It's been going on for 30 years, but it's been going on BADLY for 30 years with totally lame software.
Example: A few years ago I looked at education software for teaching algebra.
The best software I could find, after downloading and examining some 10 packages:
1. Only used a single example to teach a given concept.
2. In the teaching mode the only option was 'Next'
3. In the practice mode only presented 2-3 problems.
4. Had NO provision for showing work. Kid had to do work on scrap paper, then pick one multiple choice answer.
5. The correct answer didn't vary. (Psst, is C the answer to #2? No, it's D"
6. If you guessed 3 times, it told you the answer.
7. It's record keeping consisted solely of recording which kid had completed which module.
I wrote to ALL of the publishers
I told them that an algebra instruction system had to have:
1. Provision to type math as rapidly as you can write it. (Yes, this can take time to practice and learn. You didn't learn to use a pencil overnight, did you?)(FrameMaker for Linux (Beta 5.56?) was good enough that I could type a quiz almost as fast as writing it out by hand, and far more neatly. Not TeX in it's quality, but still pretty good.) 2. Requirement that the teacher can set the number of practice problems. 3. The teacher can set the difficulty of the problems for mastery. 4. The package adjusts difficult to fit the needs of the student. 5. The package records the mistakes of the student for the teacher to examine later looking for patterns. (Lots of kids will have 'bugs' in the algorithms they use. E.g. in subtraction they will subtract the smaller digit from the larger, instead of borrowing) 6. Every instruction module has multiple paths through it. 7. The package is good enough at symbolic manipulaion internally to identify when a kid has made a mistake. (Mathematica as part of the back end?) 8. Constant review. E.g. You don't just practice today's problems, but you also do a few of yesterdays, and a couple from last week. 9. No multiple choice answers. 10. Variable in how much work that needed to be shown, with more steps required for students who were not consistently right, or too slow.
I heard back from ONE of the publishers who said, "Sorry we don't have any plans to offer anything like that."
1. You don't have to work in print runs of 10,000 or so. So your entire catalog can be ready for sale. Way too often I've tried to find an 'out of print' book.
2. Publishers don't make a lot of money off a paper book. Printing, and shipping and commissions to the bookstore take a huge chunk of the price.
Anyway, I'm willing to pay 30% of the paper price for an ebook.
Worked as the main system admin for a dot com company. I got an offer that didn't pay as much, but offered far more flexibility, 12 weeks annual vacation, and an hour shorter commute each way.
I knew that my leaving would disrupt things, so instead of the usual 2 weeks notice, I gave them 6 weeks notice. I suggested that they hire someone, and that I would act as mentor, getting him up to speed. Management decided that Unix sysadmins were a fungible commodity, and it could wait until after the hiring freeze. Two days before my 6 weeks was up, I was fired. (!) It had come to the attention of the president of the company that I was openly critical of their procurement policies. (3 signatures on 3 different forms, and two faxes to head office required to purchase a $100 network card.)
They were so afraid that I had left trapdoors that my workstation was wiped clean, and my directory wiped from the rsync servers. (We did backups every 3 hours using rsync.
Two days later I get a phone call about the setup. I told them that it was all in a 40 page document in my home directory/SysAdmin. They asked how to restore from backup tape. I told them that it was in the Admin documentation file, and that many versions of it were on the rsync server. (Once every 3 hours for the previous 24, one per day for the previous 2 weeks, one per week for the previous 2 months, one per month for the previous 6 months. Not as bad as it sounds. If a file hadn't changed, it was a hard link)
Was I interested in coming in and sorting out the mess? No, I had a new job at 3/4 of the pay, and 3 times the freedom.
Half a year later I met my replacement. He'd lasted for 3 months before he'd been laid off. He said he was on the verge of quitting when he'd been laid off. The management style hadn't changed.
Six months later, the company was a research subsidiary of EMC, and had 12 employees.
The problem isn't greed as much as it is short sightedness
Executives are paid in dividends off a block of stock for 20 years for each year of work. Now they have an incentive to make the company profitable for the long term.
E.g. I get a salary equivalent to a good engineer, plus dividends off of 100,000 shares for 20 years.
Next year, I get my salary, dividends off the first block, and dividends off a second block.
After 20 years, I'm getting my salary, plus dividends off of 20 blocks of stock.
After 25 years, I'm still getting dividends off of 15 blocks of stock...
*** If you accept a directorship in a company then ALL of your personal assets are available for reparations if you screw up.
A director may not sell stock he holds in a company while serving nor for a period of a decade after he stops serving. It's to his advantage to make sure the company is well run. ***
Speculation is hard to regulate.
The only way I can see to limit speculation is to tax assets as opposed to taxing income. Thus, if you aren't using something, you are still going to pay tax on it. And this is only partially successful. Real Estate is taxed already, (property taxes) and it doesn't stop people buying land and sitting on it.
Modified flat tax. There is ONE deduction: A warm body deduction. The basic exemption. It applies to ALL people. So a family of 4 gets 4 deductions if the parents file jointly. Kid deductions are the same size as adult ones.
The basic deduction is equal to 1000 hours a year of minimum wage. Two parents, two kids both parents working 40 hours/week (2000 hours/year) of minimum wage pay NO tax.
FLAT rate for everything above that. This gives the working poor a break.
Corporations don't pay income tax. Corporations pay tax on gross sales.. Flat rate on total sales, probably fairly small. No deductions for the cost of production. If sales tax is 10% and it costs you $901 to produce a $1000 widget, you just lost money.
The huge simplification of tax law puts hundreds of thousands of our best minds to work doing something productive.
Yeah, but the residence time for water vapour in the atmosphere is only about 10 days, whereas CO2 has a residence time ranging from 50 years to thousands depending on who's lies you believe. (Crucial factor is CO2 exchange between ocean surface and deep water.)
