exactly. i have a dell mini 9 fitted with a 64gb ssd and with a slab of pre-preg carbon fibre araldited to the lid to make it more robust. it is literally half the size of the smallest laptop that apple make, for example. It goes into my briefcase unnoticeably, it does all I need to do while travelling (it's just done a 10 day family holiday, archiving pictures, watching iplayer, etc), and is just perfect. Dell now don't make anything as small...
As I walked into the Yamaha shop in Ginza an hour ago there was a CD player whacking out bloody Susan Boyle massacring John Stewart's Daydream Believer. There should be a law, there really should.
I get really annoyed by people getting this the wrong way around. If something's easy to learn, it has a steep learning curve: your ability rises rapidly over time, repetition or whatever your measure of effort is. If something is difficult, it has a shallow learning curve: your ability increases slowly against time, repetition or whatever. Yes, I know ``steep learning curve'' sounds all difficult and stuff, but you'd expect that Slashdot readers would at least think about that particular metaphor a little more carefully.
I was in the warehouse for the Birmingham reserve collection last week. There's a lot of old computer iron: essentially most hardware Birmingham council had in the fifties and sixties. There's what appears to be most of a Ferranti Orion there, for example.
Sandwiches are reputedly named after the 4th Earl of Sandwich, not the place in Kent. I don't think there's much relationship between the him and the town in Kent (a county, not a region), any more than there is much relationship between the Duke of Devonshire and Devon (note to tourists: Chatsworth is a long way from Devon, in every way). The 1st Earl, Edward Montagu, received his title for navel derring-do, so presumably received the Earldom of one of the cinque ports as an appropriate name.
The Earls of Sandwich are, on the other hand, who the Sandwich Islands are named for.
So I don't see where I had to buy any hype in order to think that the Battle of the Atlantic could have been won by the Germans if they were just a little bit more careful with their encryption protocols.
There were so many factors its hard to disentangle them.
The breaks into the Triton/Neptune key were important in determining the location of U boats and routing convoys around them, not doubt. But everything came good at the same time. Willow Run was turning out a B24 Liberator every few hours by then, and the VLR long range variants were starting to fill the air gap. Centimetric radar was in production, and meant that surfacing a submarine was extremely risky. Leigh Lights and 60lb rockets and other fairly low-tec solutions were widely deployed, and the Bay of Biscay was covered not just by B24s but by other coastal command aircraft. An escort carrier was being launched every two weeks, providing more and more convoys with local air cover, and the aircraft carried by those ships were becoming more capable.
Ultimately Liberty ships were being launched with such frequency that the U boats even operating unopposed simply didn't carry enough munitions to sink enough, and by then they most certainly weren't operating unopposed.
Germany lost the war because it took on one opponent with unlimited manpower (Russia) and another with unlimited industrial power (USA). Some individual technological developments make a big difference --- probably the cavity magnetron and the Enigma breaks --- but most of the rest of what won the war was good solid manufacturing strength. Little boys admire the fit and finish of the Panther tank, but historians note that the tanks didn't work, were in short supply and most critically had almost no spares provided. Meanwhile, T34s and M4A3s were turned out in such quantity that quality was almost incidental. There were cast, pressed and welded M4 Shermans because that made the best use of the factories available. There were versions with different engines, suspension and guns in part to improve the breed, but also just to make mass production easier. It wasn't the best tank of the war --- too high, not enough armour, undergunned --- but it was available in quantities such that that mattered less.
Germany may have built the weapons that defined the post-war era (Me262, StG43, MG42, the Walther-powered submarines, the V1 and the V2) but unfortunately they couldn't make enough effective planes, tanks and ships. As air superiority was ceded to the US and UK --- Goering said that he knew the war was lost when he saw a P51 over Berlin --- more and more of the German aircraft effort had to go into mass producing point-defence interceptors like the increasingly outclassed Me109 and the rather more successful FW190D `long nose'. There were no bombers, no heavy fighters, no ground attack aircraft of remotely the quality of US, British and even Russian equivalents. By the time the Me262 was available there was so little fuel and so few experienced pilots that it made essentially no difference, and the faster piston fighters of the allies (Tempest V, P51 in its later models) although outclassed had the benefit of unlimited fuel, well trained pilots --- much easier to learn to fly in Canada than in German airspace --- and again huge numbers.
