Hey, let's talk about a curated archive

[Al Kossow]

If not that what the he!! was AI made for?

Pump and dump schemes.

 

[cjs]

Plenty of 8-bit machines in the 80s....

Yes, exactly. Those machines were around before the time of people born in the '80s. Unless your proposing that people somehow feel nostalgia for machines that were around only when they were toddlers, before they could even read and write.

Nostalgia is not the only factor in this hobby, but it is significant.

Do you have any data for this, or did you just make this up?

I've asked around amongst my retrocomputing group, and only a minority of us own (or are interested in buying) the computers we used as a child, and even amongst that group, we're usually much more interested in different computers, with the one we owned mostly sitting on a shelf rather than getting any use or development attention.

I'm not sure why you think someone who doesn't care about old computers when they are 20 would suddenly start caring when they are 40.

Well, I know it can happen, since it did for me. Do you have data to show that these kinds of changes in interests are particularly unusual?

I think this is what AI was made for. Just grab everything and pump it into a giant ingestor and then you can ask it questions all day long.

Well, it rather depends on if you want correct answers to your questions, or you're happy with whatever the AI "hallucinates."

If not that what the he!! was AI made for?

Very generally, because large software companies, particularly SaaS companies, need the promise of massive growth to keep those stock prices going up, and they're running out of things that are actually useful. (And, in some cases, destroying the usefulness of what they have—see Google search.) Ed Zitron goes into all the details (probably more than you want) in "The Other Bubble" and various other articles.

Even though I was a real gearhead when I was a kid (broke gearhead, unfortunately) I was never interested in any vehicle made before I was born.

Nor, when I was young, was I interested in computers made before I was born. That came much later.

 

[hunterjwizzard]

I'm plenty interested in computers from before I was born. Unfortunately I either turned down or kept but later junked all the ones I've come buy affordable. If I find a nice one for a good price you can bet I'll be all over it.

 

[Plasma]

Yes, exactly. Those machines were around before the time of people born in the '80s. Unless your proposing that people somehow feel nostalgia for machines that were around only when they were toddlers, before they could even read and write.


Do you have any data for this, or did you just make this up?

I've asked around amongst my retrocomputing group, and only a minority of us own (or are interested in buying) the computers we used as a child, and even amongst that group, we're usually much more interested in different computers, with the one we owned mostly sitting on a shelf rather than getting any use or development attention.

Surely you understand that nostalgia doesn't just mean "I owned one as a child." I'm not going to beat this dead horse further.

 

[pbirkel@gmail.com]

This thread evolved into something less useful than intended (IMHO).

What would be useful is a set of meta-data to be added to files or whole directories which can help finding stuff. A small set of data like brand, model name, year, type or record (photo, disk image documentation, ...) in a more or less standard format.
.....

Just so. Having worked professionally in an unrelated field tacking the same sort of problem I can completely relate to (1) the importance of capturing good metadata that gives insight into the content of the item-to-be-indexed (and typically the best metadata can't be automatically extracted from whatever item-content is immediately accessible) from the perspective of search-users, and (2) the laboriousness of actually capturing _quality_ metadata. Tossing whatever item-content is immediately accessible into an algorithm-engine (or AI engine for that matter) isn't going to give good results -- and tossing compiled software code into such an engine is just going to be a waste of time, IMO. Extracting text from images of disk labels and/or packaging materials is probably a better path. If those materials still exist. Someone(s) knowledgeable in the domain who can bring contextual information and experience into play is necessary, IMO. Getting the right subject matter experts to do the necessary work is hard; for free is even harder. Doing that work in a distributed manner adds additional complexity, particularly WRT ensuring metadata-assignment consistency. Assembling and motivating a dedicated team will be challenging ...

But if the job is worth doing at all, then we ought to do it right. Unfortunately, as the saying goes, "There is never enough time to do it right, but there is always enough time to do it over.” (attributed to John W. Bergman)

You could use the softlist files in MAME as a starting point for the schema.

Guidelines for Software Lists — MAME Documentation 0.271 documentation

Thanks for pointing this out Al. Very interesting!

 

[cjs]

Surely you understand that nostalgia doesn't just mean "I owned one as a child."

Sure. It has another meaning which is "a bittersweet yearning for the things of the past," regardless of their age. But if you intended that meaning, why do you go on about it apparently applying to older people and not younger? By that meaning, it doesn't matter if it's an '80s or a '60s or a '00s computer.

I'm not going to beat this dead horse further.

