@blacklight The knowledge and attention to details in today onboard software systems is equivalent to that of those days, I assure you 🛰️

@blacklight Only 69Kb of RAM? So I'm guessing you never ran VisiCalc on a 48Kb Apple II …

@blacklight @somcak The skills and knowledge are still there, but they're being blocked to save a bit on expenses.

@blacklight this. There was a brief period of time when knowledge, discovery and the advancement of science was the goal. It was replaced by profit, greed and the enrichment of a few.
Engineers and scientists still have the same level of commitment to detail and perfection, but they are mostly prevented by greedy, bastard #captilalists that don't see further than their own selfish interests.

@blacklight
Great toot! The Voyager was truly an amazing feat!
And not to be picky, but the Pantheon is Ancient Greek. 8^)

@blacklight V-Ger? Maybe watch Star Trek the Motion Picture again. Scary!

@blacklight Well said! There was an engineering elegance back then - excruciating detail simply implemented in ways that minimized potential catastrophic points of failure.

<siiiiigh>

@blacklight I laugh every time I remember ”eight megabytes and constantly swapping” was supposed to be a terrible burn against Emacs. Eight. Megabytes.

@blacklight While I agree that Voyager is a marvelous example of code, I disagree that we lost something over time. 😉

1. In those days there are also created a bloated or buggy software, but today we don't remember it. Using the comparison to masonry, because it collapsed. 😀

2. It is hard to compare two completely different types of software. If desktop software would be created in the same way as airspace one, it would be a few thousand times more expensive and slower to deliver.

@blacklight "If the young but knew..." My first computer had a whopping 4K (yes, 4 Kilobytes) of main memory. Today I'm forced (occasionally) to deal with a circle (jerk) of Agile coders who are powerless when the IDE on their MacBook Pros won't launch, with absolutely no idea where to start in fixing the problem; the file servers, the datastores, their Macbook, the network, or a hundred other things they know nothing about. Oh, and trying to teach them about IPv6 is basically pointless.

@blacklight my iphone 5s that can barely start up now has about the equivalent flops of a cray 1 supercomputer that was used to simulate nuclear warhead yields in the late 80s

@blacklight computational constraints are a source of wonderful and clever creativity. Our current abundance is sometimes a trap.

@blacklight some time in the 90's I sent an email to the Voyager team asking some technical questions about the systems. I got a detailed answer back answering in detail my questions.

I wish I still had that email. Would be interesting to look up who it actually was that sent it. The team wasn't big even then as far as I know.

@blacklight

we ran a $20 million/year in sales (a lot then) business in 1967 with an 8kb computer the size of 2 home freezers. We leased it from UNIVAC for $1400/mth on a 5-year lease. This would be $12,789/mth today or $767,340 total for 60 months

economy of code was key to success for all but basic tasks, and there were a few legendary tricks (a very cool hack in one case) I learned from others.

optimizing code was in such cases cheaper than scaling up capacity

@blacklight 69k? Not even enough for the windows boot manager?

@blacklight i have a to do list app that is 50 meg and charges an annual subscription.

@blacklight It starts with "If I had a few more bytes, I could add feature X" moves through "we need to build garbage collection and memory checking into the system to make it safe" and leads to "we have so much capacity now, we don't need to worry about memory usage or instruction counts."

@blacklight one thing that analogy is missing though is that maybe the medieval engineer looking at the Parthenon has been tasked with building a shop for the local shoemaker. He doesn't have the resources or time available to a wealthy Greek city state.

We should admire the consummate skill with which the engineers of NASA past did their work. But they weren't building software for the masses. It's important to keep that in mind as well.

@blacklight forward thinking... missing in today's software engineers

@blacklight The Voyagers are amazing, but trust me, that attention to detail is alive and well in some places!

@blacklight
These engineers and scientists who did this should be honored....

@blacklight

What is especially sad is the way older programmers are less often hired.

I worked with some brilliant older programmers in the past, many of them came from other fields such as math or physics. They would likely not get past resume screening today.

@blacklight , it is still possible to enjoy the skills it takes to master old systems by coding for a Commodore 64. 64kb ram, simple 6502 instruction set and a nice set of chips like the SID, VIC-II and IO. It certainly puts today's terribly inefficient systems into perspective.

@blacklight

No skills have been lost.

@blacklight Medieval times gave us all kinds of cool inventions like the printing press and eyeglasses.