The second world war was won by Russian manpower and American manufacturing genius. Without those, everything else is uncertain. With those, Germany cannot win.
A hundred comments and no-one seems to have mentioned the problem we're seeing in the UK with postal votes: `heads of family' or even `community leaders' using it as an excuse for block voting. Postal voting was made available on demand, rather than requiring a reason, in the UK a few years ago, with the best of intentions. What's happened now is that oppressive fathers, oppressive husbands and in some cases soi-disant `community leaders' are able to force people to apply for a postal vote (or simply apply for a postal vote in their name and rely on their not complaining) and use it themselves. Internet voting has precisely the same problem: I can take my partner's vote, or my children's vote, and use it. We're now seeing horrendous corruption in certain parts of the country --- decency forbids me from saying where, but let's just say ``Inner city areas where every third shops is a sweet centre''.
It makes substantial differences over distances or as rates increase. A lot of our cables worked at 100Mbps but failed on GigE. Stuff that was just a TDR curiosity at 100TX failed badly with GigE rates/
I've had the same phone number for twenty-three years, but it's been ex-directory for the last twenty years. It's ex-directory, although I suspect that unlisted, where it is available for directory enquiries but not published, is as effective given that DQ costs 50p a shot. It's been TPS since the inception of the scheme. I doubt we get more than one marketing call per year.
We have TDR equipment and appropriate tools, but we still buy patch cables in bulk. We tested an assortment of ones we had made with cheap crimping tools, and they were all horrible. We can make decent ones, but it takes longer and costs more than buying them pre-tested.
Indeed, my distaste for Arial caused a colleague to pointedly mail me a document in Helvetica. Which of course, because Windows told him lies, was actually in Arial. So I printed two copies of it: one in Arial, one put correctly back into Helvetica. Everyone I showed the pair to claimed that they were essentially the same, but if asked to express a preference chose the Helvetica version. Every one.
Just done a week in California, back home for three weeks in the UK, then a week in the Pacific Northwest. I saw planes, hotels and offices. That's pretty well it. I'd have been quite happy to have slept in my own bed, avoiding 40 hours on planes and 32 hours of timezone dislocation.
The scene: two good old boys have discovered Doctor Who, and have found that once they've finished lusting after Billie Piper --- it's OK, we all do it here in the UK --- that there are other Doctors and, of course, other assistants.
``Hello, Mrs Dawkins, we've been watching some Tom Baker-era DVDs and we wonder if you'd like to come to Oklahoma. No, no, leave him at home. What, he's booked to come here anyway? Look, why don't you come instead: we'll get your ticket swapped over. No? We can legislate, you know...''
I've never understood the provision of paths from `inside' to `outside' in any work environment. We wash everything through application relays with RFC 1918 on the inside and no NAT. It's not perfect: a _lot_ tunnels through HTTP, for example, and we're fairly permissive with CONNECT to our proxies. But at least we have logs of every connection.
Short answer: it would break their business model.
So long as people behave in the way they expect, yes. The problem now is that their business model is essentially boiling down to ``give us a lot of money, or give us nothing''. Five years ago they could rely on ``a lot''. Now they're getting nothing. By ``us'' I don't mean Microsoft alone, by the way, I mean them and their partners: the Wintel ecosystem assumes that each upgrade is so compelling that it drives not just OS but also application, hardware and infrastructure change. It's not just about a recession: I think Wintel has the same problem in any economy, because XP+Office2003+2GHz+2GB+200GB (ie the computer you bought two or three years ago) is ``good enough'' for most purposes, and today's equivalent isn't compelling in any substantial way.