Water vapour acts an an amplifier for other effects. Increased temp increases the amount of water vapour in the air, which increases the temp some more. But it's an exponential series, with an exponent less than one. At least for now.
I expect that in 500 million years, or less we'll be able to move the damn planet.
Let's see: To compensate for a 30% brightness increase we have to move it less than 15% further from the sun. Roughly 13,000,000 miles. If we postpone starting for say, 110 million years, that leaves only 390 million years to get things done. 1/3 of a mile per year.
Or we can break enough chunks off the moon to make a screen.
Or we can increase the overall albedo of the earth.
Vote buying is a separate issue. The issue here is vote tampering. I vote for A but the system records me as voting for B or not voting.
The QR codes are a good idea. Take it further:
You get a number when you go to the poll. The number is on your ballot.
The number is used as a salt for a hash function. The number plus the candidate you vote for is hashed, and the result converted to a QR code. Since voting takes seconds anyway, it can be an expensive hash function.
NOW having a picture of the QR code doesn't show how you voted.
Meanwhile, every poll has a web page that lists the numbers of the ballots cast at that poll. Clicking on a number tells you what vote was recorded for that number. This allows statistical sampling of the vote.
Part of the system gives you a token with your ballot number on it. This allows you to check that your ballot was recorded correctly.
1. The 'full' capacity of my electrocar is a multiple of my normal commute. E.g. My normal town trip is 260 km. If I can be guaranteed 260 km range left in my vehicle, I'd consider this.
2. We have some battery that doesn't wear out after 500 charge/discharge cycles. I'd love it if EEStor had a real product.
In a rural setting I've got 40-50 power failures a year. Most of these are under 5 seconds long -- just enough to reset the microwave, stove, alarm clock. We get a 12 hour one once a year, and a 1 hour one 3-4 times a year. I would love to have a house wide backup system to keep the sump pump running when we are away on holiday.
Even at 5%, this is a useful feature for the grid. Right now the coal plants have to run a 2% surplus of steam power so they can ramp up to meet increased demand. We have a local hydro dam, only 60 MW, that dumps 80% of it's flow during ramp up times. If you are in the turbine room, the turbine gates change from idle to an appreciable fraction of full several times a minute.
I'm a little shaky about this Post-PC future. I have a kid who works for me on my farm. He has just bought himself a new PC. He's a gamer, and wanted something with more zip. Comes with a random generic vid card, which he will swap with the better vid card he's got already in his old PC.
I've looked at netbooks, and have considered an iPad, Macbook Air, but at the end of the day, my work pattern is to have 9 octillion windows open at a time.
As to doing your work in the cloud, forget it. Even doing work over a LAN using RDP or remote X-windows is like sucking golf balls through a garden hose as soon as you get to something with significant graphics.
My internet speed at it's best is 1/50 my local network speed; and is 1/3000 my peak local disk speed ( 2Mbit/s vs 100 Mbit/s vs 6 Gbit/s I don't think the PC is any where close to being 'Post'
They won't be as common. I imagine that the future heavy computer user will have both a desktop and a portable machine; a few people, like my father-in-law will use only an iPad, And there will be another bunch who use a computer at work, and have little interest in doing anything else.
On one hand I'm not worried. At least not about the near future (2-3 decades) Robots are getting better, but their perception of the world is still erratic. As a simple task, consider a 'robot mule' I want to put my prospecting gear on the back of this mule, and have it follow me, 6 feet behind as I walk a mountain trail, or a desert ravine.
Consider a robot gardener, one that can properly plant a petunia or prune a potentilla.
On the other hand, where the skill is knowledge based, not perception based there is some prospect for massive take over by robots/AI
Generally the progress of automation has been to create new jobs elsewhere. All fine and dandy, but the new jobs have tended to be tech jobs instead of unskilled or artisan jobs. One power loom could out-produce hundreds of hand looms. But clothing got cheap enough that there was no market for second hand clothing. And that power loom took 3-4 operators, and likely a techniician, letting a few hundred other people starve, or find a new way to make a living.
Speak of education all you want, but we already have large numbers of people who cannot/will not do any job that requires greater skill than burger flipping. What happens when only 10% of the people can find meaningful work?
I have witnessed a society in this situation. Most indian reserves in Canada. Drug abuse, alcohol abuse, wife and child abuse, widespread STDs, unintended pregnancies. Not a place I want to live.
At present the trades are places of refuge. Many trades are still composed of jobs that are 'one off'. A general purpose robot that can replace a toilet is still a good ways off.
Telus, our local POTS provider runs these cute ads on TV flogging their high speed internet. About twice a year I go to their web site and plunk in my phone number to see what package I can buy.
At present, the best they can do for me is...
56kbit dialup.
And from a few friends, they can't support this becuase of line noise, it it usually runs about 40kbit.
So I still have to use a satellite link with its 200 watt continuous power consumption (on the power supply -- may be peak only) and 800 to 1600 ms latency.
I agree that it would be a good idea for a modular home technique, probably using a larger module. This one uses a 2.4 meter side as the basic square -- not quite 8 feet. A living room that is 8 feet square is a bit cozy. A dining room 8 feet square means that with a 36" table (card table) you have 30 inches on all sides for chairs. This house would be very much like living on board a small boat.
I built my own home on a 24 foot square base. Even that had tight spaces in it, and the ground floor had far fewer walls.
The present plan is for people who don't need/want privacy. Bamboo screen partitions are not much of a block to sound. And there are no hallways. Each room connects to it's neighboring room.
PP has it right that this is a warm climate house. It would not be a pleasant house in a -40 Canadian winter.
As a modular home it has merit. Here, you can move a module 16 feet wide with pilot cars, 12 feet wide with a flasher. A 12 x 24 or a 16x32 module would work nicely for a square house. Increase the module's length to 12x30 and even a module on the north side can have a south facing window.
