Hacker News Reader

Comment by black3r | Hacker News Reader

Technology 40 years ago was so simple, that anyone with basic education in electronics could understand how this or other basic utilities work and when opened up could identify which component does what and if you identified a faulty component you could often easily replace it and thus repair it on your own, and skilled individuals could probably make their own from scratch if they could source all components and didn't care about patents....

Nowadays even the products which didn't evolve much since then are often made very hard to repair + new technologies are increasingly more difficult to understand and state of the art technology in some areas is so complicated that we're literally dependent on one or two companies who can make it for us (3nm chips for example, or web browsers in software)..

It will be very interesting to follow how much the complexity will increase in the next 40 years..

View on HN

Indeed. I always figured that even if the cost to manufacture has gone down for any given piece of electronics that cost to society has actually gone up because instead of repairing things (which also provided jobs) they are now discarded far earlier than they would have been in the past.

Much earlier in my life I earned some money on the side repairing electronics and while it is of course still technically feasible the number of one-off components in any given device is so high and the documentation and repairability so low that it just isn't worth it. This was not a good development, but it seems we're stuck with it. The one big plus from all this integration is that devices consume far less power than they used to.

It's like cars. Decades ago you might see your local mechanic once a year (if you're lucky) or multiple times a year to fix issues with the car, nowadays there's a good chance a new car will go 100k miles with only wear-and-tear items needing maintenance done.

But if they break you're in for a world of trouble, and you will in one go end up offsetting a good chunk (or more) of what you saved up to that point. In a way this is the result of designing things to outlast their warranty, once the warranty period is over anything that still works was essentially overdesigned. Fortunately for cars and other high ticket items the secondary market will punish brands that are playing it too loose and too fast but I've seen pricing for for instance headlight or taillight assemblies that make you wonder if there is an error.

My mom's fridge broke exactly one day after the warranty expired.

She was crying on the phone, then they said they would honour the repair anyway.

That made her happy again.

Old cars had more maintenance requirements and were far less reliable, but repairs tended to be simpler.

For example, you may need to tune your carburetor seasonally. Old points based distributors needed the points adjusted or replaced. Simple tools like a screwdriver and a matchbook can get those two sorted.

Fuel injection and electronic ignition eliminated a lot of regular maintenance, but they are harder for the average person to troubleshoot, let alone repair.

Gone are the days where a rough idle could be fixed with a screwdriver. But so too are the days where a mountain pass requires that same screwdriver to be pulled out on the way up and the way down to adjust the fuel-air ratio.

I have a car maintenance manual from very early in the 20th century (1910 or something) and the maintenance schedule it describes would be ludicrous nowadays. It talks about changing the oil and greasing all of the joints every couple hundred miles, for example. At the same time, there are some components that are essentially the same now. Based on the diagrams, it looks like the wheel hubs and bearings would be compatible with the ones on my modern boat trailer.

The bearing might be compatible, but new ones are so much better. The slop (if that’s the right word) in bearings manufactured 50 years ago was more than you get now.

I'm not a serious mechanic, but I suspect one of the reasons the bearings and shaft on the hub are tapered is so you can tighten it down and accommodate for minor variations in the tolerances. That would have been more important before it was all being done with CNC. The material for the bearings was probably not super either.

Things that have changed in bearings: better materials science led to more understanding of how you can harden surfaces and this in turn led to better 'races' in bearings as well as balls and rollers that are harder leaving bearings that last longer in principle. But all that wouldn't matter one bit if the bearings weren't sealed properly and that's the second improvement that led to longer bearing life. Finally, the third factor, improved lubricants, better able to deal with thermal cycling and high heat.

My Dad, born in the early 1930s, said that on any lengthy drive (I'm guessing > 100km), you just assumed that you would have at least one puncture or other minor breakdown.

He and his cousin were given a car with a wooden body; once the engine overheated and it burned to the ground.

Punctures... there is a good point! I haven't had one in years (let's not jinx this...), but in the past, up to the late 80's I would have them with some regularity. I suspect that the improved quality of the roads is also a factor in this.

I strongly believe there was a optimal point where it was both relatively straightforward to service vehicles (or possibly easier), and their reliability was much better than what preceded them (and frankly also what came after): probably in the 90s or very early 00s. An example I would kind of think of, was GM's throttle body injection, or some of the simpler port fuel injection systems and the engine and transmission technologies concurrent with these. They have basically all of the advantages of EFI, such as behaving correctly with altitude changes, not rampantly over or under fueling the engine (because of the feedback loop with the O2 sensor, etc.), while retaining the same simplicity (or better) than their predecessors. Ignition was kind of the same way: HEI distributors, and even better, wasted spark systems, are absolutely dirt simpler, and really simpler than points and condensors, while being much more efficient and robust. The electronics on these are relatively simple, well understood (basically anyone and their bother has been able to tune and flash a lot of GM EPROMS/EEPROMS for a long time now), have replacements that are also better and open (Megasquirt and similar), and very robust.