It's easy to look back at stuff like that and miss what's made possible when engineers are able to spend their energy on the useful part instead of on optimizing. I'm happy to let Firefox and Chrome use a half gig at rest for what I can do with them.

FWIW, VSCode is only at ~200MB right now with tens of documents open. It's a good trade for me.

@blacklight … 69 KB OF RAM!! 🤯🤯🤯🤯🤯🤯

@blacklight @aburka I’d hate to be the person who sent the wrong commands and (hopefully temporarily) silenced humanity’s farthest object.

@blacklight I worked briefly at Ames Research Center for a NASA hardware engineer who had hand-built discrete circuitry on those early probes. Amazing experience.

@blacklight The machine on Voyager was purpose designed for one job only. Its hardware and software were designed together by experts for a single purpose, and it works wonderfully. A PC is a general purpose machine designed to do thousands of different tasks well. A PC can run a simulator of Voyager or other space probe, but Voyager can’t play Doom. @june

@blacklight The same is true in many branches of engineering. Aerospace: we're still using the essentials of Lockheed-Martin's design work from the 50s and 60s, and having personally gone through the redesign of the Nimitz class aircraft carrier's engine room, the same was true there as well. Redundancy, structural integrity, safety factors - these are all the places where modern engineering has "trimmed the fat", not always with the best operational or safety related results.

@blacklight The software engineering on Voyager was amazing, for sure; that said, the software on missions like Juno and New Horizons is just as robust and far more capable—because it can take advantage of much better computing hardware.

@blacklight @Gerhardf 69K is loads of space - to an embedded engineer. We look at the waste of desktop PCs and weep.

@blacklight You are absolutely correct.

@blacklight

@blacklight it's because memory used to cost more than labour, now labour costs more than memory.

In the case of applications like intellij, labour is always more expensive than someone else's memory.

@blacklight so much has fallen prey to a glut of resources on modern machines.

@blacklight Programmers have absolutely thrown away almost every hardware speed and capacity advancement made in the last 50 years.

@blacklight sometimes i think they are not lost, but only taught in certain compartments. when something really really has to work

@blacklight "Memory is cheap!" "Just use this giant bloated library for one single feature!" "Install this package manager for one single package!"

@blacklight Well, at least nowaday's software still knows how to phone home.

@blacklight That's a far cry from the "move fast and break things" ethos of the modern tech-bro.

@blacklight
There was something to be said about how limited resources to work with helped hone your skills. Whether it was memory or “this task has to complete in X ms”.

@blacklight The difference between programming and engineering.

@blacklight

It's making do with the capability of one's tools.

It was cutting edge satellite at the time

Such skills are developed and honed with practice

Apprentice-ship is broken

@markmccaughrean this thread is for you 😁

@blacklight

"And they implemented a self-adjusting mechanism that a few times a year scans the positions of a few known stars to infer the position of the earth, and point back the antenna in the right direction."

Simply amazing. I never fail to be in complete awe of the Voyager program.

@blacklight
The Golden Years of Analog.

My dad worked on telemetry for the Lunar Module. Everything there was life-critical, so which meant they had humans to recover when systems lost their way. And they had a shorter mission span.

Such discussions always remind me of this song.

https://youtu.be/WOmoURHDgVM

@blacklight Yeah. I remember the day we were terribly proud of having found a saving of four bytes in a completely full 256 byte ROM ... which one of my colleagues instantly took advantage of to add five new features to the software.

@blacklight one way to reclaim the speed and space of software bloat is to rewrite as close to the silicon as possible. I feel it is WAY easier than people think. You don't need fonts, you don't need browsers, you don't need anything but C, sockets, and xcb because we can no longer get to the frame buffer reliably. The bar for rewriting stuff from scratch is pretty low. Some of us have been doing this all our programming lives. I refuse to use bloated layers, see my pinned stuff

@blacklight this is so true.

@blacklight I don't think the knowledge is lost. It's just not worth using it for deskop computer applications. A gigabyte of RAM is the same price as an engineer working for a few minutes. Even if you multiply it by the number of people running the app, it will not compensate the costs. So, no one will be paid to save that gigaby.te of RAM.

Also, there is some survivor bias here. What about all the space probes that failed because of stupid mistakes?

@blacklight I've long held the same feelings as you regarding the design and making of the Voyager spacecraft twins. Wonderful is really an apt description of the machines and their journeys.

@blacklight So true!