Apple's in a slightly better place for a recession because it's shipping incremental change, so all an upgrade costs you is the price of the upgrade. But even for them, iWork '09 and iLife '09 look like marginal changes of limited value, and Tiger to Leopard is hardly an introduction to a new world. But at least in Apple Land you can run all the new stuff on all the hardware of the past five years, which is more than can be said for Wintel.
It doesn't. It's purely aimed at static web content. Just because you can't attack the whole problem isn't of itself a reason to not attack parts of the problem. Cleanfeed removes the ``I just stumbled on it while browsing'' argument.
I was at the meeting at Portcullis House at which this was announced yesterday (the media coverage was the usual pre-event trailing). John Carr was there but didn't speak, but the minister made a decent speech. The basic position seems to be that industry agreed to use the Cleanfeed system on residential links in exchange for there not being legislation, but some of the industry is saying no on the grounds of cost and effectiveness (not, notably, on any cyber-libertarian position, although that may be their underlying motive). Those parts of the industry which have followed the Cleanfeed line voluntarily are annoyed about cost and complexity that they are shouldering which their competitors aren't.
My position is that, given that we're not going to be able to avoid the basic problem, legislation is actually not necessarily a bad thing. It would plave the IWF on a legislative footing, which would alter the governance and the contestability in potentially a good way. But people I have immense respect who know a lot more about this stuff disagree, and think the upside (judicial oversight) would not be worth the downside (ministers making positions).
Sadly, it seems that a huge part of the e-crime agenda is being devoted to child porn, which is only one past of the issue and one where the end users aren't the victims. Fraud and other issues are being subsumed.
Except most of the apostrophes in signs in Birmingham are mythical creatures. There hasn't been an apostrophe in Kings Norton or Kings Heath since the 19th century, and probably never --- source, photographs from the DJ Norton collection of station signs. And there hasn't been an apostrophe in those two examples for sure in the fifty five years since my parents moved here. Haywoods Heath, Kitts Green: any set of photographs of trams will show there were no apostrophes in the thirties, and probably well before.
What did XP really bring to the table that Windows 2000 didn't already have?
Availability. Suddenly, every machine in the shop had XP Home, and the slightly more expensive ones had XP Pro. Windows 2000 was barely available down retail channels pre-loaded on anything that the man in the street would be likely to be buying.
Oh, and applications that the man in the street wanted. And hardware support for the equipment he owned.
Yes, we all know that 2000 was Windows 5.0 and XP was Windows 5.1. But that.1 covered a hell of a lot of retail goodness.
I've had the same UK phone number for twenty-two years, and the number of marketing calls of any description I get is close to zero.
I'm in the TPS, of course. I give my work number to anyone I'm a bit dubious about, and I don't give a UK number to any company whose legal entity is outside the UK except under the direst emergency. Remember, the TPS legislation (derived from the European privacy directive) means that the responsibility to obeying the preference service lies with the ultimate client of the marketing, not the organisation that carries it out.
But I've also been ex-directory for most of those twenty-two years. The trick is to be really XD, _not_ UL (unlisted). UL numbers aren't in the phone book, but are available from directory enquiries, and given how open that business is these days I suspect the boundary is pretty permeable. The combination means that the only marketing calls I get are people stretching the prior-relationship angle to breaking point (for whom my existing business is worth more than pissing me off, so they usually get the hint and only do it once --- although I cancelled my BT ISP contract in part because they called me twice around renewal time) or people who don't know my name because they're dialling randomly. The set of calls I get from non-geos, internationals and so on is so small that if I'm not expecting a call they go to the answering machine, as do mobiles I don't recognise.
There was a fascinating article in the Graun a year or so ago (in the Saturday magazine, so not archived online) which talked about how as the TPS gets more traction, the pool of people who are targets for outbound telemarketing gets smaller, so they get more pissed off and sign up for the TPS even if they wouldn't have done at the prior level of junk calls, and so it goes downward. It's hard to get worked up about this: it's not value creating work, if anything the opposite, and the vast majority of outbound marketing is not economically productive. The ultimate end of the line is a good one for all parties concerned.
