Retro Machines Key to Rescuing Old Data

SimilarityEngine writes "New Scientist report on the virtues of old kit. From the article: 'Today's stylish PCs may perform billions of calculations a second and store tens of billions of bytes of data, but for many, they have got nothing on the 32, 48 or 64-kilobyte machines that were the giants of the early 1980s. This renewed interest in old-school computing is more than just a trip down memory-chip lane. Early computers are a part of our technological heritage, and also offer a unique perspective on how today's machines work. And within growing collections of original computers and home-made replicas, and the anecdote-filled web pages and blogs devoted to them, lies the equipment and expertise that will one day help unlock our past by reading countless computer files stored in outmoded formats.'"

Comments

First Post?

[nearlygod]

My friend John Titor told me that the IBM 5100 is going to be very popular soon.

Data?

[lachlan76]

Which storage media would last this long? What's the point of using old computers to get your data if the media is dead?

Re:Data?

[Jay Maynard]

I regularly read 9-track tapes written in the late 60s.

The tapes I have the most problems with are actually from about 1984-1987 or so...Memorex and BASF switched to a binder (the stuff that keeps the oxide on the tape) in those years that tends to migrate to the surface, making the tape stick to the read/write head and preventing it from reading correctly. There are ways of correcting the problem long enough to read the data, but I haven't been able to try any of them (the best, supposedly, is to run the tape through the same process used to freeze-dry food commercially).

old cruft

[spectrokid]

Funny that archeology in the future will be totally different. Instead of trying to maximise information out of a 2500 BC chicken bone, the art will be how to distill meaning out of gazillions of backup tapes... But true, I already spent half a day once trying to load my own thesis....

Re:old cruft

[theonetruekeebler]

I doubt that future archeologists could learn anything more interesting than what they could learn from all our print, audio and video media

Until they get to the patient records archives at the CDC or even a local hospital's TB clinic. Then they can learn a whole hell of a lot about how a disease used to spread and its epidemiological characteristics in a society that doesn't have "modern" medicine to control it.

I worked for Georgia's Division of Public Health in the 1990s. One of the most interesting projects I worked on was to recover data from the Medical College of Georgia's TB clinic. It was all on 9-track tape and was recorded from 1966 to 1973. The doctor who wrote the software was in his late 70s when I met him. He still understood the data encoding that he created for his clinic's dinosaur computer system and was working independently to import it all into a PC-based database. The concept of relational data was practically alien for minicomputers of the era; the way he had to encode the clinic's data to build statistical models out of it was fascinating, but it would have been lost forever if the original coder weren't still alive.

BS.

[tomstdenis]

You don't need a Vic-20 to read an audio cassette tape... you just need something that can capture the audio stream, some sort of analogue signal converter capable of producing a binary digit stream. Something like an "analogue-to-digital" converter if such said device exists all our problems are saved! ... /sarcasm

Yes, retro computing is cool. No, it's not required to read ancient recording formats.

Tom

Re:BS.

[mabinogi]

The point is that the interest in reto computing is keeping the _knowledge_ required to convert the stream of data to something useful.
Things like old manuals and spec sheets that might tell you exactly what encoding is used in the data. (Since manuals actually contained real information in those days, rather than being purely a vehicle for "Screw you, don't blame us" EULAs and disclaimers)

Re:BS.

[Knx]

I tend to disagree for several reasons.

1) It's probably going to be actually hard to find any tape recorder in the next coming years, just like it's not quite easy to find a Vic-20 today.

2) Many programs were (are) protected, using very specific properties of the original hardware used at the time. That's mostly true for floppies, for instance. (Just try to read a protected 3"1/2 Atari ST or Amiga floppy on today's PC floppy drives -- if your PC still has one -- and you'll see what I mean). But even some audio tapes are not that easy to decode correctly without the original tape recorder.

