Hacker News Reader

Comment by tux | Hacker News Reader

"There are only so many features a program can add until something becomes just good enough. At this point new features add less and less value."

I agree 1000% with this! But even most websites this days add way too many features that you don't simply need. Simplicity is the king in everything! I'm a firm beliaver in long term hardware, when building a computer it needs to last at least 10-15 years between builds or upgrades. Only exception is hardware failure or software no longer supporting some hardware (which is rare on linux) Good article thanks!

View on HN

>at least 10-15 years between builds or upgrades.

That wasn't realistic until very recently, after Moore's law completely died off, but I hope it becomes the standard.

I have an X220 from 2011, i7 processor & 8 gigs of RAM. I payed 180euros for it (best deal I ever made, I know) and I hope I can get 10 other years from it. It chokes on modern uberbloated webapps, but hey, it's not like I want to use those anyway.

I built a gaming computer with an i5 and a gtx 1070 in 2016, and unless I decide to go for something else than 60fps & FHD resolution I still max out every game I play. I don't see myself upgrading in the next 3 years at least. Before that I had a crappy computer built out of low-range parts from 2010, and it was perfect until Dark Souls 3.

I have a 2012 Macbook Pro that's still okay for most of what I do with it, including Android app development. Though the Nvidia GPU driver is terrible and leaks memory copiously, and that's annoying. Web apps are sometimes sluggish, but I got used to the fact that modern web just can't be fast on any hardware unless done by people who know what they're doing which is rare. I'm looking forward to whatever the refined and beefed up M1 successor would be. That would probably last me another decade if not more.

I had to abandon my 2012 macbook pro for a new intel mac earlier in the year after it was overheating and killing keyboard and trackpad input during zoom calls, but before our modern zoom based world, what a beast that computer was. Perfectly performant for typical tasks, unlike computers of old that would struggle to run the latest OS after four years.

I'm eyeing the ARM macs, but I think it would serve me well to wait at least two years for software to become more compatible and kinks worked out. People in my field have been having some issues with ARM and rosetta2, even having to return laptops.

Hm. Mine handles zoom calls just fine, including the background replacement thing. Sure, the fans do spin up a bit, but that's about it.

I think it was the battery swelling and affecting the top case more than anything. trackpad has long been cracked.

+1 for the i7 x220. I paid £150 for mine a couple of years ago for what I thought would be a throwaway bit of kit at end of life for an experimental Linux build. But what I found was the sheer utility of the thing has been unmatched by any more recent hardware purchase.

> I have an X220 from 2011

Sounds like you're in good company. The author recently upgraded from a 2015 Mac to a heavily modded X230[1].

[1] - https://thedorkweb.substack.com/p/everything-but-the-kitchen...

I was fortunate enough to buy an x220 in 2011. Don't use it much these days since I have newer machines available (though one of them is also sandy bridge), but it's still perfectly serviceable and to this day my favorite keyboard ever on a laptop.

> every new feature added was worth less than the features taken away

I think this is the core of the issue, not just adding new features. Adding features, especially in the general purpose OS that was chosen as an example, is not a bad thing as long as they stay out of sight and are there as optionals for the people who need them. But removing one feature to add another, or twisting an existing feature into something that can satisfy 100% of the users is bound to make the product worse.

The real obstacle the 100 year computer faces isn't software, there's no reason for it (or a copy) not to run for hundreds of years, providing it has something to run on. it's hardware. Making a complex device that can run for a century is close to impossible unless there's a chain of supply behind it to constantly replace failing parts. That's a big ask over 100 years.

And then there's the philosophical question: if over that 100 years every part is replaced and upgraded to fit the new trends, take advantage of new tech, etc. is it still the 100 year computer? Or the ship of Theseus?

These are all really good points. I do wonder though, is there any combination of parts that will make even the simplest computer last 100 years? (With proper maintenance, ideally not replacing parts that can't be produced for the next 100 years)

I'm defining a computer here as an electronic CPU with some sort of I/O, both user I/O and peripherals.

I'm reminded of something like a Commodore 64. I've been able to get one working after 30 years of it sitting on a shelf before, although many suffer from burst capacitors, bad chips, oxidation of circuits/pin connectors, etc.

I wonder if you could take the idea of a simple, self contained system like the C64, but use ultra reliable, beefy industrial components, like from computer systems in manufacturing factories. Maybe throw a few redundant components the computer can silently switch over to if the first component goes bad.

I imagine you could design and build such a computer (I also thought that the definition of a computer might change in 100 years so much that today's devices no longer fit it). Probably using only solid state components, having it completely self-contained, and having all the components built for longevity may be close to but not quite impossible. But it would be entirely unpractical for anything beyond the proof of concept, certainly not as a mass market device. The "longevity" part of every component likely sacrifices speed, power, and performance in exchange for reliability. Imagine things like much larger features in silicon, duplicated or triplicated (or more?) components and units, use of more non-integrated components, the system as a whole would incorporate far more redundancies that take space and possibly power just for the eventuality they need to take over, etc.

Not to mention that testing and continuous development are challenging to say the least given the timeframe. By the time you discover your design has flaws you start over with a vastly different 100 year computer.

So from all practical perspectives I can't see a 100 year old computer being built with today's tech unless we mean stuff running in hypervisors and emulators that are constantly maintained and developed over that century.

I imagine looking at the design of electronics that goes into space probes would be a good place to start although that's not quite the design point.

It might be impossible to make a 100 year computer that's practical.

If we look at other "tech" built to last, I think sewing machines. Stand mixers. Cast iron pans. And I'm sure lots of industrial equipment I don't know about.

None of these really changed much in the past 50 years that made it useless, and I can see these being useful for the next 50 years pretty easily.

The only thing I can think of is a base computer with some interface to a breakout box that can be upgraded based on how computers work.

This definitely cheats the idea of a 100 year computer, but you could have a wifi breakout board. Maybe a reader for the next storage format that's popular. Neural whatsit chip that may revolutionize computing. But, again, that's cheating. And at that point just buying modern hardware makes so much more sense.

People have been adding peripherals that allow usb and SD cards to Commodore 64s. And I believe that many of the parts for a C64 are still around.

Keep in mind that the Commodore C64 is "just" 40 years old and the rate of decay is not linear. I imagine the definition of the 100 year old computer probably targets more than "survivors".

The definition of the "100 year old computer" as described in the article is fuzzy and kind of a moving target. In the loosest interpretation anything can be a 100 year computer just as long as you care to maintain it for 100 years. If upgrading parts and building adapters for future technology fits that definition then you can do this with any machine, and the rest of the constraints are just artificial (like "must be offline first"). If you actually want to have as much of the original computer as possible for that century then it's no longer feasible.

As I mentioned in my original comment, they simply become the ship of Theseus. Heck, my own desktop is that, starting 20 years ago with one hardware config and a Windows XP and reaching now with completely different hardware and Windows 10 with no "full reset" in between, just upgrades and updates.

Reply on news.ycombinator.com