Netbook versus the early 90s
on
Less Is Moore
·
· Score: 1
Take a low-end Sun of the early 1990s, in the ELC / IPX / Classic direction. It would have a processor clocked at between 25 and 50 MHz, between 4 and 16 MB of RAM and (if not running diskless) a disk of between 140MB and 500MB onboard or hung out the back. If you were a developer or a student and had one of those on your desk, you'd be relatively content. If you step up to a medium-sized Sun of the era, something like a 10/514, you'd have four processors clocked at around 50MHz, perhaps as much as 512MB of RAM and a few GB of disk in multipacks. Networking, of course, was mostly 10Mbps, although you might have FDDI, CDDI or similar around --- we fitted that to some of our 514s.
A Netbook blows any of these away on any metric except, for the cases where you had multiple spindles plus disksuite striping, IO bandwidth. I've got Open Solaris on an elderly P5020 laptop with a gig of RAM, a 100BaseT ethernet connection and a 120GB disk drive and if I wanted to relive the 1990s, it'd do as an excellent compute server. The first Auspex I bought in 1993 will win score on raw IOPS as it spread its 14GB over about twenty spindles. But even then, the little Atom 330 with a pair of 2.5" drives mirrored with ZFS I have at home would probably run it quite close --- the 2GB of RAM (cf the Auspex's 16MB) plus the journalling will help offset the lack of raw disk bandwidth in many scenarios.
Now if in 1994 you had, say, 20-odd ELCs and IPXen plus a couple 10/514 compute-servers driven off an Auspex you'd think you were pretty hot stuff, and plenty of top-quality software got written on less. 20 SSD Netbooks running OpenSolaris (diskless, even!) and my £300 Atom home server (plus another the same for compute, let's say) would be more than the same capability --- and would have faster networking, better graphics, 1GB vs 16MB of RAM per user, etc --- and yet if you presented that as a student lab people would laugh at you. But the whole thing would probably only consume 20W per seat, total, and cost about $300 per seat, total.
But we don't think we have that choice. Perhaps a recession will hammer home the fact that in five years, with computers we now regard as little better than pocket calculators, Xerox PARC redefined modern computing. Meanwhile, with three orders of magnitude more computing power, little of the same import has emerged from a research lab this century. It's not about the computing power.
Of course, if you're using ZFS the checksums should make this less likely: as soon as you start getting back checksums off the disk it's marked bad, even if at a hardware level it purports to be OK.
Other filesystems are available, of course.
Slashdot Mirror
exactly. i have a dell mini 9 fitted with a 64gb ssd and with a slab of pre-preg carbon fibre araldited to the lid to make it more robust. it is literally half the size of the smallest laptop that apple make, for example. It goes into my briefcase unnoticeably, it does all I need to do while travelling (it's just done a 10 day family holiday, archiving pictures, watching iplayer, etc), and is just perfect. Dell now don't make anything as small...
It's a sort of first time as tragedy, second time as farce thing, isn't it?
As I walked into the Yamaha shop in Ginza an hour ago there was a CD player whacking out bloody Susan Boyle massacring John Stewart's Daydream Believer. There should be a law, there really should.
I get really annoyed by people getting this the wrong way around. If something's easy to learn, it has a steep learning curve: your ability rises rapidly over time, repetition or whatever your measure of effort is. If something is difficult, it has a shallow learning curve: your ability increases slowly against time, repetition or whatever. Yes, I know ``steep learning curve'' sounds all difficult and stuff, but you'd expect that Slashdot readers would at least think about that particular metaphor a little more carefully.
I was in the warehouse for the Birmingham reserve collection last week. There's a lot of old computer iron: essentially most hardware Birmingham council had in the fifties and sixties. There's what appears to be most of a Ferranti Orion there, for example.
I take it you're American? A little less hubris, please: how _is_ Robert J Oppenheimer's legacy looking?