3) Audio cassette tapes is just a special case, anyway. How do you read, say, a 5"1/4 floppy? How do you read cartridges without the original hardware? You may try to build a homebrew cart dumper, but you'll need detailed specifications, which may simply not be available. And if you decide to do some reverse engineering, then ... having the original hardware handy might help.

Absolutely true.

[TransEurope]

If you really want to programm in assembler and want to learn how computers work, buy an old C64 and the the Data Becker C64 Bible, or an old Amiga at Ebay. If you want to to the same on a modern iP4-machine, you'll give up faster than a SETI@home-package is analyzied ;)

Catweasel!

[mkro]

The Catweasel is a PCI floppy controller (among other things), and boasts support for over 1100 disk formats. I plan to start backing up my old Amiga and C64 disks with this one "any day now".

Data Legacy

[Anonymous Coward]

I just gave a speech to a bunch of progressive groups in Kentucky Saturday that included a screed on data loss. Twenty two years after starting a lawsuit on fair taxation and coal reserves, for example, the suit finally made it through the courts. My question was: how good a job are we doing preserving the records and data for those cases that take 30 or 50 years, like tobacco or asbestos. I'm looking ahead to the lawsuits on global warming.

If you want to see the talk:
http://www.hollowground.net/tecactv

wh

Rubbish

[onion2k]

There's certainly good reason to keep old data readers about the place.. I once spent a very dull weekend with a cassette->parallel interface loading some old ZX Spectrum code onto a pc and encoding the files into .z80 format. But there's no good reason at all to keep the rest of the hardware around. Every system before about 1995 has been emulated on faster, more stable modern system that afford us things like memory save points, video output recording, and other pleasentries.

Old hardware is dead.

Re:Rubbish

[pandrijeczko]

It's actually copyright that's the problem here.

Unfortunately, unless you have been given express permission by the author/owner of the software to distribute the program freely, then the only way you can keep a copy of the program is to have the original intact also.

Personally, I believe that if there's no chance of a piece of software making the owner any more earnings then it should be released into the public domain automatically, say after 15 years or so. (Incidentally, I'm not necessarily talking about source code also, just the program).

At least then the genuine people who want to preserve the old games and software can do so openly without fear of legal action.

Retro Links

[hedgehog2097]

I'm surprised the article didn't link to old-computers.com:
http://www.old-computers.com/news/default.asp

Plenty of "Replica"-esque machines on mini-itx. The best two are probably
http://mini-itx.com/projects/bbcitxb/
http://mini-itx.com/projects/sx64/

Testify

[PakProtector]

When I was working at the local Humane Society, I saved a Apple Mac II/ci from the dumpster. It had been donated to the thrift store and was thrown away because it was 'too old' to interest anyone.

I like playing certain old games, mainly because if a game is done right, it doesn't matter how outdated the graphics get -- Classics never change.

There's just something you get out of playing the Zork Trilogy on the old hardware that you don't get on the new stuff.

Don't throw out those old tapes!

[Jay Maynard]

You'd be amazed at what we've got running under Hercules...there's a lot of computing history being lost because people threw away old round tapes, thinking "Oh, we'll never run THAT again". A guy used an emulator to rescue old census data from Africa (was the story reported here? It wasn't that long ago), and that kind of thing will be only seen more as time goes on.

If you know of old IBM mainframe software on tape, drop me a note; chances are I can recover it. I've got 9-track and 3480 cartridge tape drives on a PC just for that purpose.

Media Degradation Is The Issue

[pandrijeczko]

It's not so much that data is held in an "old" format, it's more that the media that it's stored on like tapes and floppy disks of varying shapes & sizes will degrade much quicker than, say, optical media.

The BBC here in the UK did a radio program about getting music and video from old recordings and vinyl, even old 78 RPMs. The problem, once you've got the data off, is how you store it on a media that won't degrade over time. Even CDs are thought to have a limited lifespan of possibly only up to 100 years.

The only practical solution for "permanent" data storage currently are huge RAID hard disk arrays where you can replace a drive as it goes faulty.