Give it a shed roof so that snow slides to the outside. An option for a 'courtyard cap' gives you another room. Glaze it, and you have an interior greenhouse, although one that doesn't get a lot of light in winter.
Put a vent in the courtyard cap, and you can barbecue year round.
I use computers to get work done. I want no more than a significant release every two years, one that is well out of beta, with no bugs that I can casually find in the first 10 minutes.
Security fixes yes. Major bug fixes, yes. These are minor releases. But a minor release should NEVER change the API for extensions.
I'll go further: APIs should NEVER change. If there is need, you change the name, provide both APIs for a full (or more) major release cycle, meanwhile, announce to the developers that certain APIs will be discontinued, and give them a couple years to do it.
Worst case this means there is a version of the extension for each Major release, which if they are reasonably infrequent is not onerous to do.
So can you get better performance with Intel chips by bypassing the old crufty instruction set? If so, then just redoing the system libraries of the OS might make a major difference in overall performance.
Can a compiler be set to produce 'universal' binaries that can fall back to CISC instructions, but detect and execute faster instructions when available?
Brings new meaning to "blazingly fast performance" Or the latest 'hot thing.'
And firewall...
And daemons...
Of course an OS that supports migrating processes from one machine to another will need some form of checking out the imported process for malware. Purgatory?
It depends on the pace and the pixel size. Obviously, the 'pixel' of concrete has to be larger than the largest aggregate. So you plop down 2" blobs of stiff concrete.
Slump of concrete varies with the stiffness of the mix. The machines that continuously cast curbs seem to be able to leave behind an 8" tall layer of concrete that is solid enough to cure.
So visualize a 'printer' that casts whatever wide by 6" tall layer at a time. Horizontal rebar is placed on top of a layer, and embedded on the next pass. Vertical rebar is held while the machine makes it's pass. Probably short lengths overlapping.
Such a printer would be more like 'turtle graphics' than like raster images.
It would work at a speed that allowed the previous layer to have started to set, but not cure, so that the next layer would bond to it, much like the continuous cast of a large dam is done.
At this point I don't see how to get clean breaks (door, window openings)
But visualize a simple machine that builds a concrete silo by going round and round, adding 6" every two hours.
To calculate density they need both mass and volume. Mass is in principle calculated by the wobble of the source star, and assumes that you have an accurate measure of it's mass. However circling a binary, the wobble is that of the pair. The uncertainty of this is substantial.
To calculate the volume, you need the diameter. This would require having a reasonable curve of light drop as the disk of the planet transits one of the stars. Is Kepler sensitive enough to pick out the effect of the shadow of a single planet on the solar disk? E.g. if the planet is 100,000 km in diameter, and moves at 10 km/second then time from first contact to occlusion would be 10000 seconds. If the star itself is 1,000,000 km in diameter then the planet's disk represents 1% of the star's area, and that will be the decrease in illumination in transit. However to guage the time of first contact to occulsion, you are working with small chords overlapping. Which means that you need to record variations in output of something like 3 orders of magnitude smaller.
The orbital speed is also subject to error. A far out planet will not be observed for even a single orbit. So you have to deduce speed from the speed change of the orbited star. Is dopler that good.
A 10% error in diameter means a 30% error in volume. And I'm skeptical of even getting within 10%.
Local is better unless the computation/data ratio is very large. If it isn't you pay for moving data sets too and from your cloud, and you pay for storage time at the cloud.
Some notions for you:
1. I once spent a day analyizing benchmarks on Tom's Hardware, and found that Front side bus was a bigger determinant of benchmark performance than was CPU speed. Using the 2nd or 3rd fastest CPU in order to use a faster FSB was always a win.
2. Cache size is critical. If you can optimize your software to run in the processor's cache, your performance goes up dramatically. At one place I worked, one researcher was doing computational theoretical physics. The 200 MHz pentium was getting 25% of the performance of the Stardent Titan 4 core machine that cost 40 times as much. But when he made the inner loop just a bit bigger (he had to -- now he was looking at rotating holes) the performance fell by a factor of 20. The titan with it's much larger cache barely slowed down.
3. Another place had a Beowulf cluster for doing protein folding. It consisted of 60 machines. During the day they were windows boxes used as student labs. At 5 p.m. they rebooted as linux boxes, and picked up the computation where they left off that morning. The cluster's name was werewulf because it only came out at night.
This is one way to defray the costs if the researcher is part of an educational institution.
4. If a machine is going to boot frequently or start code frequently, an SSD disk helps a lot.
5. Don't forget to calculate your cooling requirements. A large number of CPUs working hard generates a lot of heat.
6. Beware of debugging costs. We had a 64 core machine once (Myrias SPS-3) that during the day we broke up into 4 single core queues, 1 4 core queue, and two larger queues. Process dispatch was expensive on this machine. The rules for the researchers was that they had to get their software working correctly on a single core, then get the parallelism working on 4 cores, before they could run on the larger two blocks. Researchers had their wrists slapped if they ran a job on 32 cores, and it aborted 30 seconds after starting. At night and weekends we ran either a sngle 64 queue, or if some researcher was working late, we'd run a 60 and a 4 or 60 and 4 singles.
The better solution would be to run a lead to the case, and have a button that you had to press down while the BIOS was being flashed.
[Semi-off topic digressesion follows.]
In general 'read only' is your security friend. I tried intermittently to figure out a FreeBSD hack that would allow me to boot a production server as read only on the root partition. OpenBSD has the ability to mark files as immutable when in security level 2.
I like the idea that I can have a set of programs that cannot be compromised by malware. I once had a linux intrusion that rewrote ls to hide directories that had names... (three dots) It also would show normal sizes of ps, which was hacked to not show certain processes. We were scratching our heads for a while. They hadn't hacked lsof or find however.
