Still Andrew (I guess) is a user on retro.social. You can follow them or interact with them if you have an account anywhere in the fediverse.

This is a thread about computers.

I have a lot to say.

I might not be able to finish right now.

I'm going to post unlisted.

I make no secret of the fact that I love old computers, and that I think modern computers have lost their way in terms of providing utility to users. To that end, I write about, and think about, computers and the way to move computers forward, without losing site of the fact that computers should serve their users. I grew up with hand-me-down computers, from Atari to Apple to Dell, and in the process I got to experience a sizable portion of computer history very quickly, in my teen years.

This left me with Opinions.

I write about things that are informed by these opinions often. When I talk about building a World Wide Web analog without the internet, about reviving the BBS or when I lament the fact that Gopher was doomed to obscurity by the modern web, it is in response to my experiences with an array of computers from the dawn of the home computer revolution up through to the modern age. There was a time when computers were magical.

I had come, in recent months, to suspect that I might just be an old fuddy-duddy. I'm approaching 30 years old, and I had begun to feel like I was looking at modern computers and modern software through the lens of someone who was being left behind, shouting at the sky, shaking my fists at the kids on my lawn. I was coming to the conclusion that my love of these computers of my childhood, and of ones that I had never had the chance to interact with, was some kind of rose tinted nostalgia.

I had not fully subscribed to this theory, but it seemed more likely that I was romanticizing something that was actually Not Great that it was that nearly every modern software and hardware platform had inexplicably gotten shittier.

I am now prepaired to say, with conviction, that every modern hardware and software platform has gotten shittier, and that it's not inexplicable. I'm going to try to explain how I came to this conclusion, and give some potential explainations.

First, let me lay out a little bit about my technical literacy and my profession, this might help explain some of the assertions that I'm going to make. I started using computers, and the internet, in 1995. Our family computer, my first computer, ran Windows 3.11 (for workgroups). Later, in the late 90s, I was given an Atari 400 and reams of books and magazines on basic, followed shortly by an Apple II GS and dozens of disks of software.

Later still, I started collecting computer detritus, and assembling frankenstiend linux boxes, and emulating some of the the machines I read about in the magazines I had as a kid.

I loved computers. I loved making weird little programs, and silly applications and games. I'd build things in GW Basic or Freebasic, and distribute it to my friends on floppy disks. Even in the latter half of the 00s, I was passing half broken games around on floppy disks (or collections on CD-Rs, when I could talk someone in to buying some for me.) Computers were, by and large, ubiquitous in my life. Nearly everyone had an old one they didn't want, and a new one they didn't understand.

For a teenager and an aspiring computer programmer, the 00s were a great time to learn.

I collected cast offs from neighbors, from thrift stores, from office upgrades. I rebuilt them, installed useful or fun software on them, and sold them or gave them away. All of my friends had computers of their own, because I had access to these machines, and I cared enough to outfit them with the appropriate software.

(It must be said, at this point, that 'useful' and 'appropriate' are relative terms. In 2009 I gave a good friend a computer that had been built for Windows 98. It was running Puppy Linux from a CD, and saving data to a USB flash drive over USB 1.1. It did word processing, email, and basic web browsing. It had a whopping 64MB of RAM, and was covered in glitter, googley eyes, and carpet samples. But it was free, and it wasn't useless, and that was important.)

I went to school to become a programmer, and discovered that I don't enjoy programming as it exits today. I understand it well enough, and I *can* do it, but I don't *want* to. I make websites, and I build tools to help other people use computers.

I make my living as a systems administrator and support engineer. (and I'm looking for a new gig, if you're hiring.) That's a fancy way of saying that I solve people's computer problems. Professionally, I'm responsible for identifying and mitigating the shortcomings of various computer systems.

Guess what?
There are a lot of these shortcomings. Like, a lot. More than I ever expected.

Some of these shortcomings are legitimate bugs. Some of them are bafflingly short sighted or poorly considered architectural decisions. Just as many are cases of a divergence between the needs of the user and the abilities of a program. Modern programs are often feature incomplete, poorly supported, and difficult or impossible to customize. Modern computers are often slow, and cranky. I'm responsible for handling the fallout of this unfortunate situation.