Re:Media Degradation Is The Issue

[vaceituno]

CD are not as durable as many think. Check this article for some wake up.

http://www.rense.com/general52/themythofthe100year .htm

From the article:

"But an investigation by a Dutch personal computer magazine, PC Active, has shown that some CD-Rs are unreadable in as little as two years, because the dyes in the CD's recording layer fade."

Re:Media Degradation Is The Issue

[pla]

The problem, once you've got the data off, is how you store it on a media that won't degrade over time.

Simple - Redundant serial copies. Unlike analog, digital copies don't lose anything from generation to generation.

I use DVDs for backups, but don't actually "trust" them to work, only as a last-resort fallback. I keep my old files by keeping them on live systems.

My first HDD held 10MB. My second held 40MB - So I just copied the entire contents of the first over. My next drive held 340MB, again, just copied the entire 40 to it, complete with the final state of the 10MB drive. Then a 1.2GB, same process again.

Now my home file server holds over half a TB (though I'll soon need to add a bit more space to it). I had started to worry about not having a good complete backup of that (I have 90% of it backed to DVD, but like I said, I'd rather not need to actually depend on that)... Until I recently upgraded my SO's desktop machine. Poof, threw in a 400GB drive, she needs about 20GB, and has a complete mirror of all my files up to March of this year.

I see no reason for that trend not to continue... The original media (the floppies from which I loaded files onto the 10MB drive) have long since vanished, and even the fifth generation of the above sequence (the 1.2GB drive) has vanished into the landfill. Yet I still have all the files I would need to run a vintage XT clone with MS-DOS 3.3, neatly filed away with no fewer than three redundant copies still in existance.


So I see the problem of how to store something "forever" as a bit of a red herring - We don't need any particular medium that lasts forever, only to last a few years and then we can make another new copy of it.

Universal Format

[ThosLives]

Interesting quote from the summary: "countless computer files stored in outmoded formats" led me to an interesting train of thought I've been mulling around for a while, somewhat affected by me recent reading of G.E.B.

The universal format for documentation, I believe, is the printed hard-copy document. Think of it this way: If we received the Rosetta Stone, or bits of the Torah or Quran, on some electronic media, would we have been able to get the content off - especially if it was encrypted somehow?

I think the only universal format is the printed page, which requires no "special equipment" to read (it might not be interpretable, but it can easily be recognised as a document) whereas a computer-recorded pile of numbers, while perhaps recognisable has having meaningful content, will probably, in the future, have no context in which to extract its meaning. Consider this: you receive some piece of hardware in the future which you realise stores binary data. Is it numbers? Is it a program? Is it sample data from atmospheric noise collection? All you know is there is binary data. All you know is there is binary data, and you don't even know if it is stored in 8-bit blocks, 16-bit blocks, 3 bit-blocks, or whatever. You don't know if it's in ASCII or some weird encoding of, say, Farsi. You might try running some statistical analysis on it to see if it's some kind of language, but against what do you compare the 'glyphs' of the numbers? When you see a stone like the rosetta stone, it's obvious what you've got; when you've got a list of numbers, there is no way to tell what it is other than a list of numbers.

This is a great danger of the digital age, in my opinion, and it is good that there is still expertise floating around about the "old" equipment. But remember, the "old" equipment is still less than a century old: what will happen in 100 more years? 400? I have this nagging concern that data integrity of digital media will not last the thousands of years that printed material lasted for future generations. I think this is why I really don't like the idea of digitising the libraries, or even digitising photography.

Definitely something to consider for all those folks concerned with "the best data format" and if .DOC or .PDF or XML or whatever is better.

The best format is one that contains enough information to clue the interpreters how to interpret it rather than relying on something else. Right now, all digital documents are merely a string of numbers, and a string of numbers is not sufficient to contain meaning to interpret itself - those numbers rely on some interpreter to receive meaning (as an excersise to prove this, take any file on your computer and look at it in a debugger - on various systems, a hex-editor, and a program that will use the contents of any file as raw image or audio data. It might not be rendered sensibly (I don't know that I'd want to listen to the "song" that, say, Firefox would be), but there is no effective way to tell if the string of numbers has meaning by using trial and error.

