Hacker News Reader

Yeah, I hear where you're coming from, but all of those parts are still serviceable. If your starter goes on your Ford car, you might have to buy a Ford starter (or a third party starter for a Ford car), but it can still be fixed with a commonly accessible set of relatively inexpensive tools.

If the RAM in my Macbook gets cooked, either I can buy an entire new motherboard, or a new laptop. It's throwing the baby out with the bath water, and the truck/lawnmower example you've mentioned doesn't extend to that.

I agree; it sucks for the reason you say.

"Moore's Law" is probably part of the reason we're okay with this: by the time RAM/CPU/mobo start failing, a laptop is 5 (or 6 or 7 or 10) years old and the laptop's a dinosaur anyway. Heck, by that point a lot of us are probably hoping our laptops fail so we have an excuse to get a new one.

I wonder if we'll stop being okay with this eventually, now that desktop/laptop CPUs aren't getting faster at the rate they used to. My 2011 Macbook Pro is still plenty fast enough for me. It would be nice if it lasted a long time.

Is that actually true?

We've hit a plateau on overall compute speed. There are incremental improvements, but the difference in performance over 5 years today is much less than it was 7-10 years ago.

My 10 year old laptop is awesome, and it's a Thinkpad too. I have fixed it, replacing a fan, upgrading RAM, etc... all good, but it's performance isn't so hot. Buying new made a lot of sense.

But, my current 4-5 year old, high end, laptop is still competetive in a way that didn't used to be true.

I've a 2012 MBP, and it's still plenty fast. The newer machines are nice, but it's not like the 2012 is irrelevant, and that particular MBP is servicable. Just barely, but possible. I've done a number of things on it.

The new machines now just aren't. So even the expert user service is off the table, and I find that hard to deal with personally, because I know I absolutely do plan on running the machine for 5+ years.

Apple is still selling the old model user upgradable MacBook Pro for $1100 http://www.apple.com/shop/buy-mac/macbook-pro the same price I bought one for in 2011. Except I bought refurbished and smaller HD.

Some of the new machines Apple is selling for $1000 are actually slower at many tasks. But the market has spoken and Apple is selling hardware that suits most user's narrow tasks of email, browsing and video. It also streamlines manufacturing, as they're already focused on mobile where everything must be stuck together in one chip.

But they're enabling 4K video editing on iPhones and iPad Pros with what I'm assuming is custom hardware. And that could be a very smart approach if it was applied to more tasks on Pro computers. Performance could actually move forward again.

Indeed. Custom hardware can move us forward, and that's an interesting trend that may actually drive me to buy new machines sooner.

One of my daily driver tasks is CAD. Mechanical CAD. Due to the age and massive human investment in geometry kernels, single thread performance really matters. It is difficult to apply multi core processing to CAD problems.

It would be interesting to understand what special hardware could improve on this.

Well, increasingly not even independent mechanics can fix various parts of cars, let alone enthusiasts with only cheap tools.

and a lot of that is by design, which is sad, and also is why I am buying older cars.

Amen to that. I am seriously thinking about going out and buying a 1975-1985 or so Chevy C-10 pickup with a small block V8. Those things are dead simple to work on, parts are cheap and plentiful and readily available, and there's plenty of room in the engine compartment to get in there and work.

Contrast that with my 2000 Ford Expedition where you have to jack the truck up and crawl underneath just to change spark plugs. Never mind the need for more specialized tools when you're dealing with fuel injection and messing with the ECM, etc.

Yeah, I think it's time to go back to older cars. I want something I can work on myself without needing a $5,000 tool that I'll use once, and that you probably can't buy in the first place. :-(

They are actually working on that. With DRM, DMCA, shitty software, etc. They want you to go back to the vendor to make sure things are fixed. Repairable? Only by authorized service dealers. At a monopolistic cost.

And in the US you can thank the EPA for helping with that.

How exactly is the EPA involved?

Probably means CARB, where states that have adopted California's emissions regulations require certain parts of the vehicles be replaced by special CARB-approved versions when they fail. Even if they're exactly the same as non-carb approved versions yet cost 2-3x as much.

Who is "they"? Apple? Because they USED to sell laptops that were serviceable, and have slowly over time migrated to once that aren't. Not because of software, but in order to reduce the size.

And EPA? What?

If you want a 1kg notebook that's 100x the power of a room-sized $40,000,000 Cray 2 from back when, you'll have to leave off things like connectors, removable covers, generalized internal interfaces, empty space (for varying-sized components), beefier batteries (to run whatever you plug in), etc. Some of us value compact over modular.

You want "repairable by user for cheap"? Such is eminently available, and will decidedly not weigh 1kg.