Yes, I think that, given you seem to have no evidence whatsoever to support what you're saying, you've certainly beaten it more than enough already.

 

[cjs]

...and (2) the laboriousness of actually capturing _quality_ metadata. Tossing whatever item-content is immediately accessible into an algorithm-engine (or AI engine for that matter) isn't going to give good results...

No, those results are useless for determining, e.g., when something was published. One needs contextual clues to narrow down a date range for something not explicitly dated and, even for something with explicit dates, the dates are not necessarily the ones you're looking for. (For example, the date on the publisher page in most of my Japanese books seems to be the date of that printing, not the date of publication, unless it happens to be the first printing.)

Someone(s) knowledgeable in the domain who can bring contextual information and experience into play is necessary, IMO. Getting the right subject matter experts to do the necessary work is hard; for free is even harder. Doing that work in a distributed manner adds additional complexity, particularly WRT ensuring metadata-assignment consistency. Assembling and motivating a dedicated team will be challenging ...

What we need is something slightly in the style of a wiki (i.e., accessible to and editable by all), but that allows people to suggest things like publication date for an item and explain why they think that's the correct date, keeping all of the suggestions and explanations available rather than forcing to just one. That allows others to go in later, link things together (e.g., 1st printing to 3rd printing of a book) and clean up the data.

 

[Zare]

For voodoo games I use my actual Voodoo3 system. I've tried those games with the nGlide driver and they look pretty much identical to how they played on my 433mhz celeron with software rendering back in 1998.

Cool, but not what I was aiming for.

nGlide to modern DirectX is a far more stable interface than old DirectX to new DirectX because of all the differences in underlying graphics card drivers and hardware generations. DirectX is not a fat transparent API, it serves to give hardware access to GPUs too, meaning things are not perfectly back-compatible if chips and drivers themselves aren't.

In practice I experience this with NFS5, 25 year old DirectX on modern machine fails, there are unofficial patches, but you can get nasty gamebreaking bugs with loading textures on maps that are hours into the singleplayer mode...and that's a crap situation to find yourself in when doing a playthrough. But with Voodoo mode in game, and nGlide, these problems go away.

So nGlide is quite useful for games in that age range...end of Voodoo but still ancient DirectX enough to have problems in some cases.

 

[Al Kossow]

Extracting text from images of disk labels and/or packaging materials is probably a better path. If those materials still exist. Someone(s) knowledgeable in the domain who can bring contextual information and experience into play is necessary, IMO. Getting the right subject matter experts to do the necessary work is hard; for free is even harder. Doing that work in a distributed manner adds additional complexity, particularly WRT ensuring metadata-assignment consistency. Assembling and motivating a dedicated team will be challenging ...

And this is why I've been trying to get the community involved, because there are people here with deep knowledge in specific domains that aren't getting what they know archived.
Twenty years ago, doing good resolution images of media, packaging and documentation was not as practical as it is today. Camera quality is amazing and storage and bandwidth
costs makes preserving a lot more than we used to practical. OCR and media file analysis/extraction allows you to build proper search indexes with attribution to their original source material.

NIST has the software reference library https://www.nist.gov/itl/ssd/software-quality-group/national-software-reference-library-nsrl which lets you determine the validity and de-dup software
distribution files.

Pictures of the disks imaged help to establish the provenance of their contents. Check-summing from known reference sources helps determine overwritten or modified content.

Then, there are virus infected disks to deal with.

 

[Plasma]

Sure. It has another meaning which is "a bittersweet yearning for the things of the past," regardless of their age. But if you intended that meaning, why do you go on about it apparently applying to older people and not younger? By that meaning, it doesn't matter if it's an '80s or a '60s or a '00s computer.

Not "regardless of age." Do you have nostalgia for covered wagons and muskets?

Yes, I think that, given you seem to have no evidence whatsoever to support what you're saying, you've certainly beaten it more than enough already.

There is plenty of evidence in this thread alone, but you'd rather engage in bad faith arguments with veiled insults. Thanks for reminding me why I ignore your posts.

 

[DougM]

It's kind of interesting looking at people's response to the AI question here. AI is like the cybertruck of the software industry. When people hate it they really hate it. But a few people love it.

I do believe there's a ton of uses for AI. Many more than can be dismissed out of hand. Is it a bubble? of course. All these things are cyclical. First the hype, then the excitement, then the massive crash, then the productive use cases. Something like that. The only difference being that these days any new technology is turned to evil before its beneficial uses become mainstream.