Sandwiches are reputedly named after the 4th Earl of Sandwich, not the place in Kent. I don't think there's much relationship between the him and the town in Kent (a county, not a region), any more than there is much relationship between the Duke of Devonshire and Devon (note to tourists: Chatsworth is a long way from Devon, in every way). The 1st Earl, Edward Montagu, received his title for navel derring-do, so presumably received the Earldom of one of the cinque ports as an appropriate name. The Earls of Sandwich are, on the other hand, who the Sandwich Islands are named for.
There were so many factors its hard to disentangle them. The breaks into the Triton/Neptune key were important in determining the location of U boats and routing convoys around them, not doubt. But everything came good at the same time. Willow Run was turning out a B24 Liberator every few hours by then, and the VLR long range variants were starting to fill the air gap. Centimetric radar was in production, and meant that surfacing a submarine was extremely risky. Leigh Lights and 60lb rockets and other fairly low-tec solutions were widely deployed, and the Bay of Biscay was covered not just by B24s but by other coastal command aircraft. An escort carrier was being launched every two weeks, providing more and more convoys with local air cover, and the aircraft carried by those ships were becoming more capable. Ultimately Liberty ships were being launched with such frequency that the U boats even operating unopposed simply didn't carry enough munitions to sink enough, and by then they most certainly weren't operating unopposed. Germany lost the war because it took on one opponent with unlimited manpower (Russia) and another with unlimited industrial power (USA). Some individual technological developments make a big difference --- probably the cavity magnetron and the Enigma breaks --- but most of the rest of what won the war was good solid manufacturing strength. Little boys admire the fit and finish of the Panther tank, but historians note that the tanks didn't work, were in short supply and most critically had almost no spares provided. Meanwhile, T34s and M4A3s were turned out in such quantity that quality was almost incidental. There were cast, pressed and welded M4 Shermans because that made the best use of the factories available. There were versions with different engines, suspension and guns in part to improve the breed, but also just to make mass production easier. It wasn't the best tank of the war --- too high, not enough armour, undergunned --- but it was available in quantities such that that mattered less. Germany may have built the weapons that defined the post-war era (Me262, StG43, MG42, the Walther-powered submarines, the V1 and the V2) but unfortunately they couldn't make enough effective planes, tanks and ships. As air superiority was ceded to the US and UK --- Goering said that he knew the war was lost when he saw a P51 over Berlin --- more and more of the German aircraft effort had to go into mass producing point-defence interceptors like the increasingly outclassed Me109 and the rather more successful FW190D `long nose'. There were no bombers, no heavy fighters, no ground attack aircraft of remotely the quality of US, British and even Russian equivalents. By the time the Me262 was available there was so little fuel and so few experienced pilots that it made essentially no difference, and the faster piston fighters of the allies (Tempest V, P51 in its later models) although outclassed had the benefit of unlimited fuel, well trained pilots --- much easier to learn to fly in Canada than in German airspace --- and again huge numbers. The second world war was won by Russian manpower and American manufacturing genius. Without those, everything else is uncertain. With those, Germany cannot win.
A hundred comments and no-one seems to have mentioned the problem we're seeing in the UK with postal votes: `heads of family' or even `community leaders' using it as an excuse for block voting. Postal voting was made available on demand, rather than requiring a reason, in the UK a few years ago, with the best of intentions. What's happened now is that oppressive fathers, oppressive husbands and in some cases soi-disant `community leaders' are able to force people to apply for a postal vote (or simply apply for a postal vote in their name and rely on their not complaining) and use it themselves. Internet voting has precisely the same problem: I can take my partner's vote, or my children's vote, and use it. We're now seeing horrendous corruption in certain parts of the country --- decency forbids me from saying where, but let's just say ``Inner city areas where every third shops is a sweet centre''.
It makes substantial differences over distances or as rates increase. A lot of our cables worked at 100Mbps but failed on GigE. Stuff that was just a TDR curiosity at 100TX failed badly with GigE rates/
I've had the same phone number for twenty-three years, but it's been ex-directory for the last twenty years. It's ex-directory, although I suspect that unlisted, where it is available for directory enquiries but not published, is as effective given that DQ costs 50p a shot. It's been TPS since the inception of the scheme. I doubt we get more than one marketing call per year.