You're right, my Thinkpad X201t weighs 2.89 lbs. That's 1.31kg, it's so difficult to lug around but at least I can replace the hard drive, RAM, keyboard, CPU fan, battery, and fundamentally keep it maintained until the processor fails.

If your engine block is damaged in your Corvette, you'll probably need a whole new engine block. Modern engines are using increasingly exotic materials & construction that are ever less repairable.

Your RAM can be replaced by an expert with the right tools, but computing hardware has become so cheap that it often simply isn't worth hiring an expert to perform serious repairs.

>Almost no other category of consumer good works in the modular way that we nerds wish that everything would.

One exception I can think of are bicycles.

Although, even bicycles (at least those designed for racing) are starting to become more and more proprietary. Try mixing and matching Shimano and Campagnolo shifters and derailleurs or using a seatpost from an aerodynamic frame in another one.

Have Compagnolo components ever been compatible with other maker's parts? As far as I know Shimano and SRAM are compatible, and most of the smaller players work in the same way. Compagnolo, AFAIK, is the only outlier here.

As for the seat post, it's these molded carbon frames that seem to be built for short use and then the trash that are problematic in my eyes. They're just wasteful and most of the people buying them are doing it just to buy some time for their various races.

Having gotten in to triathlons a bit lately, I'm constantly annoyed that someone can literally buy gear to make themselves minutes faster, even though our engines may be comparable. I wish the organizing bodies would standardize the equipment for competition.

The flip side to this (triathlon gear) is ITU draft legal racing, where it's basically like a criterion and requires the same UCI compliant bike setups as a road race would. This solves the problem in one way but creates a new one. Take a look at the world standings (for women, especially). Gwen Jorgensen is a beast, for sure, but she wins because she's a ridiculously strong runner who managed to get her swim good enough so she could hang with the leaders. On the bike she just has to ride with the pack. ... then, like Mirinda Carfrae in Ironman, leaves everyone in her dust on the run. I'm not sure this is fairer or not.

Haven't (particularly longer distance) triathalons always been a runners race?

Not really. Exceptional cyclists who are average runners still do very well on full iron distances. "Average runner" meaning they can crank out a 3:15-3:30 marathon after the first two events (so not really "average at all).

That said, people who come into triathlon from a running background do tend to do better than swimmers or cyclists who have to learn the other two events.

In the days of friction shifting, it used to be common to mix and match components across the entire drivetrain. You could have shifters, derailleurs, chains, freewheels, and hubs all from different manufacturers.

Drivelines are more integrated now, with derailleurs, cog tooth profiles, chains, and shifters designed as a system with minimal thought given to cross compatibility. At best you could use a non-Campy/non-Shimano chain.

As far as I know Shimano and SRAM are compatible,

My understanding is that they aren't, at least where we are talking about indexed shifting (which is pretty much everything these days). The amount of cable pull per "click" is different between SRAM and Shimano, which means you can't use a Shimano shifter with a SRAM derailleur. Or maybe that's just true in MTB world and not road bike world, not sure.

OTOH, chain, cogs and chainrings are pretty much compatible, at least within the constraints imposed by the changes in chain width to accomodate the newer, denser cassettes. Eg, 10 speed chain or 9 speed chain won't work with an 11-speed drivetrain.

Hi Fi stereo sound is strongly out of cultural favor, tinny little speakers and 64K mp3 quality means you're conspicuously consuming smartphones, which is more important. However, component hi-fi stereo or 80s era home entertainment systems were a very modular thing in the past. From my wife's point of view its one octopus like appliance, but to mine it was a collection of amp, speakers, and legacy media players.

Can't imagine why you've been down voted for this completely accurate observation.

Probably because the observation is accurate but it's mixed in with snark like "conspicuously consuming smartphones." Component systems were certainly modular although I honestly don't know if I were starting over today I'd go with a big multi-speaker stereo system over something like a Sonos.

I remember seeing a vintage studio recorder at one point; the manual it came with included all of the schematics of the electrics, so that you could make your own replacement parts. That's some future proofing right there - the device's creator (probably Philips) knew they wouldn't last forever, while their device probably would. It did cost a fortune back when though, I have to add.

If you buy a pair of Sony MDR-7506 headphones -- they've been around since 1991, still being made, under $100 on sale -- you get an exploded diagram with part numbers for everything.

At the end of the cable is a 3.5mm TRS phone jack connector, which has a passive adapter to 6.35mm (1/4"). The 1/4 inch phone jack was invented in 1878, for use in the repetitive insertion environment of a manual telephone switchboard.

Basically every set of headphones you've used in your life has used a phone plug/jack connector. It's a pretty good bet that a hundred years from now there will still be people using phone plug/jack systems to connect up systems where some part needs 2-4 wires and will be repeatedly removed and re-inserted.