A better model for a secure system:
1. Certain directories are not writable in normal operating mode./bin,/sbin/lib are a good start.
2. No program that requires elevated privledge can run from a writable directory.
3. The chmod system call cannot mark a program set{ug)id in normal operating mode since any such program is in a directory that is un-writable.
4. Data is not executable. Executable code is not modifyable in memory. Don't know how much of a restriction this makes.
5. To make a change you have to lower the operating status of the OS to 'insecure' In practice this should mean similar to single user mode, with a set of parameters that can be chosen by the operator. E.g. in security change mode, the computer cannot route outside the local subnet. Or if truely paranoid, has no network connection at all. No background processes run in SC mode.
This doesn't make the computer secure. Lots of bad thinks can be done by programs running in user space. But this gives you a kit of trusted tools that the OS can use to examine the running processes.
ancestor poster said that sunlight is limiting in rainforests. Perhaps under the top canopy, but I'm skeptical about it being limiting at the top of the rain forest.
In my bio class, many years ago, so the guestimates may have changed, the prof said that few plants increase photosythesis at all above insolation levels of about 25%. Of course in a Rain Forest it's frequently heavily cloudy. Perhaps on cloudy days, light is limiting. Hmm.
For most plants CO2 is limiting. Commercial greenhouses will run CO2 up to levels of 10000 ppm (1%)
Temperature is often limiting too. Lots of plants basically shut down at temps above 90 F, by closing their stomata to reduce water loss. Sure, in a rain forest there is lots of water, but you have to pump that water from the roots to the leaves. Water loss is highly non-linear with temp, so at some point you can't generate enough energy to replace the water.
Reducing the amount of sunlight is unlikely to reduce photosynthesis directly, unless it forms a lot of dense cloud.
***
One Discover channel doc pointed out that the amount of aerosols produced by China and India has masked the warming substantially. And with increasing prosperity, these countries are cleaning up their air. It is likely that we are going to see a rebound effect with much faster global warming for a year or so.
***
Ancenstral poster commented about melting ice caps reducing the temperature difference the drives the winds. Possibly, but not by much. The equator still gets a lot more radiation per square meter than the polar regions. They are also a lot larger. Melted north polar ice cap reveals a lot of water, which with low angles of incidence is still a pretty good reflector. However vapour pressure of water/air is large compared to ice/air. This may result in warmer, wetter sub-arctic regions. The circumpolar tundra's climate may become more like Sweden's with much heavier snowfall. (Much of the tundra is cold desert -- less than 10 inches annual precipitation)
Back to wind: The equator receives about 1.4 times as much radiation as does latitude 45. If mass transport didn't haul the heat away, it would have to radiate. To radiate 1.4 times as much heat it would have to increase in temp by a factor of the 4th root of 1.4, or about 1.09. Increasing the absolute temp by 9% works out to about 26 C temp rise.
Obviously this model is very flawed. I've ignored a bunch of things. Long before the temp rises that much you will get convection cells. They just run at a higher temp.
There are several factors that make pruning the bloat difficult for the average user:
1. Spurious dependencies. I have on occasion installed a command line utility that called in the entire KDE source tree and did a build world.
2. Lack of documentation. Much of the time there is no good documentation outside of the program in terms of how to use it. You have to install it to test it, then uninstalling reveals that 9 billion things will break
3. Proliferation of libraries. I tell you Linux is getting as bad as windows with it's DLLs.
4. Incompatibility across upgrades. This is the flip side. Do an upgrade, and foo-3.2.11.lib changes the syntax for the bar call compared to foo-3.2.10.lib (of which both on install were symlinked to foo.lib...) Come on guys. Minor revisions should not change syntax.
5. Spurious change at the OS level. My computer is a tool. I use it to get work done. Changes for the sake of eye-candy is anathema. The visible part of a distribution should go unchanged for many years at a time.
6. Security fixes and bug fixes should be segregated from feature changes.
Steve Jobs, when running NeXT commented that people won't change for a 10% improvement. Or even 30%. People *will* change for a 500% improvement.
If you want to sell Linux to the masses:
1. It has to stop interfering with how things work. 2. It has to have good, clean, fast applications that are easy to learn to use, and incredibly stable.
There may be little problem with newer motherboards, but I can tell you there is a lot of combinations of stuff out there that won't work with a 3TB drive.
I just got burned on this. Had a not very old 500 GB Lacie external drive. The 500 GB drive in it was getting noisy, so I bought a Hitachi 3 TB drive, popped the case, and swapped drives. My Mac recognized it as a 801 GB drive. WTF!
A couple hours later I knew more about LBA. (Hitachi has good info on their web site.)
You need an OS that supports 48 bit LBA. You need drivers that support 48 bit LBA. You need adapter cards that support 48 bit LBA.
In my case the Lacie is a multi protocol box, so IT has firmware. And that firmware does not support 48 bit LBA, so bit 33 of the capacity is stripped off and I see 1 TB - 2 TiB -1 = 801 GB
I find it amusing.
3 TB for $150 bucks. 50 bucks a TB. My first hard drive added 1200 to the price of the computer. It was 10 MB and even that had to be logically divided into 8 chunks to be addressable by the 2 MHz Zilog Z-80.
I remember when CHS limitations restricted disks to 32 MB. There have been a series of limits since. Some of the limits were clearly stopgaps for a short respite, (We'll write bigger sectors and get a factor of 8) But several have been on the basis of "This should fix the problem for a few decades."
ZFS uses 64 bit addressing. Bets on how long before the first company ships a disk that won't address it.
Slashdot Mirror
Go onto an Indian Reserve if you want to see a society that is driven by welfare.
It's been going on for 30 years, but it's been going on BADLY for 30 years with totally lame software.
Example: A few years ago I looked at education software for teaching algebra.