@blacklight @jrconlin you're also discovering the diferrence between single-purpose and very general purpose computing

@blacklight
It sounds like Bill "640KB should be enough for anyone" Gates was off by about 10x.

@blacklight I bring Voyager up a *lot* in ops discussions at work.

@blacklight I feel Margaret Hamilton judging me from behind her giant stack of printed code every time I need to reflash firmware because I forgot a minus sign or when I can't fit my LED blinking function on a microcontroller with multiple megabytes of flash

@blacklight perfect example of limitation breeding innovation

@blacklight

Yes and no. Such abilities have been lost to mainstream business, because terminal stage capitalism can't find uses for such skills, but they remain in existence outside of it.

@blacklight

But consider:

Huygens landed on Titan all by itself -- mission control found out later if it worked, after setting it up.

Likewise the Pluto high-speed near encounter by New Horizons. Took months to send back what it collected.

Perseverence, Ingenuity, and Curiosity are AI robots navigating alien terrain with minimal human interaction.

All these machines ALSO have low-level failsafes of the type you describe. That skill never went away.

@blacklight No. I'm calling this out. It wasn't lost. It takes TIME and a MANDATE.

They made it right because they were give a huge amount of time to do very little. Even despite constraints we still do marvelously well when given the time to do those things.

If you want software like this? Stop pushing for new things and start polishing what we have to a MIRROR finish. Slow down and do it, but in that case don't expect a new iPhone every year.

Past people weren't smarter they had more *time*

@blacklight we feel like we lived through the changes, yeah. it's... upsetting.

@blacklight
Can I put in a word for Clive Sinclair (RIP) when every byte counted for something.

@blacklight Good thing to keep in mind next time one reaches for a bloated OS or framework to solve a problem that can be solved better and more reliably with a simple small program running bare metal on a microcontroller. Every byte and line of code is one that can hide a bug. Very often, less is actually more and better.

@blacklight My father-in-law was an engineer who worked on the Apollo lunar excursion module. He died shortly before the 50th anniversary of the first lunar landing. At lunch after his funeral, some of his engineer friends who also worked on Apollo were telling some stories. Truly amazing work was done with the technology restrictions of those times.

@blacklight

"Make do."

@blacklight I agree with your point - and we're even in a worse place with the web taking over the app space - even more fragile and needs a browser that's eating half my memory...

BUT, it does have to be noted, the Voyager software is absolutely single purpose, very tightly crafted, minimalist, engineered into the ground at high cost and sold to exactly two customers (Vgr1 and Vgr2) and no one else.

So comparing it to wide audience tools running on a general purpose OS is a little unfair.

@blacklight My Dad was one of these men. I thank you for what you wrote. He, and his buddies at JPL were brilliant and deserve the accolades. What WE need to do as their descendants is save the planet with smart minds and tight programming to cool oceans and scrub the atmosphere. My Dad went on to teach climate sci and help invent neural networks. We had so much HOPE and he believed he was saving the planet. I wish he had been able to.

@blacklight I am very much not a techie, so treat with bucket of salt, but it has always *seemed* to me that programs these days - for whatever - are MASSIVE and unwieldy. I'm led to believe they don't have to be.

I'm also led to believe, by bits here and there that I've scanned over, that this is a *culture* rather than a necessity. People, insist on goldbricking (adding pages and pages of crap because they can and think it looks good) rather than building in simple redundancy.

@blacklight IMO this is a bad take. We still have very high reliability microcontrollers and people who know how to program them, and we make use of them all the time in all kinds of situations: pacemakers, MRI machines, elevator controllers, autopilot in planes, brake assist in cars, autoswitching the electrical grid, etc. But working that way is expensive and slow so we use other tools most of the time. You are pining for a past that is still right here!

@blacklight We built things, on earth, like switchgear, designed to function properly for 100 years. Key to this is use of well know components, whose failure rates are know, and they are designed to fail gracefully if/when they do.

@blacklight I want to believe with all my heart that we can still be this clever! If we had people who were really in love with it, they’d feel that sense that what they were doing was so much bigger than anything else going on here on this rock, and pluck more of this amazing out of the ether. :)

@blacklight depends on who is paying the bills. Most businesses spend as little as possible on everything including software. This has led to a entire generation of lazy developers and project managers.

@blacklight
I imagine someone entering commands centuries from now.