But in the end I do believe that AI will have several if not many incredible use cases. Already I use it to write my code. I am the sole proprietor of my projects, which means I do the electronics, I do the mechanical, I do the software and I do the design. As a result of this I don't code very often - it's my least favorite part and I am *really* rusty. So being able to ask an AI to code something up for me as a starting point is immeasurably useful.

And being able to interact with it in realtime is very impressive.

This will expand to many aspects of life away from the computer - some for the better, some for the worse. I'm about as interested in a chatbot as I am in a musket but I keep seeing areas where AI could make my life so much easier.

A bit of morning rambling, thanks for riding along

 

[cjs]

Not "regardless of age." Do you have nostalgia for covered wagons and muskets?

No, of course not. I don't have nostalgia for anything I don't remember using myself. You're the one making claims that kids born in the 80s are "nostalgic" for computers that disappeared before they could read or write, and I'm not clear why you're doing this, or, apparently, what you mean by "nostalgia." It's clearly something distant enough from the mainstream meanings that you should explain yourself.

There is plenty of evidence in this thread alone, but you'd rather engage in bad faith arguments with veiled insults.

There is not. I've been through the thread carefully and outside of you, I mainly see people claiming (when they touch on the topic at all) that nostalgia is not the main driver of interest in retrocomputing. So let's look at what you have to say about it.

Well, the first thing to do is load it up with the software you remember. Nostalgia and whatnot.

Here it sounds like you're saying that nostalgia is about things you remember. Yet elsewhere you say nostalgia covers things that the vast majority of people wouldn't remember. You need to clarify here.

Nostalgia is not the only factor in this hobby, but it is significant. This is not unique to vintage computing; for example classic cars are on a similar trajectory. Yes, there will always be some younger people interested, but it's trending downward over time.

Again, completely unsubstantiated facts. What I see, based on prices on Yahoo Auctions and other site, is that interest in vintage computing is trending way up over the last ten years. What I see in user groups is that a lot of these people, particularly those in the 30s and 40s are interested in computers for which they cannot possibly be nostalgic, because they never knew these computers existed until long after their childhood.

Now this is of course anecdotal evidence, and thus should be taken with a grain of salt, but it's still more evidence than you've provided in your pronouncements. If you feel insulted because I call you out for making claims with no support whatsoever, well, I'm sorry you feel bad about being called out for that, but there is a way to keep that from happening: just point out when you're giving opinion that you have no idea is fact or not.

Teens in the 80s and 90s were not pining for muscle cars, they were buying them. Because they were dirt cheap. Those are the same guys who are nostalgic for these cars today.

Well, teens in the '80s and '90s certainly were not buying cheap 8-bit computers in droves, as far as a I can tell.

 

[hunterjwizzard]

First the hype, then the excitement, then the massive crash, then the productive use cases. Something like that. The only difference being that these days any new technology is turned to evil before its beneficial uses become mainstream.

This is indeed why AI is so reviled by many. Because its this big, hip, trendy thing, everyone wants to use it as an all-solving hammer when it really, really isn't.

And unfortunately that's actually blocking legitimate uses for the technology. We're going through this at work right now. The decision-makers want to throw AI at every problem, and the guys like me down in the trenches know its not going to do jack for most of those problems. So some of my coworkers respond by assuming AI won't work for any problem, and the company is going to end up wasting a fortune developing tools no one will ever use.

 

[segaloco]

Going back to the topic at hand, the curated part of this quandary is ringing more and more in my mind. I think one of the barriers there is the fact that curation is a big personal investment, and regarding stuff like this of dubious providence and copyright, I could see individuals not wanting to paint too large of a target on their backs by formally taking a prominent curation role. Granted, we see several successful examples of archival operations out there that are not drawing unwanted scrutiny, but it's still a risk, especially for those who might be in the positions in this world to wrangle together more of these arcane materials.

On the flip side, the more hands you have providing metadata, content, etc. the more wildly the quality of curation is going to vary, so it really makes me ask myself what effective community curation of important historical ephemera looks like, how that can be done in a way that doesn't put anyone in some sort of painful legal jeopardy when their intentions are clear, and in a way that they know their hard work is going to last and be respected and effective. This is where even the smallest amount of engagement from the commercial sphere would be helpful, even if it's just clarifying stances by corporations saying "Any software developed prior to XXXX or targeting platform ABCD is relicensed under terms BlahBlahBlah". This happens occasionally, but not as much as one would hope.