The best software I could find, after downloading and examining some 10 packages:
1. Only used a single example to teach a given concept.
2. In the teaching mode the only option was 'Next'
3. In the practice mode only presented 2-3 problems.
4. Had NO provision for showing work. Kid had to do work on scrap paper, then pick one multiple choice answer.
5. The correct answer didn't vary. (Psst, is C the answer to #2? No, it's D"
6. If you guessed 3 times, it told you the answer.
7. It's record keeping consisted solely of recording which kid had completed which module.
I wrote to ALL of the publishers
I told them that an algebra instruction system had to have:
1. Provision to type math as rapidly as you can write it. (Yes, this can take time to practice and learn. You didn't learn to use a pencil overnight, did you?)(FrameMaker for Linux (Beta 5.56?) was good enough that I could type a quiz almost as fast as writing it out by hand, and far more neatly. Not TeX in it's quality, but still pretty good.)
2. Requirement that the teacher can set the number of practice problems.
3. The teacher can set the difficulty of the problems for mastery.
4. The package adjusts difficult to fit the needs of the student.
5. The package records the mistakes of the student for the teacher to examine later looking for patterns. (Lots of kids will have 'bugs' in the algorithms they use. E.g. in subtraction they will subtract the smaller digit from the larger, instead of borrowing)
6. Every instruction module has multiple paths through it.
7. The package is good enough at symbolic manipulaion internally to identify when a kid has made a mistake. (Mathematica as part of the back end?)
8. Constant review. E.g. You don't just practice today's problems, but you also do a few of yesterdays, and a couple from last week.
9. No multiple choice answers.
10. Variable in how much work that needed to be shown, with more steps required for students who were not consistently right, or too slow.
I heard back from ONE of the publishers who said, "Sorry we don't have any plans to offer anything like that."
On top of that an ebook:
1. You don't have to work in print runs of 10,000 or so. So your entire catalog can be ready for sale. Way too often I've tried to find an 'out of print' book.
2. Publishers don't make a lot of money off a paper book. Printing, and shipping and commissions to the bookstore take a huge chunk of the price.
Anyway, I'm willing to pay 30% of the paper price for an ebook.
Worked as the main system admin for a dot com company. I got an offer that didn't pay as much, but offered far more flexibility, 12 weeks annual vacation, and an hour shorter commute each way.
I knew that my leaving would disrupt things, so instead of the usual 2 weeks notice, I gave them 6 weeks notice. I suggested that they hire someone, and that I would act as mentor, getting him up to speed. Management decided that Unix sysadmins were a fungible commodity, and it could wait until after the hiring freeze. Two days before my 6 weeks was up, I was fired. (!) It had come to the attention of the president of the company that I was openly critical of their procurement policies. (3 signatures on 3 different forms, and two faxes to head office required to purchase a $100 network card.)
They were so afraid that I had left trapdoors that my workstation was wiped clean, and my directory wiped from the rsync servers. (We did backups every 3 hours using rsync.
Two days later I get a phone call about the setup. I told them that it was all in a 40 page document in my home directory/SysAdmin. They asked how to restore from backup tape. I told them that it was in the Admin documentation file, and that many versions of it were on the rsync server. (Once every 3 hours for the previous 24, one per day for the previous 2 weeks, one per week for the previous 2 months, one per month for the previous 6 months. Not as bad as it sounds. If a file hadn't changed, it was a hard link)
Was I interested in coming in and sorting out the mess? No, I had a new job at 3/4 of the pay, and 3 times the freedom.
Half a year later I met my replacement. He'd lasted for 3 months before he'd been laid off. He said he was on the verge of quitting when he'd been laid off. The management style hadn't changed.
Six months later, the company was a research subsidiary of EMC, and had 12 employees.
The problem isn't greed as much as it is short sightedness
Executives are paid in dividends off a block of stock for 20 years for each year of work. Now they have an incentive to make the company profitable for the long term.
E.g. I get a salary equivalent to a good engineer, plus dividends off of 100,000 shares for 20 years.
Next year, I get my salary, dividends off the first block, and dividends off a second block.
After 20 years, I'm getting my salary, plus dividends off of 20 blocks of stock.
After 25 years, I'm still getting dividends off of 15 blocks of stock...
***
If you accept a directorship in a company then ALL of your personal assets are available for reparations if you screw up.
A director may not sell stock he holds in a company while serving nor for a period of a decade after he stops serving. It's to his advantage to make sure the company is well run.
***
Speculation is hard to regulate.
The only way I can see to limit speculation is to tax assets as opposed to taxing income. Thus, if you aren't using something, you are still going to pay tax on it. And this is only partially successful. Real Estate is taxed already, (property taxes) and it doesn't stop people buying land and sitting on it.
Modified flat tax.
There is ONE deduction: A warm body deduction. The basic exemption. It applies to ALL people. So a family of 4 gets 4 deductions if the parents file jointly. Kid deductions are the same size as adult ones.
The basic deduction is equal to 1000 hours a year of minimum wage. Two parents, two kids both parents working 40 hours/week (2000 hours/year) of minimum wage pay NO tax.
FLAT rate for everything above that. This gives the working poor a break.
Corporations don't pay income tax. Corporations pay tax on gross sales.. Flat rate on total sales, probably fairly small. No deductions for the cost of production. If sales tax is 10% and it costs you $901 to produce a $1000 widget, you just lost money.
The huge simplification of tax law puts hundreds of thousands of our best minds to work doing something productive.
Yeah, but the residence time for water vapour in the atmosphere is only about 10 days, whereas CO2 has a residence time ranging from 50 years to thousands depending on who's lies you believe. (Crucial factor is CO2 exchange between ocean surface and deep water.)
