
The idea that people would stay on 5.25" disks instead of 3.5" does seem a little silly. But in 1984, they were still a bit non-standard and uncommon. The transition was not easy, especially when copy protection was involved.
Click to expand...

Yea, that's what I found most strange... He claimed that we would never be able to convert all applications from 5.25" to 3.5". I guess that is also a sign of the software industry itself being so young and new. The pattern of every major piece of software getting a big update every few years at the least was not known yet, I suppose.
Because otherwise obviously the next version could just be distributed on 3.5" disks right away. In fact, for a long time, software was included on both 5.25" and 3.5" media in the same box.
Another interesting way of distributing software... perhaps some of you have seen that?
It was a small machine that could be placed at a local store. It contained a HDD(?) with all the software on it, and drives for all the common media (5.25", 3.5", tape). When you bought the software, it would be written to the media of your choice on the spot.
I should still have an article on that in an old magazine, I think the name for it was 'EDOS' or something, can't quite recall, and a quick Google didn't turn up anything.
 
Removable storage has the problem of opening the drive to insert the media which risks contamination or moving the read/write heads. The floppy was absurdly over-engineered with tracks many times wider than actually necessary just to compensate for minor alignment issues. The Zip Drive ran into a lot problems trying to spin flexible media fast. The Orb drive showed just how vulnerable hard drive style removable cartridges were to even tiny levels of contamination. The limits of removable disk technology were clearly recognizable by about 2000.

I reserve worst predictions ever to things that were clearly wrong at the time. SyQuest's estimate that the Zip drive would fail because no one wants affordable storage disputing an industry tradition for many years of cheaper beating out better is almost certainly the worst of all time.

Edit: The projection that 3.5" would not replace 5.25" drives matches the earlier changes in drive type. The companies that added 5.25" drives to their minicomputers did not see much conversion of existing data from the 8" disks to the new 5.25." 9-track tape had a similar pattern. The willingness of PC customers to discard old useless data took a lot of companies by surprise.
 
Last edited:
Even if MTBF for flash goes up, storage time seems to go down. There's no way flash can replace harddisks for storage anytime soon. Tried to read your old USB memory sticks recently? And it's much worse for SSDs that are not powered on, and stored at high temperatures. A couple of weeks in Miami in the summer with a power outage was an example I saw mentioned..

The problem is of course that flash is based on keeping a charge. Won't last. Magnetic does better in that respect.
 
SomeGuy said:
They were mainly talking about bubble memory, and as I recall there as a lot of buzz and hype about bubble memory back then, but the tech at the time simply wasn't practical. And removable disk and hard disk sizes kept increasing, outpacing that sort of thing for a while.
Click to expand...

Bubble Memory was practical, its just they never got to a critical mass, and the price never came down. They were always targeting 10% more capacity at 10% more price compared with floppies.
 
Tor said:
Even if MTBF for flash goes up, storage time seems to go down. There's no way flash can replace harddisks for storage anytime soon. Tried to read your old USB memory sticks recently? And it's much worse for SSDs that are not powered on, and stored at high temperatures. A couple of weeks in Miami in the summer with a power outage was an example I saw mentioned..

The problem is of course that flash is based on keeping a charge. Won't last. Magnetic does better in that respect.
Click to expand...


I'm not sure I agree. Spansion reports up to 20 years data retention time on their MirrorBit flashes with a single erase cycle, up to ten years with 1000 erase cycles...all at high temps (55C). Clearly this technology is maturing.
 
Customers discarded data because they were changing apps too much to worry about importing it to the next one. Look at all the proprietary data formats of that time and all the different apps doing the same exact thing. These days people have been using MS office for ages so keeping that old data is more important.
 
Bubble was non-volatile, but power-hungry and the basic support logic was pretty complex.

However, today we do have MRAM and FRAM.

And I do still retrieve data from tapes and floppies written in the 1970s.
 
FRAM still has the limited-life-cycle limitation that Flash does, though. MRAM supposedly doesn't - guess I'll find out when I ever get around to doing one of my daydreamed homebrew projects with the samples I got from Everspin.
 
I note that TI is packaging some models of their MSP430 MCU with built-in FRAM. It probably doesn't get written too often.

Most consumers don't understand that there are basically two types of NAND flash (much less NOR flash).

Consumer-level NAND flash is usually MLC (multi-level cell) where 2 bits are represented by 4 charge levels in a single bit cell. No simple 1s and 0s, but rather dibits. It's cheaper for a given capacity, but slower and more prone to error. I've seen MLC flash being sold as SLC (single level cell) on eBay for what seems to be too-good-to-be-true prices (so look out!). On the other hand, I have some fairly old flash cards (e.g. SmartMedia) that still have the same error-free contents on them after more than a decade in storage--they're pretty small--32MB and 4MB) but they still work fine.
 
Last edited:
commodorejohn said:
FRAM still has the limited-life-cycle limitation that Flash does,[..]
Click to expand...
Not according to the 'FRAM for dummies' book TI sent me. I don't have it right here so I'll have to google.. and this PDF claims 'number of write cycles: 100 trillion'.
(the PDF is from 2011 though. [Edit: Wait - last modification time is 2011, the doc itself is from 2009]).
This PDF from Fujitsu says '1 trillion read/write cycles'. And 'data retention 10 years'. That's actually not as good as I hoped.

The problem with FRAM seems to be that the activity isn't really there.. most docus are old, and it's not like you can find FRAM many places. There are some chips you can buy though, and TI sells MCUs with FRAM (as has been mentioned in another thread before, I believe).

Edit2: Write cycles are stated as 10^13, 10^14, 10^15, 10^16 according to where you look. TI still states 10^15 for the FRAM in their MCUs, and 100 times the speed of flash.

-Tor
 
Last edited:
You must log in or register to reply here.
Back
Top