A printed document unequivocally has more information than this - a schemaatic diagram is different than a picture of an apple is different than a poem... and while we may not know 'apple' or the language of the poem or have the capability to understand the diagram, we know that those things aren't, say, a random paint splatter.

So, again, while I applaud the efforts of these guys for writing down their knowledge, if they don't do it in a "universal" format, who will be around to interpret their blogs and digital records in 1000 years?

"Retro-Machines": Good learning tools

[ndogg]

Old machines are good learning tools, even if only on paper, although they were easier to work upon in my electronics class.

Hardware concepts haven't significantly changed over the years. What has changed, significantly, is that everything has become smaller. Once the basics are understood through learning of these old machines, the more complex concepts of more modern machines can be more easily understood. Good Computer Architecture classes will start off on the hardware of these old machines first, and build off those concepts as the class moves into understanding newer machines.

Odd disk formats, etc.

[JoeCommodore]

One of my side hobbies seems to be converting PET documents to text files or PET disk/tapes to emulator friendly images.

Tapes are relatively easy as the 64 can read most of the, the hard part is that sone disk formats are hard to come by, the Commodore PET has several different format drives, the most popular are the 4040/2031 which a Commodore 64 can read, but the 512k single sided 8050 and double sided 8250/SFD-1001 disks are another matter both using quad density drives (nowhere related to the PC HD format) and GCR encoded to increase capacity. These drives (unless you are a hardware whiz) communicate exclusively using IEEE-488 so A PET/CBM or B128 are best employed.

I myself use the PC-to-pet interface the C2N232 with related software to get the files fron the PET to the PC, from there it's a matter of some home spun chipmunk BASIC programs to get the files tidyed up and in ASCII.

To be consistently successful at it you have to not only have the tools but knowledge of the various disk and file formats and system quirks that you are dealing with, which will help you get around the unexpected.

I've had requests to help convert 64 related software, but have passed on that as I am not into real time programming work (some sort of lighting program on a cartridge) but there are others up to that challenge.

Same goes for other platforms like old 400k Mac disks which use a varialble speed drive and can only be read IIRC on a 68k mac using System 6 or lower. There are also the protected disks or those that were recorded with utilities to improve speed or capacity (which makes the disks/tapes differ from any knwn standard format). Not everything can be done with an emulator.

Re:Commodore...

[blakespot]

I recently put together a C64 system. I was an Apple II guy in the early days, but wanted to have some time w/ a C= 8-bit.

http://www.blakespot.com/list/images/c64c_1.jpg
http://www.blakespot.com/list/images/c64c_2.jpg
http://www.blakespot.com/list/images/c64c_mobo.jpg

Have a C-One going as well:

http://homepage.mac.com/blakespot/PhotoAlbum24.htm l

Good stuff. Other machines as well:

http://www.blakespot.com/list

blakespot

Re:Buy a Model-T to learn about combustion

[VAXcat]

Heh? You could learn far more about internal combustion engines using a Model-T rather than a modern car...in like wise, you would also learn far more about aviation with a DC-3 than a modern jet...In both cases, the engineering is much more accessable and the devices are much closer to the physics involved than their modern counterparts. On the DC-3, for example, you would learn a great deal more about adverse yaw and use of the rudder to deal with it than you would on a modern plane with sophisticated flight controls. Navigating using a sectional chart, dead reckoning, piotage amd VORs on the DC-3 will teach you much more about navigation than just punching a destination in on a GPS. The Model-T, with its simple carburetor and Kettering cycle ignition, all very exposed, will teach you far more about physics than attempting to work under the hood of a new car, with its fuel injection and computer controlled spark. In general, the first few generations of something are much more useful for learning than the products produced after many decades of engineering have overlaid the first principles.