I've seen how revolutionary a computer can be, if it is designed with the needs of the user in mind, and how disastrous the same can be when it is not. I've seen computers used to empower people, and used to oppress. I've seen computers be Good, and the consequences of when they are not.

So that's who I am, and my experience with computers so far. Those are my credentials, and my qualifications.

Before we go any further, let's talk about The Computer Chronicles.

The Computer Chronicles was a TV show that ran from the early 80s through the early 00s. Over it's nearly 20 year run, The Computer Chronicles covered nearly every facet of the newly developing Computer industry. It was hosted by people with Opinions.

The guests were, frequently, people who were proud of the things they made, or the software they represented.

Watching the developer of CP/M and DR DOS talk to a mainframe engineer who worked at IBM in the 50s about the future of computers as seen from the 1980s was eye opening.

On the one hand, this show serves as an excellent introduction to, or reminder of, the capabilities of computers 35 years ago. It helps us see how far we've come in terms of miniaturization, while also demonstrating again that, in many ways, there is nothing new under the sun.

Before the advent of the internet, reporters were writing their stories on laptops and sending them in over phone lines, 25 years before the release of the iphone HP released a computer with a touchscreen, three years before microsoft released he first version of windows Apple and Visicorp demontrated GUIs wih features that Windows wouldn't be able to approach for another 9+ years.

And, of course, I'm reminded again of Douglas Engelbart's 1968 "Mother of all Demos", in which he demonstrated the mouse, the GUI, instant messaging, networked gaming, and basically every other important development of the following 50 years.

It took 5 years for Xerox to refine and miniturize Engelbart's ideas to the point that they thought they could market them, and another 10 years before Apple refined and further miniturizaed the same ideas, and brought us the Mac.

Nothing is ever new.

The whole video of Engelbart's Online System (NLS) is available on youtube. Some of it is *really* interesting. Most of it is unfortunately dry. It's easy to forget that this was 50 years ago, and also mindblowing that it was only 50 years ago.

Anyway, back to Computer Chronicles. In an episode about Word Proccessors, the man they were interviewing said "There's a lot of talk about making people more computer literate. I'd rather make computers more people literate." There's a phrase that resonated with me in a big way.

It sounds like the kind of semantic buzzword shuffling so common in standard corporate speak, but I got the impression that the guy that said it, believed it. He believed that computers had gotten powerful enough that they no longer had to be inscrutable.

There were others working around the same time on similar ideas, or at least from a similar philosophy. Working to make computers, if not intuitive, at least comprehensible. I think this is a noble goal.

The computer is often thought of as a tool, but it is more like a tool shed, in which we store a collection of tools, a source of power, and a workspace.

The tools of the 60s and 70s were primitive, partially because of the limited space and limited power our toolbox could provide for them, but also because our ideas and understanding of how these tools should work were limited by the audience who was using the tools.

That is to say, in the 60s and 70s, computers were weak and slow and computer users were also computer programmers. A small, tight knit circle of developers and computer scientists were responsible for the bulk of the progress made in that time, and the idea of designing tools for non-technical users was never considered.

Computer culture had, by and large, a kind of elitism about it as a result of the expense and education required to really spend much time with a computer. This changed, slowly, starting in the mid 70s with the development of the Microcomputer Market and CP/M.

Computers became more affordable, slowly. Affordable computers became more powerful, quickly. Within 10 years, non-technical users were interacting with computers on a daily basis. It was against the beginnings of this backdrop that the phrase I mentioned earlier was coined. "Human Literate Computers" or "Human Centered Computing."

Ease of Use was the holy grail for a lot of computer companies. A computer that was so easy to use that they could sell it to grandma. But, to me at least, Human Literate and Easy to Use are distinct ideas. Many modern applications are Easy to Use. Netflix is Easy to Use. Facebook is, for all it's faults, pretty easy to use. The iPhone, the iPad, and ChromeOS are super easy to use.

Well, they are easy to use as long as you use them in the prescribed way. As long as you let them tell you what you want to do, instead of the other way around.

That, IMO, is the distinction.

I think that many of the steps towards demystifying the computer of the 80s and 90s did good work, but ultimately, the computer industry left the whole idea behind, in favor of making some tasks Very Easy while making other tasks Practically Impossible, and turning everything into a surveillance device.