We have TDR equipment and appropriate tools, but we still buy patch cables in bulk. We tested an assortment of ones we had made with cheap crimping tools, and they were all horrible. We can make decent ones, but it takes longer and costs more than buying them pre-tested.
Indeed, my distaste for Arial caused a colleague to pointedly mail me a document in Helvetica. Which of course, because Windows told him lies, was actually in Arial. So I printed two copies of it: one in Arial, one put correctly back into Helvetica. Everyone I showed the pair to claimed that they were essentially the same, but if asked to express a preference chose the Helvetica version. Every one.
Just done a week in California, back home for three weeks in the UK, then a week in the Pacific Northwest. I saw planes, hotels and offices. That's pretty well it. I'd have been quite happy to have slept in my own bed, avoiding 40 hours on planes and 32 hours of timezone dislocation.
``Hello, Mrs Dawkins, we've been watching some Tom Baker-era DVDs and we wonder if you'd like to come to Oklahoma. No, no, leave him at home. What, he's booked to come here anyway? Look, why don't you come instead: we'll get your ticket swapped over. No? We can legislate, you know...''
ian
I've never understood the provision of paths from `inside' to `outside' in any work environment. We wash everything through application relays with RFC 1918 on the inside and no NAT. It's not perfect: a _lot_ tunnels through HTTP, for example, and we're fairly permissive with CONNECT to our proxies. But at least we have logs of every connection.
So long as people behave in the way they expect, yes. The problem now is that their business model is essentially boiling down to ``give us a lot of money, or give us nothing''. Five years ago they could rely on ``a lot''. Now they're getting nothing. By ``us'' I don't mean Microsoft alone, by the way, I mean them and their partners: the Wintel ecosystem assumes that each upgrade is so compelling that it drives not just OS but also application, hardware and infrastructure change. It's not just about a recession: I think Wintel has the same problem in any economy, because XP+Office2003+2GHz+2GB+200GB (ie the computer you bought two or three years ago) is ``good enough'' for most purposes, and today's equivalent isn't compelling in any substantial way.
Apple's in a slightly better place for a recession because it's shipping incremental change, so all an upgrade costs you is the price of the upgrade. But even for them, iWork '09 and iLife '09 look like marginal changes of limited value, and Tiger to Leopard is hardly an introduction to a new world. But at least in Apple Land you can run all the new stuff on all the hardware of the past five years, which is more than can be said for Wintel.
Except it doesn't block lots of things, etc. The `Virgin Killer' debacle was the first, and so far only, time it's crossed onto the public radar.
It doesn't. It's purely aimed at static web content. Just because you can't attack the whole problem isn't of itself a reason to not attack parts of the problem. Cleanfeed removes the ``I just stumbled on it while browsing'' argument.
My position is that, given that we're not going to be able to avoid the basic problem, legislation is actually not necessarily a bad thing. It would plave the IWF on a legislative footing, which would alter the governance and the contestability in potentially a good way. But people I have immense respect who know a lot more about this stuff disagree, and think the upside (judicial oversight) would not be worth the downside (ministers making positions).
Sadly, it seems that a huge part of the e-crime agenda is being devoted to child porn, which is only one past of the issue and one where the end users aren't the victims. Fraud and other issues are being subsumed.
Except most of the apostrophes in signs in Birmingham are mythical creatures. There hasn't been an apostrophe in Kings Norton or Kings Heath since the 19th century, and probably never --- source, photographs from the DJ Norton collection of station signs. And there hasn't been an apostrophe in those two examples for sure in the fifty five years since my parents moved here. Haywoods Heath, Kitts Green: any set of photographs of trams will show there were no apostrophes in the thirties, and probably well before.
Availability. Suddenly, every machine in the shop had XP Home, and the slightly more expensive ones had XP Pro. Windows 2000 was barely available down retail channels pre-loaded on anything that the man in the street would be likely to be buying.
Oh, and applications that the man in the street wanted. And hardware support for the equipment he owned.