The relicensing of UNIX by Caldera is a good example of how this sort of thing might shake out. Caldera didn't have the sources to actually provide release specimens of the various versions of research UNIX, but their licensing changes clarified what rights archivists and researchers have with regards to the things they had squirreled away all those years. Now a number of iterations of research UNIX and BSD are out there in the open in a way folks can actually materially understand their rights. On the other hand if IBM or whoever came down saying no, thou shalt not in any way do anything even remotely resembling distributing IBSYS, even if it is decades old, that would give those with trepidation some certainty that okay, that one is off limits, so I feel more secure in working with my other stuff because I have been given guardrails.

However, at some point or another, the folks who are actually making these decisions are so far removed from the folks who were involved in the creation of these systems and technology back in the day that the fact that they still hold unilateral sway over these things is a bit insulting. They have about as much to do with it as any one of us does, they just have that company's name at the top of their paychecks. It's a ship of Theseus situation for me, is there any meaning in a copyright exerted by a company in 2024 named IBM regarding an operating system released by a materially different company also called IBM in 1960 when that entity IBM in 2024 bears so little material commonality with the environment in 1960? Sure, they share lineage, but I share lineage with monkeys, I'm not living in the jungle picking bugs out of my neighbor's hair. No offense to IBM or any curiosities surrounding IBSYS by the way, simply a (possibly inaccurate) example for the sake of demonstration.

Pardon the length, I thoughts on this specific matter of historical preservation going back a couple decades...

 

[Chuck(G)]

AI is a tool, like many others, such as web search engines--or like the little girl:

There was a little girl,
Who had a little curl,
Right in the middle of her forehead.
When she was good,
She was very very good,
But when she was bad she was horrid.

Use it, but verify. That second part can often consume just as much time as doing things "the hard way".

One thing that irritates the hell out of me is the prevalence of AI-implemented customer support portals. "No, you @#$%^! ijit, that's not what I'm asking about!"

The other day I received a junk call where, after the synthetic "Hello", a bot-voice asked "Are you a bot?" repeatedly. AIs in search of other AIs?

 

[hunterjwizzard]

I could see individuals not wanting to paint too large of a target on their backs by formally taking a prominent curation role.

My current (not particularly)"brilliant" idea is to take the archive offline. Distribute it via physical media.

For example, once the archive is created, it could be put on SD cards in flashy, neat-looking cases. These could be sold at festivals, traded, mailed back and forth. There are obviously many downsides and doing it this way is completely unnecessary in the modern era.

But the upside is - its highly unlikely to attract anyone's attention.

 

[Al Kossow]

I could see individuals not wanting to paint too large of a target on their backs by formally taking a prominent curation role.

If you don't have pointers to where the software actually is, I don't see the problem.
People have been generating annotated catalogs of programs for a long time.
The only list I could imagine that could run afoul of the DMCA are lists of copy-protected programs if those were in some
way a circumvention of copy protection. What if you just cataloged the image hashes? That is exactly what the NIST
software reference library is.

 

[Al Kossow]

My current (not particularly)"brilliant" idea is to take the archive offline. Distribute it via physical media.

For example, once the archive is created, it could be put on SD cards in flashy, neat-looking cases. These could be sold at festivals, traded, mailed back and forth. There are obviously many downsides and doing it this way is completely unnecessary in the modern era.

But the upside is - its highly unlikely to attract anyone's attention.

That assumes it is cast in stone. any collection changes (grows, is revised) over time
I can't imagine there will ever be The Archive

 

[RetroAND]

My five cents about archiving software:

• All pieces of software should point to the CPU they are targetted. For instance, I made a dump for the AS/400 C-10 panel, which is targetted at MCS-48.
• Whenever possible vinculate it to the entity/es who made it.
• Related to the previous, I think it would help an onomastic index.
• If possible, vinculate the piece of software to some place. For instance, for a Soviet game, the Soviet Union.
• Related to the previous, maybe add a toponymic index.
• Vinculate all related entities together. For instance different ports of the same game between platforms.

I think most of those things would be covered by the ISAD-G standard, but unfortunately I think it is not yet prepared for this specific kind of documents, at least at the level of detail required. However, with some extensions, it could work.

 

[RetroAND]

My current (not particularly)"brilliant" idea is to take the archive offline. Distribute it via physical media.

For example, once the archive is created, it could be put on SD cards in flashy, neat-looking cases. These could be sold at festivals, traded, mailed back and forth. There are obviously many downsides and doing it this way is completely unnecessary in the modern era.

But the upside is - its highly unlikely to attract anyone's attention.

How many hard disk drives would you like to share in the meetings?