When I was a kid I was brought up with computers that showed you how they worked.

You booted in to a command prompt or a programming language, or you could get to one, if you wanted to.

I got to play with GW Basic and qBasic and, a little, with hypercard.

I got to take apart software and put it back together and make things that made people happy.

I got to make things that I needed. I got to make things that make me happy.

Today, the tools to do that are complex to compensate for the vast additional capabilities of a modern computer, but also to reinforce technical elitism.

I often wonder why Hypercard had to die.

It was because Jobs wanted the Computer to be an Appliance. A thing only used in prescribed ways.

Letting people build their own tools means letting people control their own destiny.

If I can make what I want, or if someone else can make what they want, and then I can take it apart and improve it, why would I pay for an upgrade? Why would I pay you to build something that doesn't meet my needs?

I'm mentioning hypercard specifically because I've been relearning hypercard recently, and it is *better* and more useful than I remember it being.

It's honestly revelatory.

Hypercard, if your unfamiliar, is powerpoint + instructions.

Here's a great introduction/example: loper-os.org/?p=568

The author walks you through building a calculator app in about 5 minutes, step by step.

Warning: There's a bit of ableist language tossed around in the last paragraph. Skip it, there's nothing worth reading there anyway.

You use the same kinds of tools you would use to build a slideshow, but you couple them with links, multimedia, and scripting.

Want a visual interface for your database of client data? Great! slap together a roladex card, and drop in a search function.

Go from concept to presentation ready in an hour or two (or less, if you've done this before!)

Hypercard was easy to use. Everyone who used it loved it. It was integral to many businesses daily operations.

Jobs killed it because he couldn't control it.

Microsoft doesn't ship any tools for building programs with their OS anymore, either.

They used to. There was a time when you could sit down at any windows or DOS machine and code up a program that would run on any other Windows or DOS machine.

But we can't have that anymore.

In the name of Ease of Use, they left out the Human aspect.

Use your computer how you're told to use it, and everything is easy.

Do anything new or novel and it's a struggle.

My nephew has an ipad.

He asked his dad how to write games. His dad didn't know. His dad asked me how to write games on an iPad. I told him not to bother.

My nephew asked me how to learn to write games.

I gave him a raspberry pi and a copy of pico 8.

Now he writes computer games.

He couldn't do that on his iPad.

Hypercard would be a perfect fit for the iPad and iPhone.

Imagine it!

Imagine the things you could build.

But we aren't allowed to have computers that are fun to use, that are easy to build for, that are human centric, or human literate.

The last 10 years of development in computers were a mistake. Maybe longer.

Instead of making computers Do More, or making them Feel Faster, we've chased benchmarks, made them more reliant on remote servers, and made them less generally useful. We brought back the digital serfdom of the mainframe.

In the first episode of computer chronicles (youtube.com/watch?v=wpXnqBfgvP) the mainframe guy is real adamant about how mainframes are good and micros are bad.

The host, a microcomputer legend, disagrees pretty strongly.

Later, when they talk about the future of networking, the mainframe guy talks about it as a return to mainframes. The micro guy talks about BBSs, peer to peer networks.

The mainframe guys are winning.

(this is not to say that I think mainframes are bad. I don't. Mainframes can be really good and interesting! Plato was wonderful, as were some of the early unix mainframes.

But IBM style Mainframe culture is The Computer as a thing you Use but don't Control culture, and I am very against that.)

I have to step away for a while. I'll continue this later.

@ajroach42 I want to respond, elaborate, & discuss at length here. I spent about 10 months some years ago immersed in the computing literature around the history of debuggers, during which I went from EDSAC to Visual Studio, but also all the other half-dead ends ends of computing history such as, e.g., Lisp machines.

Naturally, I came out of it a Common Lisper, and also naturally, with Opinions about modern computing.

Up for the discussion? It could get wordy and over a few days. :)

@pnathan for sure.

I haven’t gotten in to lisp machines yet, but I’m always down for discussion.

@ajroach42 @pnathan
This thread is going to be gold :)
(I'm replying here so that I won't forget about it...)

@ciaby @pnathan I hope you enjoy! I'm looking forward to the discussion as well.

@ajroach42 @ciaby
OK, so, I'm about a decade older than you, Andrew: I taught myself QBasic in the mid 90s, got online late 90s, never really looked back.