Water vapour acts an an amplifier for other effects. Increased temp increases the amount of water vapour in the air, which increases the temp some more. But it's an exponential series, with an exponent less than one. At least for now.
Hang the disks on fishing line. Best if you have several vintages of disks. Add some old table saw blades for a different tone.
I expect that in 500 million years, or less we'll be able to move the damn planet.
Let's see: To compensate for a 30% brightness increase we have to move it less than 15% further from the sun. Roughly 13,000,000 miles.
If we postpone starting for say, 110 million years, that leaves only 390 million years to get things done. 1/3 of a mile per year.
Or we can break enough chunks off the moon to make a screen.
Or we can increase the overall albedo of the earth.
Politicians buy your votes with promises.
Vote buying is a separate issue. The issue here is vote tampering. I vote for A but the system records me as voting for B or not voting.
The QR codes are a good idea. Take it further:
You get a number when you go to the poll. The number is on your ballot.
The number is used as a salt for a hash function. The number plus the candidate you vote for is hashed, and the result converted to a QR code. Since voting takes seconds anyway, it can be an expensive hash function.
NOW having a picture of the QR code doesn't show how you voted.
Meanwhile, every poll has a web page that lists the numbers of the ballots cast at that poll. Clicking on a number tells you what vote was recorded for that number. This allows statistical sampling of the vote.
Part of the system gives you a token with your ballot number on it. This allows you to check that your ballot was recorded correctly.
,,, if:
1. The 'full' capacity of my electrocar is a multiple of my normal commute. E.g. My normal town trip is 260 km. If I can be guaranteed 260 km range left in my vehicle, I'd consider this.
2. We have some battery that doesn't wear out after 500 charge/discharge cycles. I'd love it if EEStor had a real product.
In a rural setting I've got 40-50 power failures a year. Most of these are under 5 seconds long -- just enough to reset the microwave, stove, alarm clock. We get a 12 hour one once a year, and a 1 hour one 3-4 times a year. I would love to have a house wide backup system to keep the sump pump running when we are away on holiday.
Even at 5%, this is a useful feature for the grid. Right now the coal plants have to run a 2% surplus of steam power so they can ramp up to meet increased demand. We have a local hydro dam, only 60 MW, that dumps 80% of it's flow during ramp up times. If you are in the turbine room, the turbine gates change from idle to an appreciable fraction of full several times a minute.
I'm a little shaky about this Post-PC future. I have a kid who works for me on my farm. He has just bought himself a new PC. He's a gamer, and wanted something with more zip. Comes with a random generic vid card, which he will swap with the better vid card he's got already in his old PC.
I've looked at netbooks, and have considered an iPad, Macbook Air, but at the end of the day, my work pattern is to have 9 octillion windows open at a time.
As to doing your work in the cloud, forget it. Even doing work over a LAN using RDP or remote X-windows is like sucking golf balls through a garden hose as soon as you get to something with significant graphics.
My internet speed at it's best is 1/50 my local network speed; and is 1/3000 my peak local disk speed ( 2Mbit/s vs 100 Mbit/s vs 6 Gbit/s
I don't think the PC is any where close to being 'Post'
They won't be as common. I imagine that the future heavy computer user will have both a desktop and a portable machine; a few people, like my father-in-law will use only an iPad, And there will be another bunch who use a computer at work, and have little interest in doing anything else.
On one hand I'm not worried. At least not about the near future (2-3 decades) Robots are getting better, but their perception of the world is still erratic. As a simple task, consider a 'robot mule' I want to put my prospecting gear on the back of this mule, and have it follow me, 6 feet behind as I walk a mountain trail, or a desert ravine.
Consider a robot gardener, one that can properly plant a petunia or prune a potentilla.
On the other hand, where the skill is knowledge based, not perception based there is some prospect for massive take over by robots/AI
Generally the progress of automation has been to create new jobs elsewhere. All fine and dandy, but the new jobs have tended to be tech jobs instead of unskilled or artisan jobs. One power loom could out-produce hundreds of hand looms. But clothing got cheap enough that there was no market for second hand clothing. And that power loom took 3-4 operators, and likely a techniician, letting a few hundred other people starve, or find a new way to make a living.
Speak of education all you want, but we already have large numbers of people who cannot/will not do any job that requires greater skill than burger flipping. What happens when only 10% of the people can find meaningful work?
I have witnessed a society in this situation. Most indian reserves in Canada. Drug abuse, alcohol abuse, wife and child abuse, widespread STDs, unintended pregnancies. Not a place I want to live.
At present the trades are places of refuge. Many trades are still composed of jobs that are 'one off'. A general purpose robot that can replace a toilet is still a good ways off.
Telus, our local POTS provider runs these cute ads on TV flogging their high speed internet. About twice a year I go to their web site and plunk in my phone number to see what package I can buy.
At present, the best they can do for me is...
56kbit dialup.
And from a few friends, they can't support this becuase of line noise, it it usually runs about 40kbit.
So I still have to use a satellite link with its 200 watt continuous power consumption (on the power supply -- may be peak only) and 800 to 1600 ms latency.
The advantages of rural life.
I agree that it would be a good idea for a modular home technique, probably using a larger module. This one uses a 2.4 meter side as the basic square -- not quite 8 feet. A living room that is 8 feet square is a bit cozy. A dining room 8 feet square means that with a 36" table (card table) you have 30 inches on all sides for chairs. This house would be very much like living on board a small boat.
I built my own home on a 24 foot square base. Even that had tight spaces in it, and the ground floor had far fewer walls.
The present plan is for people who don't need/want privacy. Bamboo screen partitions are not much of a block to sound. And there are no hallways. Each room connects to it's neighboring room.
PP has it right that this is a warm climate house. It would not be a pleasant house in a -40 Canadian winter.