A lot of the newer technologies, like variable valve timing, are cool, but you can still have (and did for a time) a lot of the ease of service, with much better running behavior and reliability. We can have our cake and eat it, manufacturers are just apparently very incentivized against this.

I strongly agree with your assessment. You can have a lot of fun and reliability with circa 1995 technology combined with modern manufacturing techniques.

Much of the newer stuff is about meeting emissions and fuel mileage mandates. There's a reason VVT lockouts are a thing in the aftermarket

But both restricting emissions and improving fuel mileage are good things, right?

> Technology 40 years ago was so simple, that anyone with basic education in electronics could understand how this or other basic utilities work and when opened up could identify which component does what and if you identified a faulty component you could often easily replace it and thus repair it on your own,

You’ve never seen an ISDN phone then (came only a few years after this one was made)

Yeah, the beauty of simplicity... but OTOH, you had to dial all those phone numbers by hand, the quality for long distance calls was sometimes so bad that you had trouble understanding anything at all, when the phone rang, you had no idea who was calling (or who had called when you didn't get to the phone in time) etc. Newer technology is orders of magnitude more complicated, but it enables you to transmit not only voice, but also several Mbps of data over the same two thin telephone wires.

Comparing state of the art semiconductors to web browsers feels like an insult honestly

In my opinion browesr are just huge mess of shitton of technologies that are constantly changing/moving, yet fundamentally it is just about processing some letters, executing some code and drawing colors on the monitor.

Nothing novel that we weren't doing before or in other software (OSes, compilers, etc)

Meanwhile semico feels to be close to the limits of the physics.

" just about processing some letters, executing some code and drawing colors on the monitor"

Sure, nothing more. Except that this drawing colors on the monitor is plattform independent and used worldwide for all kinds of sandboxed applications and is the plattform for most commercial transactions and normal communications now. So quite trivial and surely anyone can do it.

Sure I made it sound trivial, but that's in the context of comparison with state of the art semiconductor engineering.

Semico. engineering is current "rocket science" of 70s.

Cross-platform GUIs aren't novel. Frameworks for this shit probably appear every year and are as messy as existing ones.

It's a mess, hard mess, but not novelty, not the edge of science / engineering. The problem is to get people to agree on something.

>and is the plattform for most commercial transactions and normal communications now

What's so novel about it? SMS or even your Master/Visa cards were doing it for decades

IRC, Email too.

And of course still: we were using Web browser as platform those things for decades.

It's not a bleeding "edge" technology in $currentYear.

WebAssembly ecosystem? yet another collection of VMs/Compilers/Intermediate Languages?

We were doing it for decades.

>sandboxed applications

Ok, that's fair point - there's novelty needed.

Very often because we try to write technology that's... easy to fuckup? idk - check how huge % of CVEs is related to JIT compiler

or because we write it in unsafe languages - check how huge % of CVEs is related to mem. issues.

But yea, I agree on security - stuff like postquant crypto and other protections.

> Semico. engineering is current "rocket science" of 70s.

Even rocket science is just Newtonian physics, for the most part.

That's the theory behind, the execution is hard.

Meanwhile, unlike moon-landings - we're constantly doing web browsers:

there was IE, there was Edge, there is Firefox, there is Chromium and its forks.

There are smaller browsers with less resources than Google.

Yet, we are not trying to figure out how to make web browser possible.

We know how to do it, we are doing it for decades.

It's just hard due to reasons - money, effort required, moving target, backward compatibility, and probably many other.

Can same be said about semico? Do we know what needs to be done to reach "state of the art" for year 2033?

Do semico competitors know their roads to match its competitors capabilities 10 years from now? I dont think so.

I get what you mean, but my original thought was about complexity as in how much effort (work & time & money) would it take to re-create the thing and start a competing business, given you already have all the necessary knowledge. Cause it increasingly feels like working on state-of-the-art is only achievable by big players...

Under these assumptions it's not really being bleeding edge, how much research went into it, or how hard of a science is behind that make it complex, but rather the amount of components and parts needed to assemble it (and in case of hardware also the rarity of them). When I am talking about state-of-the-art technology there I just mean the newest competitive version of a thing that most people use daily...

Modern browsers have code execution environments that very much push the state of the art in compilation and os-like capabilities.

>Modern browsers have code execution environments that very much push the state of the art in compilation

By which metric?

Well, I’m not a practitioner in that space, and maybe my perception is out of date.

But didn’t the semantics of JavaScript lead to some pretty novel ways of just-in-time compiling code at runtime? When the V8 engine was introduced, for example.

I have a Siemens phone that's similar to this (I turned it into a mobile phone). When I opened it up, it had a leaflet with the schematics of the entire device inside.

Maybe, maybe, there will be a compression/folding event, that squashes all the useless complexity into simpler artefacts.

"useless complexity," spoken like a true software engineer.

You make the case for the GPL very well.

Or WiFi. Or the USB protocol.

Reply on news.ycombinator.com