Voice recognition is hard. Image comparison is hard. Working with video is hard.
Kodak has a digital camera back for view cameras that has 50 mega pixels 16 bits each per channel. That's 300 Mbytes per photoshop layer. Todays computers cannot render that fast enough. Most people don't use view cameras. Most of those don't use 20,000 buck digital backs. Those that do can put fractional terabytes of RAM in their computers.
People have gotten used to mediocrity in computers, and they don't expect much from them, and they are so used to their limitations that they are unaware of them. It will take some 'killer app' to move that forward.
I mentioned the phone market mostly to emphasize that reasonable network speed is NOT a given, and that even service of any kind is far from ubiquitous outside metro areas.
To a certain point you are right. The market is no longer one of, "replace my computer because it can't keep up" and more of one of "replace my computer because it's worn out." The number sold each year is declining to replacement units rather than new units.
Re: desktop. I want the desktop to BE the computer. 5 x 3 feet rolled into a quarter cylinder, touch sensitive, hyperbolic scaling. (If I move a window toward the edge of the screen it shrinks.) This is not a commodity item yet.
Would you replace your laptop if doing so it would weigh less than a pound, be as thin as a clipboard, be usable in daylight, was waterproof, and had enough battery to allow you to fly to Perth, Australia watching Blu-Ray movies all the way?
I disagree about cheap commodity computing being the death knell of hardware.
1. I still don't have real time voice recognition on my computer. I can't say, "Max: email Heather at Tree Time: Start message: Heather can you increase my order for aspen to 200 number 5 pots." End Message. Show. Send. Go to sleep.
2. One of the biggest problems with the 'paperless office' is the lack of desk size monitors: a monitor system that has the information capacity of a desk. 5' x 3' x 200 pixels/inch (at least on the centre blotter size area.
3. I still don't have an application that I can tell, "Find me the near duplicates of this image, despite crops, and edits in the contrast, brightness and hue.
4. I live in Canada. 3G communication may nominally be 6 mbit/s but my iPhone takes 30-60 seconds to load a page. I barely get cell phone coverage at home. Edge Network (2G?) covers about 75% of the paved roads outside the cities. I don't want my life to depend on cell phone coverage.
Home broadband? Currently my options are satellite and dialup. I pay $85/month for 1 Mbit download speed with a 50 MByte per hour cap. If I exceed the cap, I'm reduced to dialup speeds until there is essentially NO activity for an hour. I'm looking at a non-satellite option. It will require that I build a 60-75 foot tower to get line of sight to the provider's tower.
Tablets may be a partial answer, but there's portable and there's portable.
For me, when I'm in the field portable means it MUST fit in my pocket. I use two hands at a time most of the day. I can't have something that requires that I grab it and carry it. The iPhone is too small. I think I want something that just fits a shirt pocket that opens like a paperback.
I can't depend on networks, so it has to be self contained.
From a security/military point of view, the key is what they are capable of. Intentions are hard to determine, and can change. Capabilities may be hard to determine, but overestimated capabilities leaves you still safe.
However, the ability to travel in large groups between stars implies a level of energy control that is daunting. Any efforts to resist need to be based on making it just too much work.
If the aliens can use a blasted planet, then we're hooped. However if they want to have some form of 'economic' dealings with us, there is some space to work. (We have economic dealings with cows: We exchange steady food, disease prevention, good water for steak.)
The first level short of blasting could involve taking out most of the world's electronics with EMP pulses. At one point I read an estimate that a 100 MT bomb exploded 120 miles over Nebraska would take out most of the electronics in North America. This would essentially stop our economy, probably without killing more than 5% of the population directly.
I’m sorry. I don’t get really excited about microbes and brine shrimp. If you do NOTHING, they will become crunchy. Dike off a square mile or two as a preserve.
In general solutions that pay for themselves work faster. Bringing water in from either the Red Sea or the Mediterranean can generate power twice — once by the elevation difference, and once by the dilution process. Sure, you’ll end up with a salinity gradient — less salty where the resalinization plant comes in.
Done right, it can be a source of water and power in a region that needs both. (There's a trade off - more fresh water produced, less power)
The only way we as a species can have zero impact is to all commit suicide. I suspect that many people would object. So we have to choose at any given time between workable alternatives.
30 years of watching catastrophists predict the End of Life As We Know It has convinced me that eco-systems are much more robust than previously thought:
Most impacts from serious oil spills vanish in a couple years. A decade brings fish back in the Aral Sea. Lake Erie is coming back to life. Polar bears that lived on baby seals are now eating bird eggs,
The big disruptions are ones when we move a pest from one habitat to another, and it takes a while for a new balance to be struck. We still finding the balance from having earthworms in North America, honey bees in North America. Dutch elm disease. White Pine Blister Rust.
Much of the time natural selection for 5-10 generations works out a resistant type. E.g. We’re starting to see resistant elms, resistant chestnuts.
Critters with short life cycles — microbes and brine shrimp and their ilk, will adapt to changing salinity fairly quickly.
Study the Dead Sea options carefully, yes. But this looks like a viable soltuion to me.