As a modular home it has merit. Here, you can move a module 16 feet wide with pilot cars, 12 feet wide with a flasher. A 12 x 24 or a 16x32 module would work nicely for a square house. Increase the module's length to 12x30 and even a module on the north side can have a south facing window.
Give it a shed roof so that snow slides to the outside. An option for a 'courtyard cap' gives you another room. Glaze it, and you have an interior greenhouse, although one that doesn't get a lot of light in winter.
Put a vent in the courtyard cap, and you can barbecue year round.
I really really really hate software releases.
I use computers to get work done. I want no more than a significant release every two years, one that is well out of beta, with no bugs that I can casually find in the first 10 minutes.
Security fixes yes. Major bug fixes, yes. These are minor releases. But a minor release should NEVER change the API for extensions.
I'll go further: APIs should NEVER change. If there is need, you change the name, provide both APIs for a full (or more) major release cycle, meanwhile, announce to the developers that certain APIs will be discontinued, and give them a couple years to do it.
Worst case this means there is a version of the extension for each Major release, which if they are reasonably infrequent is not onerous to do.
So can you get better performance with Intel chips by bypassing the old crufty instruction set? If so, then just redoing the system libraries of the OS might make a major difference in overall performance.
Can a compiler be set to produce 'universal' binaries that can fall back to CISC instructions, but detect and execute faster instructions when available?
Inferno. Gotta like it.
Brings new meaning to "blazingly fast performance" Or the latest 'hot thing.'
And firewall...
And daemons...
Of course an OS that supports migrating processes from one machine to another will need some form of checking out the imported process for malware. Purgatory?
Is code 'blessed' or 'damned'
It depends on the pace and the pixel size. Obviously, the 'pixel' of concrete has to be larger than the largest aggregate. So you plop down 2" blobs of stiff concrete.
Slump of concrete varies with the stiffness of the mix. The machines that continuously cast curbs seem to be able to leave behind an 8" tall layer of concrete that is solid enough to cure.
So visualize a 'printer' that casts whatever wide by 6" tall layer at a time. Horizontal rebar is placed on top of a layer, and embedded on the next pass. Vertical rebar is held while the machine makes it's pass. Probably short lengths overlapping.
Such a printer would be more like 'turtle graphics' than like raster images.
It would work at a speed that allowed the previous layer to have started to set, but not cure, so that the next layer would bond to it, much like the continuous cast of a large dam is done.
At this point I don't see how to get clean breaks (door, window openings)
But visualize a simple machine that builds a concrete silo by going round and round, adding 6" every two hours.
To calculate density they need both mass and volume. Mass is in principle calculated by the wobble of the source star, and assumes that you have an accurate measure of it's mass. However circling a binary, the wobble is that of the pair. The uncertainty of this is substantial.
To calculate the volume, you need the diameter. This would require having a reasonable curve of light drop as the disk of the planet transits one of the stars. Is Kepler sensitive enough to pick out the effect of the shadow of a single planet on the solar disk? E.g. if the planet is 100,000 km in diameter, and moves at 10 km/second then time from first contact to occlusion would be 10000 seconds. If the star itself is 1,000,000 km in diameter then the planet's disk represents 1% of the star's area, and that will be the decrease in illumination in transit. However to guage the time of first contact to occulsion, you are working with small chords overlapping. Which means that you need to record variations in output of something like 3 orders of magnitude smaller.
The orbital speed is also subject to error. A far out planet will not be observed for even a single orbit. So you have to deduce speed from the speed change of the orbited star. Is dopler that good.
A 10% error in diameter means a 30% error in volume. And I'm skeptical of even getting within 10%.
Local is better unless the computation/data ratio is very large. If it isn't you pay for moving data sets too and from your cloud, and you pay for storage time at the cloud.
Some notions for you:
1. I once spent a day analyizing benchmarks on Tom's Hardware, and found that Front side bus was a bigger determinant of benchmark performance than was CPU speed. Using the 2nd or 3rd fastest CPU in order to use a faster FSB was always a win.
2. Cache size is critical. If you can optimize your software to run in the processor's cache, your performance goes up dramatically. At one place I worked, one researcher was doing computational theoretical physics. The 200 MHz pentium was getting 25% of the performance of the Stardent Titan 4 core machine that cost 40 times as much. But when he made the inner loop just a bit bigger (he had to -- now he was looking at rotating holes) the performance fell by a factor of 20. The titan with it's much larger cache barely slowed down.
3. Another place had a Beowulf cluster for doing protein folding. It consisted of 60 machines. During the day they were windows boxes used as student labs. At 5 p.m. they rebooted as linux boxes, and picked up the computation where they left off that morning. The cluster's name was werewulf because it only came out at night.
This is one way to defray the costs if the researcher is part of an educational institution.
4. If a machine is going to boot frequently or start code frequently, an SSD disk helps a lot.
5. Don't forget to calculate your cooling requirements. A large number of CPUs working hard generates a lot of heat.
6. Beware of debugging costs. We had a 64 core machine once (Myrias SPS-3) that during the day we broke up into 4 single core queues, 1 4 core queue, and two larger queues. Process dispatch was expensive on this machine. The rules for the researchers was that they had to get their software working correctly on a single core, then get the parallelism working on 4 cores, before they could run on the larger two blocks. Researchers had their wrists slapped if they ran a job on 32 cores, and it aborted 30 seconds after starting. At night and weekends we ran either a sngle 64 queue, or if some researcher was working late, we'd run a 60 and a 4 or 60 and 4 singles.
Agreed, sort of.
The better solution would be to run a lead to the case, and have a button that you had to press down while the BIOS was being flashed.
[Semi-off topic digressesion follows.]
In general 'read only' is your security friend. I tried intermittently to figure out a FreeBSD hack that would allow me to boot a production server as read only on the root partition. OpenBSD has the ability to mark files as immutable when in security level 2.