First, I want to say this: older computer systems - considered as systems - were generally more capable.

But to be clear, they were limited in use for those who didn't take an interest in learning them. I'm talking about things that weren't Windows 3.1+.

@ajroach42 @ciaby This was the Great Debate that was largely won by Microsoft. "Everyone can 'use' a computer.". That is to say, everyone can operate the appliance with preinstalled software. *everyone*. Apple pioneered the notion, but it turns out to be the preferred mode for businesses, who really rather don't like having specialized experts.

@ajroach42 @ciaby It is my contention that Windows (& *nix) computer systems are designed to be administrated and managed by sysadmins, and the user experience in this case is great.

When you have sysadmins, there are no driver problems. There are no printer problems. There are no problems, as a matter of fact: it's all been taken care of by the admins.

This is exactly how executives like it.

Apple does the same, with their iPhone.

Apple is the sysadmin, metaphorically.

@pnathan @ciaby This is a good point, but I think it deserves scrutiny.

I am employed as a support engineer and a sysadmin, and I still run in to driver issues, printer issues, etc.

I take care of them, eventually, when I can.

But, even after doing this for 10 years, I still encounter problems that I can't solve (because there isn't a solution.)

but the metaphor of Apple as sysadmin, I'll accept. I disagree with someone else admining my phone, but that's another issue.

@ajroach42 @ciaby your users pay you so they don't have to care about sysadmin issues. their world is great!

@ajroach42 @ciaby I'm glossing over the 1% failures to get at the core point: sysadmins are designed into the windows and unix world so users can focus on their core competency.

@ajroach42 @ciaby

Hi, I'm probably near the age of @pnathan, and while I'm not a lisper anymore (ages went from my emacs fluency) I agree with all he said.

To give some context, I'm a polyglot programmer currently working on a brand new operating system jehanne.io

Now, the assumption that you seem to share is that people cannot learn how to program. I used to think this too.
Now however I realized that it's like we were scribas of Ancient Egypt arguing that people cannot write.

@Shamar @pnathan @ciaby I never said people can't learn to program.

I'm saying that some people don't want to learn to program, and that what we call "programming" is needlessly difficult for some tasks, in the name of corporate profits.

@Shamar @pnathan @ciaby I feel like you think this was a clever point, but I don't understand what you mean.

Programming is a specialty, and some people have other specialties. Expecting them to also become expert programmers because our current expert programmers can't be arsed to make extensible and understandable tools is unreasonable.

@ajroach42 @ciaby @pnathan

This is the assumption I challenge.

For sure programing is a speciality right now.

But it's a speciality just like reading, writing and counting.

Not everybody can be a novelist, nor a professional mathematician.

But people should be able to program, just like they are able to read, write, compute a volume, reason about an average speed, a length...

Programming is harder then math because we are still using primitive tools.

It's sad that we are happy with them.

Still Andrew (I guess) @ajroach42

@Shamar @pnathan @ciaby

Some kinds of programming (just like some kinds of math) will remain hard.

But better tools are what I'm after, yeah.

· Web · 0 · 2

@ajroach42 @ciaby @pnathan

We need a lot of #research.

We should #hack more.

And we need better #math too.

It will take some centuries.

Much more, if each generation keep being satisfied with the shit it slightly improve (or messup, as we did with the #web when we give out #XHTML for #Javascript).

Because, to me, the tool we need are as different from today mainstream tech as our writing system is from Egyptian #hieroglyphs.

@Shamar
I feel that it's not only a matter of research, but also a point of throwing away some tech that we take for granted (x86, for example) and rebuild from scratch with different assumptions in mind. In the current economic system I find it quite hard to do...
@pnathan @ajroach42

@ciaby @pnathan @ajroach42

You are largerly undervaluating the #power of #technology.

Most programmers are not well versed about #history, and it's a pity. There's a lot to learn for us, from history.

Technology is probably the most powerful and effective way to change the world. Most changes in human organizations have been allowed by technological innovations: from fire to boats, from bronze to iron, through argricolture, writing, counting, roads, from sword to guns...

Technology can ....

@ciaby @pnathan @ajroach42

Technology can change the world for the better or for the worse.

It can disrupt "the current economic system".

