Then you start running into failure. Modern electronics isn't really built to last, so having a laptop that will last you more than 10 years is extremely unlikely.
Until market demand changes to quality, long-lived electronics, you will still be rolling over your computing device every couple of years. And then there's always the trends and cycles of fashion...
My desktop has already hit this point. Granted I'm not much of a PC gamer or anything, but prior to my current desktop I would upgrade and/or totally replace my main desktop system at least once per year. Now I've got a 4 year old desktop running on a Q6600 and Nvidia GeForce 9-somethingOrOther that I'm still insanely happy with. The only upgrade made to it in the past 4 years was replacing the system drive with an SSD.
Laptops are pretty much "done" for me too if you don't consider size/battery issues -- a Core i5 laptop is more than enough for pretty much everything I do and likely will be for the next 4+ years, but now I just need a smaller, lighter, more battery efficient one.
Just because an increase in CPU power won't lead to a perceptible improvement in performance, that doesn't mean that your CPU is fast enough. It just means that the bottleneck isn't your CPU at the moment. If all your data were stored on high-end SSDs, and your network connections were multi-gigabit, and your display was 300dpi 120Hz, you would find a faster CPU useful.
Only when programmers are the bottleneck, ie. computers can run the most complex algorithms in realtime, rendering their output at the limits of human sensory capabilities, will we have computers that are "fast enough". Until then, our machines are merely tolerably (or not) slow.
Calling a laptop that still takes more than a second to launch Photoshop "fast enough" is too forgiving - the room for improvement is so obvious that we can only forget it by willfully ignoring and excusing it.
I'm not sure if all of these assumptions follow. Marketers are in the business of finding reasons to compel you to buy new stuff, and marketers are the ones making products.
In the past, people held on to things longer, not because they lasted longer or were viable longer, but because they didn't have the resources to buy new stuff. In order for "repair culture" to set-in in any significant way, consumption culture would need to be displaced. The factors that would drive that are largely unrelated to technology. For example, a global economic depression would result in people keeping their possessions longer simply because they don't have the money to buy new.
Look at the many other products in our lives that people regularly replace before their usefulness has expired. Cars are probably the best example. When the economy is booming, people replace their cars rapidly. As it slows down, they hold on to them longer. I can't find any long term data, but the graph and caption on this page imply that the two are correlated: "The Changing U.S. Auto Industry Series: Consumer Sentiment During Challenging Times."
I don't see any reason computers will be different. "Hackers" will remain a niche community, but I still think the many other reasons cited by the author will result in better tools for us to play with. For example, development of reasonably priced and performant FPGAs would be huge. Look at the Arduino line of products.
The future is still a bright one. I just don't see "heirloom laptops" in our future.
> people keeping their possessions longer simply because they don't have the money to buy new
This is already the case for most of the developing world. However, with increasing automation and outsourcing, it looks as if that is a future that might not be too far away here too. There might simply not be enough jobs to go around and a new class of lifestyle may arise where you need to purchase durable and repairable goods and a market would rise around that need.
> For example, a global economic depression would result in people keeping their possessions longer simply because they don't have the money to buy new.
That would seriously suck then because the stuff that we've got today for the most part was built with a very definite life-cycle in mind. So when you are dumped in that global recession you need the quality stuff that wasn't produced when the recession wasn't on yet and you can no longer afford its creation.
Absolutely, on top of that the stuff that is made today is not even meant to be repaired at all. It's all about the cost to the manufacturer and not falling apart until the day after the warranty expires.
Although I do agree that Open Hardware has a bright future, I'd just like to point out that people have been calling the end of transistor size scaling for a few decades, and technology keeps finding ways to get around the limitations (strained silicon, high-k metal gates, finFETs, etc.).
My point is that, even in that case (current silicon scaling stopping at 5nm), there are a lot of potential technologies that could allow Moore's law to continue (nanowires, III-V materials, etc.).
I like where this is going, ie. Hardware becomes just another software component.
