Ask HN: Is retrocomputing the pragmatist power user's future of the desktop?
I feel like I have recently reached an inflection point where I have turned from being generally optimistic about the future of computing on the desktop to being generally pessimistic.
For example, I have Macs from about 10-15 years ago in a box somewhere, together with a few thousand dollars worth of professional-grade commercial software that I used to run (MS Office, Adobe Creative Suite, Sibelius, ...), all on DVD with license keys and everything.
I increasingly feel that dusting off that stuff and reintegrating it into my daily driver desktop computing environment might be a good use of my time and money. I feel like it might, in fact, be better bang for the buck than subscribing to today's software and keeping the hardware and my knowledge of the software updated as the software updates roll in.
On HN, one can frequently read about how the industry is forcing changes on everyone that not everyone wants to go along with. The first reaction is usually "Open source on the desktop, now more than ever". But, looking at it from a pragmatist's point of view, I don't expect the Linux-on-the-desktop experience will, at any point in the next 10 years, match the commercial desktop experience that existed 10 years ago.
As an alternative to "Open source on the desktop, now more than ever", how do you feel about "Retrocomputing, now more than ever"?
22 comments
[ 4.6 ms ] story [ 66.3 ms ] threadComputing is transforming from enabling tools to a means of social control and enforcement.
The market has completely turned around from demand/desire driven to a "push" market, where we are told what we are expected to own and use. I call this consumer communism.
Cybersecurity is a lost cause. We just can't fix it. And it's getting rapidly worse. The more layers of security we add (whose security?) the less desirable and usable the products become.
Stability can only be found in products that are ossified, hard wired and off-network.
Software Engineering is basically broken at the consumer level. Exploding complexity has accompanied deskilling and a reckless "move fast and break things" attitude that means working products are no longer shipped (everything happens in the maintenance phase as a long-tail SasS). The "product" itself is just bait.
Many products are just about cultivating dependency, lock-in and removing features to squeeze users.
When a non-trivial number of users see a 1990s Amiga as a better choice for serious work than the latest silicon, something has gone quite wrong.
The problem is that 80s/90s technology will eventually be unservicable. What can we do to take back tech and ensure a supply of stuff that we actually want for the future?
FOSS is doing amazingly well.
I also think open hardware will probably have it's day eventually. The IoT needs a bit of a push to stop being a pile of SASS poop, and we especially need a free alternative to things like Tile trackers before the EU kills them or something.
What missing elements do you see in free software that we should be trying to patch?
The hard work is still to be done on the legal and cultural levels.
I don't want to use free software as a marginalised, "not supported" exception and constant "inconvenience" to others who have been deceived or bullied into domesticated monopoly silos.
I want a right to choose (or build for myself) software that fits my needs, practical, ethical, technical, financial and cultural, and to have those choices work at least as well as for a big-tech consumer for most common tasks in a modern digital society.
Surely the real action is now on the protocol and interoperability level? Things have essentially not improved since the late 80s when you couldn't load a DOS file on a Mac, except now the barriers are on the network. In this regard I am pleased to live in Europe which I think is doing good things to promote software freedom and diversity - though I hear you have reservations. Do you think the dangers (in Europe) are that we simply replace corporations with governments as enemies of freedom? If so, how can a free market (rather than top-down legislation) address the seemingly insoluble problem of forcing giants to open and interoperate?
But I don't like quite that level of extreme stance on privacy. I do believe it's absolutely a right, but I don't want it to be a duty.
Not all personal data is a top level military secret, and there are many legitimate uses(Look at HAM radio) for even plaintext systems.
I think I'd rather they just require honest privacy policies in a simple machine readable format with an accompanying human readable version.
Something needs to be done, because it's definitely not right to lie to consumers, but I also don't want it to be any harder to log data for debugging and attack mitigation, to use CCTV, or to use technologies that are decentralized and inherently don't support revoking rights to process.
I mostly like the GDPR, but there's a few things I don't quite like, and the possibility of even more strict stuff is a bit scary.
I'd hate to see AirTags, CCTV, and Amazon Sidewalk outright banned with no chance of even a FOSS version, and I personally know people who would love that to happen.
Here in the US most states forbid one person from secretly recording a conversation that they are part of without specific exceptions, and even that has some scary side effects, because a hidden recorder is a powerful self defense tool.
Other than that, I'm a fan of 95% of what I've seen from the EU on tech.
It's probably not enough to really fix things though.
Data portability doesn't fully solve things when so much is in the cloud. Sure you can move to some other provider, but it's gonna be a lot of work, and then you'll just have a different single point of failure.
I think what we really need is for a really good open protocol to really take over.
Something like WebDav, but extended to support synced redundant accounts, domainless self hosting(With pubkey/DHT based domains that don't require asking a registrar's permission to make one), and so on.
It's just way too convenient to make stuff sync in the cloud, without it, FOSS doesn't have full parity. We need something like SOLID, making that cloud sync a separate reusable thing and letting us choose our providers, while making it easy to add to apps, and not involving crypto tokens or 7729 different accounts on a million custom sync systems, some of which may be insecure.