"Understand that PCs were end-user fixable and upgradeable by accident."

I think if that were the case, IBM would not have released such complete documentation with the PC, XT, and AT. They could've just as well kept it to themselves, like they started to do with the MCA-based systems (PS/2).

The IBM PC design was a fabulously successful attempt to co-opt the much more accidental S-100 eco system. S-100 software and hardware could be easily ported to the PC, solving the chicken / egg problem inherent in creating a new ecosystem to compete with established ones.

The one mistake they made was assuming they could keep control by controlling the BIOS.

That information was primarily intended for OEMs, system integrators, and service technicians. They weren't really concerned about little Johnny down the street upgrading his computer... they certainly weren't trying to prevent that but it's not why they created those specs.

You're hopelessly, utterly wrong. I refer you to, for just one of millions of examples, the Apple II reference manual: https://archive.org/details/applerefjan78. The early culture of PCs was one of end-user fixing and upgrading by default, often primarily.

> the Apple II reference manual

And then the Mac was sealed. You needed a special tool just to open the case.

I upgraded the RAM in my Atari ST by solder-sucking 16*(16+2) through-holes and soldering the additional chips and caps myself. Not exactly modular by design.

The early history of PCs included sending away for a box of parts and a manual and then soldering the pieces together yourself. And then when you put the kit together, there was no software, so you had to write it yourself. Also, there was no keyboard, so you had to input the machine code for your software directly into RAM using toggle switches. And there was no real display, just a few LEDs that you could blink on and off.

https://en.wikipedia.org/wiki/Altair_8800#Description

I'm very aware of that.

For brevity's sake, I talked about IBM PC hardware because that's what our current x86 hardware is directly descended from.

(If you typed your post on some crazy descendant of the 6502 and/or the Apple II's architecture... then I stand corrected, and please share details of your setup!)

This was also true for all x86 ancestor hardware. For example, the IBM XT came with a technical reference manual that provided the schematics and source code (https://archive.org/details/bitsavers_ibmpcxt150ferenceApr83...).

In fact, the fundamental success of the PC/XT line was that it had increased user-expandability and maintainability over the competitors, as documented extensively in all accounts of the period.

And we're all typing on crazy descendants of the Apple II architecture, because pieces of it informed all subsequent architectures:

"It was also decided along the way that the machine would have an "open" architecture, like that of the Apple family. IBM would include slots under the cover of the machine that could accommodate plug-in boards that would add features or even change the entire personality of the PC. And, to make it relatively easy for outside companies to participate in the building of the PC market, IBM would publish a Technical Reference Manual with the entire set of electrical schematics for the machine and a full explanation and printout of the ROM-based BIOS (Basic input Output System) that provides the hooks into the machine for hardware and software. The ROM BIOS and the IBM logo were actually the only elements of the entire machine that bore an IBM copyright." (http://www.atarimagazines.com/creative/v10n11/298_IBM_coloss...)

Frankly, your lack of understanding of the history you're so confidently asserting facts about is beyond shocking. I'm sorry if that seems mean-spirited, but there's no other way to put it.

This is a very American point of view. German and Japanese cars are very modular; you can take certain parts from a 3 series and use them in a 7 series, and vice-versa. The transmission in the Toyota 86 is straight out of a Lexus IS.

It's not really a nationality-based thing (although Toyota and VW do have the largest volumes on a single platform). There's been a significant shift toward modular automobile platforms in general. Some numbers I saw recently indicated that almost half of passenger cars were based on the top 20 platforms.

For some brands, there are retrofit fans, who take parts from newer models and put them into older, to get features that were never offered when their cars were made.

This is actually true for American cars as well. I used "Ford Fusion" and "Ford F-150" in my example because one's a family sedan and one's a big pickup truck. I bet the Fusion shares more than a few parts with the Focus.

At the very least, each American manufacturer shares engines and transmissions across its various models.

80s and 90s American car interiors were sort of quietly hilarious. They shared common parts like crazy. You'd see the same gearshift, steering wheel, and so forth on various models. Maybe that's why it was a really ugly era in automotive interiors; their designs were constrained by all the common parts.

And the transmissions from e36 and e39 (early-mid 2000s) are built by GM. There is actually a lot of cross platform and cross manufacturer parts rattling around in our vehicles.

Honda is a great example of modular components. Taking a d-series and swapping it with a b-16/18/20 and even k-series with just a change in motor mounts/axles/ecu. It takes an afternoon to swap it over.

Of course as you get further into the aftermarket realm things like bolt-in conversions for swapping 2jz into 90s mustangs becomes possible. Fabbing up some brackets isn't that hard, conversely, desoldering a cpu or dealing with non-modular computer components is a much more daunting task to me. The parts are too small and fragile.