The key item thats left is some sort of standard interface to port programs to a FPGA style processor, i.e. a way of interfacing between a language compiler and HDL.
I like the look of the Reduceron (http://www.cs.york.ac.uk/fp/reduceron/) which looks to be funded by Xilinx and Lava( by Satnam Singh, who works for Microsoft Research), purely because I think its easier to visualise a functional language translating into circuitry.
Although I do wonder whether it will take a start-up to be able to successfully merge both the software and the hardware components into a workable model.
I like where this is going, ie. Hardware becomes just another software component.
Unlikely. You can't run software without hardware. Even if the computers of tomorrow are a mess of FPGA's that can be reconfigured on the fly, somebody's got to make the FPGA's.
I am somewhat annoyed by the author's portrayal of individuals or small companies as the innovators, and large companies as the converse. Who do you think drives Moore's Law? It's not some force of nature. The material properties of silicon are not changing every 18 months. Do you think it will not take innovation to produce a viable 14nm production line?
Additionally, even if we hit hard limits in transistor gate size, do not expect progress to stop. There are alternative technologies aplenty that put silicon to shame, that are only avoided because of how cheap manufacturing is with silicon.
One more thing. The reason early computers and radios shipped with manuals and lists of replacement parts is because the technology was young, and things broke. As the technology has matured, it has integrated into our lives and failure rates have gone down. As a consequence, fewer and fewer people are interested in the guts, and people expect it to "just work". They have no interest or use for a repair manual with their Macbook Pro.
It is much the same as with the juvenille stages of any technology. Automobiles were once notoriously unreliable; old British cars were said to require work every weekend to keep running. Next, American cars had 5-digit odometers, because cars didn't last beyond 100,000 miles. Today, cars routinely go 200,000 miles or more. There is talk that soon engine wear will cease to be the limiting factor in a cars' life. As a result, your average citizen has never seen the inside of their engine bay, and could care less about how to fix their car- they expect it to just work.
The fail rates may have reduced some interest, but I believe there's other bigger reasons:
+ Students are increasingly brought up to do memorisation over creativity.
+ It's hard to hack integrated microchips. Especially with proprietary software. Want to repair your own car? Need to take it to a Brand(tm) Certified car repair service to even get the proprietary diagnostic. Even then, they may refuse to give the logs to you.
+ Loss of ownership with IP laws. Try to break open Wii? You're a pirate! Jailbreaking? Bad! Also run the risk of being sued for talking about breaking open hardware despite owning it.
It's largely a myth that you need proprietary scans. OBDII/III are standards that give instructive error codes - it's actually pretty easy with them and a good multimeter. The manuals are step by step - I'd say parts monopolies in German cars, etc. hurt worse.
> Students are increasingly brought up to do
> memorisation over creativity.
Is is so, or is this just a popular thought mindlessly repeated over and over?
The thing is, for creativity you need a foundation of knowledge and there is no way to avoid memorization. Also, you may need to memorize what was already done:
if you keep recreating everything yourself from scratch you may not have time to create anything new.
My money is on massive FPGAs, think a whole wafer of silicon with just a bunch of general purpose IO and power.
Once stuff like that becomes available at a reasonable price (a big FPGA will cost you a lot of money at the moment) open hardware will be as simple as downloading a bitstream on to your 'general purpose' rig.
24 comments
[ 1.6 ms ] story [ 62.5 ms ] threadUntil market demand changes to quality, long-lived electronics, you will still be rolling over your computing device every couple of years. And then there's always the trends and cycles of fashion...
Laptops are pretty much "done" for me too if you don't consider size/battery issues -- a Core i5 laptop is more than enough for pretty much everything I do and likely will be for the next 4+ years, but now I just need a smaller, lighter, more battery efficient one.
Only when programmers are the bottleneck, ie. computers can run the most complex algorithms in realtime, rendering their output at the limits of human sensory capabilities, will we have computers that are "fast enough". Until then, our machines are merely tolerably (or not) slow.