I work with a lot of purists who run Linux on their desktops. I can totally relate, and I have a lot admiration for how they have their principles and are willing to put in the work to uphold them. I frequently find myself in video conferences where someone can't get their audio to work, and, predictably, it's always one of those guys.
I can equally-well relate to the frustration they must be feeling when that happens, and I remember it well from when I ran Linux on the desktop.
I installed my first Linux on desktop when I was 12 years old (that's 25 years ago) and even developed some OSS for it a few years after that. I used Linux on desktop as a daily driver for work-related stuff ca 2005 - 2012 when I was an academic. And I do periodically try out new distros like elementary OS and so forth. I've been running Linux Mint on a laptop I use for trying out nonessential things. (It recently had to give way to QubesOS).
The purist in me has always been at war with the pragmatist in me.
When OS X and later WSL came out and gave you the option of running a Linux-like userland on top of a commercial-grade desktop, this established a truce. The obvious thing for me to do was to use OS X or Windows, and then install whatever software I wanted with no regard to whether it was open source or not, as long as the price tag was worth it for me. A price of zero for open source frequently leads to a great cost/benefit ratio, but with some things like InDesign, the price tag is just worth it for me for the functionality. And when things fall short of the standard of "it just works" then the true cost can quickly skyrocket, and I've often found this to be the case for open source.
With regard to the software that I most deeply care about because it's most immediately connected to the software that I myself develop, it has always been clear that I wanted to stick with Linux and OSS. But we're talking terminal-stuff here, not GUIs, since I'm a software engineer who deals mostly with data and pure abstraction (like math, NLP, etc).
It seems like recently the industry has turned a corner that is upsetting that truce. OS X and Windows, in terms of the corporate politics behind them, have just become too bad to bear, and I am expecting they will get even worse.
So this is what is creating the motivation in me to look to the past of commercial software rather than its future, or the future of open source, for my daily computing needs.
I finally put linux on my gaming laptop this weekend, because microsoft decided to push linkedin resume builder in my face when I was updating my resume in word.
Windows 11 is now bursting with flashy distractions that make it hard to get work done.
Screw that noise. Pop!OS worked with the nvidia graphics so I went with that.
Me too, Pure nostalgia. Just like to log in sometimes and party like its 1989. :)
It's not... it just seems like it because our current way of using computers seems to make sense, when actually it's insane.
If we thought of money the same was as compute, image going to buy bread at the store. You make your selection, then hand over your wallet, and sign a power of attorney, have it notarized, and hand it to the clerk. This clerk has been checked to me a good clerk by both your AntiBadClerk service. You get your wallet back, and take your purchase home, confident it has had no bad side effects, and the correct amount has been taken from your wallet.
---
Or... you could take $5 from your wallet, and hand it to the cashier, very confident there can't be any larger side effect than the loss of the $5.
---
We use the principle of least privilege when we take money from our wallet to pay in a transaction. The largest side effect possible in a normal transaction is not getting the correct change back.
---
We use the principle of "don't trust the user, blindly trust the software with the user's rights" with our programs on computers. We hope that our firewalls and antivirus have stopped anything bad from happening. Unfortunately the halting problem is impossible to solve, and you can't enumerate bad software, or limit side effects using this model.
---
Every time I bring this up here, objections are made that it's too complicated, wouldn't actually work, etc... Fine, I expect those, and will try to answer them, like always.
My main point is, least privilege works. WASM seems to be holding up fine, and it uses it. Virtual Machines and Containers are a way too course model of it.
Cybersecurity CAN be solved. Please don't give up hope.
Thank god there are people like you who still have the courage to think this way.
My comment "Cybersecurity is doooomed" is actually a little more nuanced. Here's a few of the issues, as I see it, in a longer form [1].
Pragmatically, what we now have is a culture aggressively opposed to the USERS's security [2] under the various veils of "convenience", "necessary efficiency" and "security". Yes, included in these "three-little-pigs of broken systems" faux "security" is also key mechanism for undermining _actual_ security).
To this end, grotesque negligence and incompetence are rewarded in the name of shipping products, making money, satiating power and exalting convenience uber-alles.
To take your analogy of money. Quite literally the opposite is being pushed under the guise of "cashless commerce". Instead of taking $5 from your wallet, you wave a contactless card at an unknown computer and hope it's "temporary power of attorney over your bank account" works out fairly.
Yes, cybersecurity can exist. But I don't think it can exist in the digital ecosystem that corporations and governments want to push onto society.
[1] http://techrights.org/2021/11/29/teaching-cybersecurity/
[2] I believe, and teach my students, that in the adversarial 21st century context of unfettered state security and surveillance capitalism security is a "fixed sum game". For example, the vendor's security is the user's insecurity, and so on. Any question of security must be qualified with three further questions:
Security for who?
Security from who?
Security to what end?
Security for the owner of the computer. A tool for that user, the person you daily uses it in their lives. Not some corporation or government.