And that's why #capitalists are in a hurry to keep #hackers under control.

So, I don't think that the "current economic system" should be a problem for hackers.

We CAN throw away the web.

I really think it (I work with browser all the day, I know the stack pretty well...).

From scratch, with the lessons learned, it will take a fraction of what it took.

@Shamar @ajroach42 @pnathan
That's possible, and somehow is already happening.
What I'm talking about, however, goes much deeper than that. I'm talking about open hardware infrastructures, where every component is documented, there are no binary blobs or proprietary firmware .
Very important is also the instruction set, because what we have now (x86/amd64) is incredibly bloated and full of backward compatibility shit.
RISC-V is a step in the right direction. If only the hardware wasn't so expensive... ;)

@ciaby @ajroach42 @pnathan

Interesting point. You're right.

We cannot actually trust the hardware, either.

But... I'm a software guy.
I'm more concerned about the way we connect and use the hardware than the hardware itself.

Indeed, everything I've had to learn about hardware while developing my x86_64 OS has been a pain.

So, YES!

We need more research on the open hardware too.

There are only 2 things that I'd like to preserve keep: little endiannes and 64bit longs.

Please. 😇

@Shamar @ciaby @ajroach42 Bold claim: open source or non-open source hardware doesn't matter when deployed at scale.

the essential problems today are, in a sense, all software, mediated by the scale.

@Shamar @ciaby @ajroach42 that said:

we have these *inter-twingled* issues: the hardware is manky, the software is manky, and the incentives to improve are perverse.

nymag.com/selectall/2018/04/da

My reckoning is that there is a space today for a sort of New System, a Unicorn OS, where the whole thing is largely rebuilt. Does the web have to exist? does tcp/ip? are there better systems?

here we see we make choices and one prioritizes those who take the time to learn the system and one ...doesn't

@pnathan @ciaby @ajroach42

What is Unicorn OS?

Oh... you cited Oberon some toot ago.

I like it a lot (but I have to admit that I've never tried it on real hardware).

I love that Wirth still work on Oberon-07 language, and that he keep REMOVING features.

Oberon inspired Pike for Plan 9 UI. I started from Plan 9, and frankly I'm not brave enough to throw away TCP/IP as Wirth did.

Still... the Wirth's approach (hack, hack, hack, challenge all assumptions, keep it simple!) is what we need.

@Shamar @ciaby @ajroach42 UnicronOS : the magic OS that we're talking about that solves the problem.

with a sparkling dash of rainbow over it, because, you know it's magic. :)(

@pnathan @ciaby @ajroach42

Oh... the most funny definition of vaporware I've ever seen!

UnicornOS: the first #vaporware with a #rainbow! 🤣

(Disclaimer for any actual developer of an OS called Unicorn: I'm just kidding... the joke was simply too good... sorry)

@pnathan @ciaby @ajroach42

I do not know actually.
I literally know nothing about #hardware.

But my insight is that probably, #cheep #OpenHardware and #simple #distributed #software #systems could change the world.

I have a dream: one low power mail server in every house.
End to end mail encryption everywhere.

Unfortunatly no one seem interested in such a huge business opportunity.

Sad.

@Shamar @ciaby @ajroach42

ah jeeze man, think of the sysadmin needs.

the mail servers fail. the administration is confusing because docs aren't perfect, so it gets misconfigured. the network goes down. baby pukes on server and it fails to boot. server is overloaded by volume of spam.

then the task is outsourced to a guy interested in managing the emails....... whoop whoop we're recentralizing.

@pnathan @ciaby @ajroach42

Oh, no I can't think that it's not possible.

It's not easy, but we buy and sell firewalls, routers, wifi spots... we can sell mail server too.

And with E2E encryption by default, do we really need spam filters?

@Shamar @ciaby @ajroach42 my Inner Young Geek wants to argue that actual configurable systems are actually not used in the home outside and that mail servers cross that barrier between appliance and administrating-needing machine.

but let's not rabbit trail onto that. ;-)

more my contention and question is: should we expect a member of cyberspace to be knowledgable in minor sysadmin?

I argue yes! we expect people to be able to refill their oil in cars, right?

@pnathan ... no?

There's a whole industry out there of shops that only exist because people don't change their own oil.