@blacklight I remember in 1992 someone explaining to me they built a hotel reservation database interface with less than 64kB. I said, "that's impossible. You need 2MB and a 386 to run Windows reliably." I was 17, and I didn't realize how different things were 10 years earlier or how short 10 years seems when you're older.

@blacklight It's a major plot point in Asimov's Foundation series: the Galactic Empire had so much raw materials available, everything they built was gigantic and inefficient. On Terminus, resources were scarce and expensive, so the Foundation technologists learned to miniaturize.

@blacklight @JustChapman
The little, tiny, baby amount of logic I struggled through with the aid of integrated circuits in my summer digital electronics class convinces me you're absolutely right.

That and a few run-ins with Turing machines have left me with a sense of awe at the intellects who created the field.

@blacklight eh, just run vi in a terminal. Or ed if you are feeling crazy.

@blacklight V-ger seeks the creator

@blacklight We used to design things for people, now we design things for money.

@blacklight one of my longest blogposts (almost 1000 words) is on how Goldratt and ToC can be used to understand why we don’t write software “better” now with more resources. I compare today with “Production of large computer programs” by Benington https://chocolatedrivendevelopment.com/2021/04/08/unnecessary-limitations-with-a-history/

@blacklight I don't think the skills are necessarily gone, but the motivation for doing that surely is. There is little incentive to be performant or use little memory in a FAANG like company and even less in a start-up environment.

@blacklight I'm not sure it's "lost". There are still some very very brilliant people writing software out there. But back in 1977, only the most brilliant people had the means to write code. Nowdays... well, you know how this goes.

On the one hand, average code quality has suffered. Obviously 😠

On the other hand, it's nice to give normal people opportunities. Because the secret is - those "normal" people - those mundane 99.9% bell-curvers - they're you and me. We get to write code.

@blacklight Sigh. Such a good post until you started shitting on IntelliJ. I'm so sick of this bloat whining with bad comarisons.

@blacklight I think we can still produce this kind of code/achievement, by redefining the standards. Today, devs are always in rush and some keep tell people to increase the resources’ size instead of optimizing the code. The engineers of NASA might think that their code must run for decades on really small computers, so optimization is the key

@blacklight At the points in my career where I've hired software developers having a background (historic or current) in the demoscene has always meant reaching the interview stage precisely because of this reason.

An understanding of what the machine is actually doing rather than just being able to use some high level APIs.

@blacklight this does nothing but illustrate my notion that we should strive for some sort of digital de-scaling, at least at the software level.

@blacklight I believe it depends on the industry. In space and flight, the attention to detail is still there just as it used to be. In other industries, there's good and bad quality products, but the bad quality products just didn't survive until now.

@blacklight It doesn't help that our industry repeatedly falls in love with the new thing, and then considers everything the old thing did bad, until years later, relearning the lessons the old thing already learned.

@blacklight I can't help but think, that many projects nowadays would be better off in deep space but without the option of re-establishing communications with Earth.

@blacklight As someone from that era, it was necessity. We had too little space to store code to run in too little RAM. So you rolled your own operating system. I did that to boot a supercomputer. I did that to cram more DAT code than would fit in CMOS to run in 16K with a paging operating system. I've calculated every tick of system clocks to ensure code finished in time for the next fetch. It's rocket science, sure. But really old rocket science on crappy hardware with cold solder joints.

@blacklight this kind of story makes me smile

@blacklight There's plenty of low level expertise, not only in embedded systems, but also at the hobbyist level of coding old home micros or new virtual systems (i.e. the Pico 8).

The Voyager systems are truly incredible, but had a huge amount of engineering effort companies are not prepared to expend on system software.

@blacklight same sentiments here https://youtu.be/ZSRHeXYDLko

@blacklight @raimue They have not been lost. It's just that nobody is willing to pay for them any more. Nowadays, we would just build a new probe if we lost the old one. ("But boss, it would take decades until it reached the location again" -- "Who is the engineer, you or me? Here's a 10M budget, go invent teleportation")

@blacklight

While this has definitely occurred in software, I think the real victim of "eh, just throw some more computing power/storage at the problem" is the field of mechanical engineering.

So many ingenious design techniques have fallen out of favor because it's easier to throw some silicon at the problem.

@blacklight there's several order of magnitude more detail available now than then. Not justifying the ram hungry nature of software, but suggesting that most of this apparent laziness is a by product of increase in complexity and a resultant reduction in ability.

@blacklight A natural result of learning to program on IMSAIs and Apple IIs.