I like the idea that I can have a set of programs that cannot be compromised by malware. I once had a linux intrusion that rewrote ls to hide directories that had names ... (three dots) It also would show normal sizes of ps, which was hacked to not show certain processes. We were scratching our heads for a while. They hadn't hacked lsof or find however.
A better model for a secure system:
1. Certain directories are not writable in normal operating mode. /bin, /sbin /lib are a good start.
2. No program that requires elevated privledge can run from a writable directory.
3. The chmod system call cannot mark a program set{ug)id in normal operating mode since any such program is in a directory that is un-writable.
4. Data is not executable. Executable code is not modifyable in memory. Don't know how much of a restriction this makes.
5. To make a change you have to lower the operating status of the OS to 'insecure' In practice this should mean similar to single user mode, with a set of parameters that can be chosen by the operator. E.g. in security change mode, the computer cannot route outside the local subnet. Or if truely paranoid, has no network connection at all. No background processes run in SC mode.
This doesn't make the computer secure. Lots of bad thinks can be done by programs running in user space. But this gives you a kit of trusted tools that the OS can use to examine the running processes.
ancestor poster said that sunlight is limiting in rainforests. Perhaps under the top canopy, but I'm skeptical about it being limiting at the top of the rain forest.
In my bio class, many years ago, so the guestimates may have changed, the prof said that few plants increase photosythesis at all above insolation levels of about 25%. Of course in a Rain Forest it's frequently heavily cloudy. Perhaps on cloudy days, light is limiting. Hmm.
For most plants CO2 is limiting. Commercial greenhouses will run CO2 up to levels of 10000 ppm (1%)
Temperature is often limiting too. Lots of plants basically shut down at temps above 90 F, by closing their stomata to reduce water loss. Sure, in a rain forest there is lots of water, but you have to pump that water from the roots to the leaves. Water loss is highly non-linear with temp, so at some point you can't generate enough energy to replace the water.
Reducing the amount of sunlight is unlikely to reduce photosynthesis directly, unless it forms a lot of dense cloud.
***
One Discover channel doc pointed out that the amount of aerosols produced by China and India has masked the warming substantially. And with increasing prosperity, these countries are cleaning up their air. It is likely that we are going to see a rebound effect with much faster global warming for a year or so.
***
Ancenstral poster commented about melting ice caps reducing the temperature difference the drives the winds. Possibly, but not by much. The equator still gets a lot more radiation per square meter than the polar regions. They are also a lot larger. Melted north polar ice cap reveals a lot of water, which with low angles of incidence is still a pretty good reflector. However vapour pressure of water/air is large compared to ice/air. This may result in warmer, wetter sub-arctic regions. The circumpolar tundra's climate may become more like Sweden's with much heavier snowfall. (Much of the tundra is cold desert -- less than 10 inches annual precipitation)
Back to wind: The equator receives about 1.4 times as much radiation as does latitude 45. If mass transport didn't haul the heat away, it would have to radiate. To radiate 1.4 times as much heat it would have to increase in temp by a factor of the 4th root of 1.4, or about 1.09. Increasing the absolute temp by 9% works out to about 26 C temp rise.
Obviously this model is very flawed. I've ignored a bunch of things. Long before the temp rises that much you will get convection cells. They just run at a higher temp.
There are several factors that make pruning the bloat difficult for the average user:
1. Spurious dependencies.
I have on occasion installed a command line utility that called in the entire KDE source tree and did a build world.
2. Lack of documentation. Much of the time there is no good documentation outside of the program in terms of how to use it. You have to install it to test it, then uninstalling reveals that 9 billion things will break
3. Proliferation of libraries. I tell you Linux is getting as bad as windows with it's DLLs.
4. Incompatibility across upgrades. This is the flip side. Do an upgrade, and foo-3.2.11.lib changes the syntax for the bar call compared to foo-3.2.10.lib (of which both on install were symlinked to foo.lib...) Come on guys. Minor revisions should not change syntax.
5. Spurious change at the OS level. My computer is a tool. I use it to get work done. Changes for the sake of eye-candy is anathema. The visible part of a distribution should go unchanged for many years at a time.
6. Security fixes and bug fixes should be segregated from feature changes.
Steve Jobs, when running NeXT commented that people won't change for a 10% improvement. Or even 30%. People *will* change for a 500% improvement.
If you want to sell Linux to the masses:
1. It has to stop interfering with how things work.
2. It has to have good, clean, fast applications that are easy to learn to use, and incredibly stable.
There may be little problem with newer motherboards, but I can tell you there is a lot of combinations of stuff out there that won't work with a 3TB drive.
I just got burned on this. Had a not very old 500 GB Lacie external drive. The 500 GB drive in it was getting noisy, so I bought a Hitachi 3 TB drive, popped the case, and swapped drives. My Mac recognized it as a 801 GB drive. WTF!
A couple hours later I knew more about LBA. (Hitachi has good info on their web site.)
You need an OS that supports 48 bit LBA. You need drivers that support 48 bit LBA. You need adapter cards that support 48 bit LBA.
In my case the Lacie is a multi protocol box, so IT has firmware. And that firmware does not support 48 bit LBA, so bit 33 of the capacity is stripped off and I see 1 TB - 2 TiB -1 = 801 GB
I find it amusing.
3 TB for $150 bucks. 50 bucks a TB. My first hard drive added 1200 to the price of the computer. It was 10 MB and even that had to be logically divided into 8 chunks to be addressable by the 2 MHz Zilog Z-80.
I remember when CHS limitations restricted disks to 32 MB. There have been a series of limits since. Some of the limits were clearly stopgaps for a short respite, (We'll write bigger sectors and get a factor of 8) But several have been on the basis of "This should fix the problem for a few decades."
ZFS uses 64 bit addressing. Bets on how long before the first company ships a disk that won't address it.