Security from everyone who is not the owner, and especially NOT any code, from anyone one, anywhere. Users should NEVER have to blindly trust code. They should ALWAYS have a way to play with code, and know it won't have any unwanted side effects, ever.
Security to what end? It's a tool, man. Think of the care free days of CP/M and MS-DOS, when you could get shareware disks full of nifty stuff to try out, and you could write programs to do anything you wanted, and share them with the world. You write-protected your OS, and could try anything, and see how you liked it, in safety. That power, in the hands of the people, is the end I seek.
I wasn’t even alive at the time most of the machines I have were manufactured. But I can understand them. I cannot understand modern computers for the life of me and I’ve tried.
Nowadays computing feels like a heap of black boxes designed to extract as much as they can from you.
I have spent probably a total of $10 or less on software in the last decade, and I don't pirate.
The last time I used commercial software was a copy of Eagle I requested at work, but I don't use that for anything personal these days, I use LibrePCB(Not perfect but still very much good enough).
https://www.riscosopen.org/ https://www.haiku-os.org/
Both descendants of former commercial OSes and - at least in my opinion - way less bloated than the modern Linux Desktop.
I personally use RiscOS as my daily driver and so far i am very statisfied.
If you don't host internet accessible services locally, and don't offend any nation states or billionaires, it's reasonably possible to run quite old software behind a NAT box without grief. Know that something might come along and infect your machine, plan for it, make and test backups, and you should be good to go.
Tools should last for a lifetime, and the computer (and it's software) is just a tool. It's time to re-normalize buying something for life.
I opined during the whole installation process: something is very wrong with the modern computing industry when a user feels the need to wind the clock back 10 years (and go to the substantial trouble of doing so) to obtain needed functionality when there is no legal alternative. (i.e. I licensed CS6 in 2012, I was used to paying 600 or so dollars every few years for an upgrade, but I'm done with paying another cent for "renting" creative cloud at it's ridiculous ever-increasing price and CPU absorbing DRMware client. Adobe had me paying since CS3 - and for eight years as a CC customer). Capitalism would dictate that there should be a viable alternative, and yet, there is not.
I also found myself thinking, what have we gained in the last 10 years? Telemetry? Desktop advertisements? New icons? Loss of backward compatibility. Mavericks - to me - seems perfectly usable even 10 years later, however I don't dare to connect it to the internet for lack of security updates.
So... I too wonder if the future of the desktop is retro.
It sounds jaded when I say that if this "rolling back" becomes a mainstream thing, surely big tech will come up with a derogatory term for it: "The Dr. Who's" who somehow undermine progress.
I can't help feeling that we're experiencing multiple business monopolies simultaneously: because in Adobe's case - there was tremendous pushback against their subscription model... all to no avail. Their vendor-lock-in is as strong as it gets.
Adobe's behaviour is also enabled by Apple and Microsoft in this case, newer versions of the OS's do not support CS6, and to use newer versions of Photoshop you'll need a newer version of the operating system.
So what do we get now as locked-in photoshop users? Creative cloud ever increasing fees, Adobe CC software that wastes computing resources, topped off with the privilege of Microsoft or Apple telemetry. Back in the day it used to be called spyware.
How are we moving in the right direction as an industry?
I am completely sick of it.
But in general I concur - old macOS + old Photoshop has a lot to recommend it.
You don't have to give this up on Mavericks! :D
https://github.com/blueboxd/chromium-legacy
---
Mavericks also has something better than dark mode—a visual design that uses the full tonal range of lights and darks together to have higher contrast. A lot of the rest is only useful if you're in Apple's ecosystem, which I'm mostly not.
MS Office is just reskinning of older versions, so yeah an older version is fine, you can do okay if you are a music maker too I suppose. But when it comes to say video editing, you do need the new stuff.
A perfect use case for old hardware and software today is banks, some in my country still use XP and IE 6. It just works, and it is not connected to the public internet so there is no issue of security.
https://www.clockworkpi.com/product-page/devterm-kit-a06-ser...
b) The industry departed from original methods/beliefs/practices/teachings and is something else today, something fake/bad
c) The media is bombarding normal people with false beliefs and false images of relationships, looks, talking, thinking and also programming (anybody can become a web dev)
d) Internet is over saturated, over populated which directly leads to piling loads and loads of e-junk
e) More (RAM, faster speeds) doesn't mean it's better
f) Algorithms don't get old, they don't decay over time. (Algorithm for gausian blur from 2002 isn't old compared to 2022 one. Algorithm is in perfect condition as it was in 2002, it doesn't rot or decay. The form can/is improved to fit today's capabilities. So you may call it less improved, slower, less efficient, more costly)
g) If you were to write a 30 page textbook (no images/few small images are supplied), would it _really matter_ if you were writing it with: (MS Office 2022 on 16GB RAM i9) or (Office 2003 on 32bit XP with 3GB of RAM) or (LibreOffice Writer on Raspberry) or (some Macintosh) or (<insert some low end old tech here>). It would not matter. The result would be the same on each system.
Just some thoughts worth exploring. Maybe it all comes down on privacy/tracking and personal preference.