There was an article these days about how companies are starting to wish for the other face of Moore's law: computers getting cheaper every year at the same performance. The article was about the next Windows, but I always thought it applies much better to laptops.
I'd say there are three levels of performance suitable for such mini-machines:
- The equivalent of the last 486 or the first pentiums: can play mp3, browse the web (without the bells and whistles), display a pdf, run simple office applications, have ethernet and usb, may not have wireless. This is what a real $100 laptop should be like, and what readers could become (keyboard?)
- Can show movies, run a contemporary OS and application suite (including last versions of Open/Microsoft Office) - pretty much everything an office computers does, without the games and other needy apps. This is already covered, as there are many cheap laptops just like this.
- Sub $100 laptops. This is a very interesting field, because anything can be done. Text-only and monochrome displays, dedicated applications, the field is huge and the prices can be as low as any hand-held chinese game ($10 is not at all absurd). There are two big issues here: compatibility and net connectivity. I don't know what an USB or Ethernet chip goes for, but it it's $2 it's too much.
You'll be surprised if you find an actual PC with 486. I have installed XP on a Pentium II machine with 256MB of RAM and it was barely usable even for web surfing. Without bells and whistles: scrolling simple text+picture pages in IE/FF was very slow.
Launching any non-trivial application also took terribly long. Not just the Office, but even Firefox felt more like Photoshop. I couldn't believe how bloated the software had become.
These $100 machines you're talking about can't run normal software, you'll need Microsoft to start selling Win98 or you'd need a stripped-down Linux with XFCE or something. The key to performance IMO is smaller binary sizes: those CPUs had tiny caches and pathetic memory throughput. Pentiums ran on 33-66Mz FSB, compare it to 1066Mz of modern machines.
Isn't it amazing how software bloats and slows down so that it only fits the latest generation of hardware?
We were doing web browsing of the simple text+pictures kind back in the early 90s. Word processing and spreadsheeting has been done since the late 70s, early 80s (I think).
Totally agree. I wonder, with a market of a billion users, does a set of software apps that are not back-compatible with our bloated legacy become plausible?
You could use old Windows or Unix apps. Or DOS apps. Or even old Apple ][ apps. Or, you could write an entirely new set of apps - a fresh start, in a fresh market, on fresh hardware. Doesn't that have a certain appeal?
Freed from back-compatibility with bloated software, does a $10 laptop become plausible?
Maybe something like the hp200lx: DOS, 24x80 LCD screen with incredible battery life (2 AA batteries), came with lotus 123 spreadsheet etc. Or, even lower specs than that.
In fact, I have an old Toshiba laptop with initially 256MB of RAM. It had both Windows XP and FreeBSD+KDE in dual boot. Windows XP takes forever to boot and runs like a snail. On the other hand FreeBSD was pretty usable.
But one day, some part of the RAM stopped to run stripping
it down to 68MB. Even FreeBSD+KDE wasn't usable anymore.
So I decided to use Enlightenment as a window manager instead. Installing only the strict necessary: xmms for listening to music, vlc for videos, emacs as a text editor and for coding, xpdf as a pdf viewer, feh as image viewer.
The performance is pretty correct. It doesn't have an internet connexion because I found out I'm more productive
without it when coding. So I use it more than I expected to
get things done.
I am not sure what you would call "normal software", but i have an old machine that is a PIII with a 550Mhz processor and 128MB of RAM(i use it as a file server). I run on it a version of xubuntu(forgot which version, but it is about a year or so old). XFCE runs beautifully on it. Along with all of its software. And with an even lower end machine, you can use even lighter desktop systems.
Of course, you'd need to substitute a few of your regular applications, for example firefox still has a few problems with memory, and so you might want to look at an alternative.
I guess a laptop like that (if it ever gets made) would probably run some lightweight Unix-like OS. And it would be more like an appliance, where you can only run the programs that come bundled with it.
So, no installing random Windows software in it.
I disagree with the OP about wireless: I think nowadays Wi-Fi and maybe even cellphone connectivity would be a must for such a device.
The problem is that there's a very real and finite limit to the "Moore's price law" as you mentioned. Yes, things get cheaper rapidly after initial R&D, but this slows down dramatically after a few years. Once the initial R&D cost is recouped you will find that there is still a significant production cost that is slow to reduce and doesn't really move much at all.
I recently bought a 200MHz ARM7 board with 128MB of RAM. At bulk rates it still costs ~$70 a pop, and you still have to add peripherals to that to make it a viable personal computer. At these prices we're hitting close to production cost.
I had a 486 Dx4 120 (AMD), eventually with 64MB of ram, it always ran NT4. I remember downloading mp3s in 1997 or 1998. The machine couldn't play the mp3, but I could decode it in bulk to wav files, which it could play.
Yup... I've been wishing for a touch-screen, 4"x3" display I can hang in all the rooms of my house to control a variety of things... I think the sweet spot in terms of price would be $20-$50.
They're launching a laptop for $10? With 2GB of RAM? You can't even buy 2GB of RAM for that price. I really hope this is real, but when you put together the costs of a display, RAM, some sort of persistent storage, heck even the keyboard is going to run more than $1.
You could get the flash, ram, processor, and most of the peripheral ports integrated into a microcontroller package at well under $3 for the right volume, particularly if they were not component packaged but chips epoxied onto the motherboard. The motherboard and keyboard would probably be the same component (think pocket calculator membrane keys), and if there was a trackpad it might work similarly (otherwise use arrow keys for free).
In this context, government would give Rs 2.5 lakh per institution for 10 Kbps connection and subsidise 25% of costs for private and state government colleges.
10kbps is obviously a tiny amount of bandwidth, especially per institution. This is clearly not enough for doing interactive browsing in a class. For this reason I suspect that the 2GB is actually flash, not DRAM. You couldn't keep a room full of 64k DRAM computers busy with 20-30% of a dialup modem.
At this stage, the price is working out to be $20 but with mass production it is bound to come down
Still it is quite unbelievable. If it was something built to browse internet alone with little RAM and processing power may be. But their claim and price doesn't match.
No markup - They don't need to make a profit or even subsidize employee wages
Their costs are local - Shipping and labour for example are going to be extremely low in the local economy
Indian has done similar projects before such as generic pharmaceuticals which they sell to other 2nd and 3rd world companies.
Government subsidies can not be ruled out - that would change the equation completely.
Additionally, this device will probably not be available for general consumption - it sounds like they plan to use it exclusively for education. If not, reselling 2 GB RAM sticks imported from India could become a lucrative business.
The $10 looks like a transcription error. It should probably be $100 and the current version costs $200 which is consistent with what an OLPC XO laptop costs:
I'd love to know how they calculated the $200 "hidden cost" of the OLPC laptops. I'd also wonder what that figure is for other platforms, and how they plan to avoid it for their $10 implementation.
I'd say there are three levels of performance suitable for such mini-machines:
- The equivalent of the last 486 or the first pentiums: can play mp3, browse the web (without the bells and whistles), display a pdf, run simple office applications, have ethernet and usb, may not have wireless. This is what a real $100 laptop should be like, and what readers could become (keyboard?)
- Can show movies, run a contemporary OS and application suite (including last versions of Open/Microsoft Office) - pretty much everything an office computers does, without the games and other needy apps. This is already covered, as there are many cheap laptops just like this.
- Sub $100 laptops. This is a very interesting field, because anything can be done. Text-only and monochrome displays, dedicated applications, the field is huge and the prices can be as low as any hand-held chinese game ($10 is not at all absurd). There are two big issues here: compatibility and net connectivity. I don't know what an USB or Ethernet chip goes for, but it it's $2 it's too much.