@ajroach42 changing oil isn't refilling oil.

one you just stick a can of oil in, the other requires draining the system, changing the filter, etc. much more specialized tooling & environment to do it right.

6 @pnathan @ajroach42

We do not learn car hacking at primary school because it's a specific skills set with little effect on a person growth.

Computational thinking is a completely different things.

Even if you don't code, the ability to think clearly about a problem, decomposing it into small pieces, debugging an argument... are all skills useful beyond the computer use.

It's literally a matter of power and freedom, because who is able to do that has a strong advantage over everybody else.

@pnathan @ajroach42

As an #hacker myself, I feel it as a #moral duty to spread such #power.

Nobody should be able to exploit #ignorance, because we know that ignorance is the precondition of #curiosity.

@Shamar @pnathan @ajroach42

in the UK they are now starting to teach this at junior school level (this is for children at biological ages 5-7, which is called Key Stage 1 here)

When I grew up in 1980s it was only taught in high school at age 14+, to those who had opted to take Computer Studies (a introductory CS course)

bbc.co.uk/guides/z3whpv4

@pnathan

@ajroach42

a better metaphor is cooking. everybody is expected to know enough about cooking to feed themselves. some people cook at a much more expert level, and people who are capable of feeding themselves pay those experts to feed them occasionally. cooking for yourself has benefits over eating out even if you aren't very good, because you can cater to unusual preferences.

@enkiv2 @pnathan @ajroach42
cooking for yourself also keeps the cost of eating out down, because professionals are competing with free. if all professional chefs started doing something (like cooking 'rare' burgers as well-done to avoid liability), home cooking isn't subject to those rules.

It's possible because cookbooks are mostly for the intermediate talented-amateur cook.

@enkiv2 @pnathan

And even still, we have tools (frozen dinners, spice blends, hamburger helper) to help folks that can't cook well still manage to cook what they want.

I want the Hamburger Helper of modern software development.

@pnathan @ciaby @ajroach42

Exactly!

Give us a little cheap fanless server and we will move the world!

@Shamar @ciaby @pnathan Pi 0 W?

$10 + a power supply.

But you still have to deal with NAT, or you have to deal with IPv6, or you have to not deal with the internet.

@pnathan
At scale, yes.
Although I feel that software is not evolving because:
The effort to develop new OS is too great, given the amount and complexity of the modern hardware (and closed specs).
Without a new OS, you can't develop new paradigms, and so we're stuck with ideas from the 70s (unix mostly, plus VMS-influenced Windows).
Programming languages are going to use the OS, and that's why we're not really progressing...
My proposal: simpler hardware, open and documented. Build on top of that. No backward compatibility. :)
@ajroach42 @Shamar

@ciaby @ajroach42 @Shamar I agree that backward compatibility has to be nixed for real research and change to occur.

now I have to debug a piece of code that is like the reification of all bad backend possibilities combined.

@ciaby @pnathan @ajroach42

What if everything was a #file for true?

What if all you need to support an #OS (and all #hardware it can handle) were 16 syscalls?

Keep this in mind and give a look at jehanne.io
(#Jehanne still needs 26 system calls, but I welcome suggestions to polish it further :-D)

@Shamar @ajroach42 @pnathan
I was actually looking at it before and find the concept quite interesting :)
Can I run it in a VM easily?

@ciaby @ajroach42 @pnathan

Follow the readme, and you should get it on QEMU pretty fast.

(First build takes 30 minutes due to GCC, later it takes 3 minutes at most)

@Shamar @ajroach42 @ciaby @pnathan

The direction this seems to be heading in is that we need to make a LISP (or Scheme) web browser.

I'm planning on starting soon...

@ixn @Shamar @ajroach42 @ciaby why should you make a web browser?

why not a gopher system for browsing files?

and for interactive work, why not dig through one of the old remote windowing & data transfer systems, and use *that* approach?

@ixn @Shamar @ajroach42 @ciaby believe me, the modern web is a *windowing* system with *HTTP* calls as data transfer protocol. kill the HTML/JS/etc side of it, and use a stateful connection

@pnathan
Or, why not just use SFTP, with a FUSE filesystem on the server?
You can use a file manager and text editor, and you can have interaction, as well as authentication through SSH.
I think it even supports FIFOs...