Calling a laptop that still takes more than a second to launch Photoshop "fast enough" is too forgiving - the room for improvement is so obvious that we can only forget it by willfully ignoring and excusing it.
In the past, people held on to things longer, not because they lasted longer or were viable longer, but because they didn't have the resources to buy new stuff. In order for "repair culture" to set-in in any significant way, consumption culture would need to be displaced. The factors that would drive that are largely unrelated to technology. For example, a global economic depression would result in people keeping their possessions longer simply because they don't have the money to buy new.
Look at the many other products in our lives that people regularly replace before their usefulness has expired. Cars are probably the best example. When the economy is booming, people replace their cars rapidly. As it slows down, they hold on to them longer. I can't find any long term data, but the graph and caption on this page imply that the two are correlated: "The Changing U.S. Auto Industry Series: Consumer Sentiment During Challenging Times."
http://www1.eere.energy.gov/vehiclesandfuels/facts/2010_fotw...
I don't see any reason computers will be different. "Hackers" will remain a niche community, but I still think the many other reasons cited by the author will result in better tools for us to play with. For example, development of reasonably priced and performant FPGAs would be huge. Look at the Arduino line of products.
The future is still a bright one. I just don't see "heirloom laptops" in our future.
This is already the case for most of the developing world. However, with increasing automation and outsourcing, it looks as if that is a future that might not be too far away here too. There might simply not be enough jobs to go around and a new class of lifestyle may arise where you need to purchase durable and repairable goods and a market would rise around that need.
That would seriously suck then because the stuff that we've got today for the most part was built with a very definite life-cycle in mind. So when you are dumped in that global recession you need the quality stuff that wasn't produced when the recession wasn't on yet and you can no longer afford its creation.
The key item thats left is some sort of standard interface to port programs to a FPGA style processor, i.e. a way of interfacing between a language compiler and HDL.
I like the look of the Reduceron (http://www.cs.york.ac.uk/fp/reduceron/) which looks to be funded by Xilinx and Lava( by Satnam Singh, who works for Microsoft Research), purely because I think its easier to visualise a functional language translating into circuitry.
Although I do wonder whether it will take a start-up to be able to successfully merge both the software and the hardware components into a workable model.
Unlikely. You can't run software without hardware. Even if the computers of tomorrow are a mess of FPGA's that can be reconfigured on the fly, somebody's got to make the FPGA's.
Additionally, even if we hit hard limits in transistor gate size, do not expect progress to stop. There are alternative technologies aplenty that put silicon to shame, that are only avoided because of how cheap manufacturing is with silicon.
It is much the same as with the juvenille stages of any technology. Automobiles were once notoriously unreliable; old British cars were said to require work every weekend to keep running. Next, American cars had 5-digit odometers, because cars didn't last beyond 100,000 miles. Today, cars routinely go 200,000 miles or more. There is talk that soon engine wear will cease to be the limiting factor in a cars' life. As a result, your average citizen has never seen the inside of their engine bay, and could care less about how to fix their car- they expect it to just work.
+ Students are increasingly brought up to do memorisation over creativity.
+ It's hard to hack integrated microchips. Especially with proprietary software. Want to repair your own car? Need to take it to a Brand(tm) Certified car repair service to even get the proprietary diagnostic. Even then, they may refuse to give the logs to you.
+ Loss of ownership with IP laws. Try to break open Wii? You're a pirate! Jailbreaking? Bad! Also run the risk of being sued for talking about breaking open hardware despite owning it.
Is is so, or is this just a popular thought mindlessly repeated over and over? The thing is, for creativity you need a foundation of knowledge and there is no way to avoid memorization. Also, you may need to memorize what was already done: if you keep recreating everything yourself from scratch you may not have time to create anything new.
Once stuff like that becomes available at a reasonable price (a big FPGA will cost you a lot of money at the moment) open hardware will be as simple as downloading a bitstream on to your 'general purpose' rig.