Yes, we all know that 2000 was Windows 5.0 and XP was Windows 5.1. But that .1 covered a hell of a lot of retail goodness.
ian
I've had the same UK phone number for twenty-two years, and the number of marketing calls of any description I get is close to zero. I'm in the TPS, of course. I give my work number to anyone I'm a bit dubious about, and I don't give a UK number to any company whose legal entity is outside the UK except under the direst emergency. Remember, the TPS legislation (derived from the European privacy directive) means that the responsibility to obeying the preference service lies with the ultimate client of the marketing, not the organisation that carries it out. But I've also been ex-directory for most of those twenty-two years. The trick is to be really XD, _not_ UL (unlisted). UL numbers aren't in the phone book, but are available from directory enquiries, and given how open that business is these days I suspect the boundary is pretty permeable. The combination means that the only marketing calls I get are people stretching the prior-relationship angle to breaking point (for whom my existing business is worth more than pissing me off, so they usually get the hint and only do it once --- although I cancelled my BT ISP contract in part because they called me twice around renewal time) or people who don't know my name because they're dialling randomly. The set of calls I get from non-geos, internationals and so on is so small that if I'm not expecting a call they go to the answering machine, as do mobiles I don't recognise. There was a fascinating article in the Graun a year or so ago (in the Saturday magazine, so not archived online) which talked about how as the TPS gets more traction, the pool of people who are targets for outbound telemarketing gets smaller, so they get more pissed off and sign up for the TPS even if they wouldn't have done at the prior level of junk calls, and so it goes downward. It's hard to get worked up about this: it's not value creating work, if anything the opposite, and the vast majority of outbound marketing is not economically productive. The ultimate end of the line is a good one for all parties concerned.
Take a low-end Sun of the early 1990s, in the ELC / IPX / Classic direction. It would have a processor clocked at between 25 and 50 MHz, between 4 and 16 MB of RAM and (if not running diskless) a disk of between 140MB and 500MB onboard or hung out the back. If you were a developer or a student and had one of those on your desk, you'd be relatively content. If you step up to a medium-sized Sun of the era, something like a 10/514, you'd have four processors clocked at around 50MHz, perhaps as much as 512MB of RAM and a few GB of disk in multipacks. Networking, of course, was mostly 10Mbps, although you might have FDDI, CDDI or similar around --- we fitted that to some of our 514s.
A Netbook blows any of these away on any metric except, for the cases where you had multiple spindles plus disksuite striping, IO bandwidth. I've got Open Solaris on an elderly P5020 laptop with a gig of RAM, a 100BaseT ethernet connection and a 120GB disk drive and if I wanted to relive the 1990s, it'd do as an excellent compute server. The first Auspex I bought in 1993 will win score on raw IOPS as it spread its 14GB over about twenty spindles. But even then, the little Atom 330 with a pair of 2.5" drives mirrored with ZFS I have at home would probably run it quite close --- the 2GB of RAM (cf the Auspex's 16MB) plus the journalling will help offset the lack of raw disk bandwidth in many scenarios.
Now if in 1994 you had, say, 20-odd ELCs and IPXen plus a couple 10/514 compute-servers driven off an Auspex you'd think you were pretty hot stuff, and plenty of top-quality software got written on less. 20 SSD Netbooks running OpenSolaris (diskless, even!) and my £300 Atom home server (plus another the same for compute, let's say) would be more than the same capability --- and would have faster networking, better graphics, 1GB vs 16MB of RAM per user, etc --- and yet if you presented that as a student lab people would laugh at you. But the whole thing would probably only consume 20W per seat, total, and cost about $300 per seat, total.
But we don't think we have that choice. Perhaps a recession will hammer home the fact that in five years, with computers we now regard as little better than pocket calculators, Xerox PARC redefined modern computing. Meanwhile, with three orders of magnitude more computing power, little of the same import has emerged from a research lab this century. It's not about the computing power.
ian
Of course, if you're using ZFS the checksums should make this less likely: as soon as you start getting back checksums off the disk it's marked bad, even if at a hardware level it purports to be OK. Other filesystems are available, of course.