1. Physical. 1a. Location. Could be a room number, a ceiling space in a hall. (leaf switches tend to live here in many buildings) If it's a server room you need rack and level information.
1b. What it is. Size, maker, asset tag number, Stuff you can't find from an ssh login.
2. Topological: 2a. Network connectivity. What wires to what. Some of this can be mapped automatically. Ping, followed by arp gives you the MAC/IP mapping. Most managed switches can give you a MAC/Port mapping. This is limited if done passively, as it won't discover redundent links, and it may be tricky to pry out the info on trunked lines. 2b. Subsystem connectivity. Which machines are connected to which disk arrays. Hopefully the array partitioning software can help you here.
3. Services 3a. Which servers run which services. At the lowest level, NMap is your friend. It can tell you at least what is running what service. When you get to a scale where you have multiple instances of services (E.g. you have 4 web servers) it starts getting trickier to do automatically. 3b. Failover boxes are extremely tricky this way. If you're confident in your setup, you can shut down a service on one machine and see if it gets picked up properly, but there are a lot of ways this can go wrong.
I solved #2 and part of 3 in one job with a bunch of perl scripts. It created stanzas much like AIX uses for a lot of config information.
Bunch of other stuff too. The master script for this could either pull data (run ping, then arp, run NMap) or push data (write dhcp files, NIS files, arp tables) Location and stuff like asset tags had to be added by hand. But it meant that my info was in one file. This worked fairly well in an academic department with 300 computers running 11 different versions of Unix.
Alberta is a province with laws similar to California. If you get into the intersection on yellow, you can proceed.
1. Left turns. Lots of times two cars will proceed part way into the intersection on green. When the light goes yellow, they wait another 2 seconds for the last car to clear, then they proceed. Usually both are still in the intersection on red.
2. Winter driving. Lots of times you can be 50 meters from an intersection, moving at 40 km/hr, have the light go yellow, and you cannot stop before the intersection or before the light goes red. Most people don't try very hard. The converse is true for the cross traffic. They spin and slide trying to get started.
The combination of the two effects is such that most lights now don't go green for several seconds after the otehr way has gone red. This allows time to clear the intersection.
I'm not opposed to all geoengineering. But it's certainly a lot better to work with things that are fairly easily reversed.
Example: There is a proposal to inject SO2 into the stratosphere. Mount Pinitubo did the original experiment for us.
Best thing: Residence time of SO2 in the upper atmosphere is only about 2 years. So if we goof, we stop, and things (hopefully) go back.
But wait, there's more.
Turns out that SO2 while reflective, also absorbs some energy, heating up the upper atmosphere. The decreases the lapse rate, making it harder for a cloud to rise enough to cool enough to precipitate. so while it makes it cooler, it also makes it dryer.
Ooops.
The problem with bubbles: This requires a major change int he chemistry of the ocean. A chemical that works in 1 part per million requires a HUGE amount of chemical to be effective when working on an oceanic scale.
Remember that the ocean averages 12,000 feet deep. a 30 foot layer of water over the planet has the same mass as the entire atmosphere. Sicne the ocean covers only 70% of the planet, we'll say that it takes 50 feet of ocean to equal one atmospheric mass. So the ocean has a mass of 240 times that of the atmosphere. It's taken us over a century to run the CO2 content of the atmosphere up by 100 ppm. That same mass is less than 0.5 ppm of the ocean. And someone is advocating deliberate polution ^H^H^H^H^H^H alteration of the ocean on the scale of ppms?
Bubbles are essentially soap.
Quick possible consequences: * slime washes off of fish. * sea birds lose waterproofing oils * shore bugs that depend on surface tension sink.
Geoengineering may have solutions in it, but THIS is not one of them.
All kids end up learning to count and do simple arithmetic just so they can handle the loose change that gravitates to their grubby paws.
Lot of merit in giving even young kids a solid foundation in arithmetic. Balance the checkbook, check the visa statement, figure out the materials list for the new deck. Do your income taxes. Estimate what 20% off a price really means. Calculate interest, figure out fuel economy, compare prices.
Young brains are better at certain things. Up to about 8 or so, young brains are language sponges. In kindergarten and early years kids should learn at least 2 other languages.
Up to age 12 kids are factoid memorizers. Rote stuff. They are good at learning it.
Around age 12 kids are ready to manipulate those facts and learn logic. They take great delight in showing up a contradiction in what you say.
Around age 15 kids start getting passionate about causes. (Yeah, other things too.) This is when you teach them to debate, persuade, sell. It's also when you teach them to recognize all the tricks used.
At the high school level, tying math into physics would be a win.
I had a colleague, a math major, who also taught physics at the high school I worked at.
I asked him why he didn't teach calculus and physics in the same course? He looked at me strangely.
"Look, velocity is the derivitive of position with respect to time. Acceleration is the derivative of velocity."
He had never thought of it that way.
Lots of high school physics is a lot easier with calculus.
In general lots of kids have real problems with abstract ideas. Even in college you run into students who learn better from getting the concrete exampels first, then the theory.
As to programming: While there is a bunch of abstract math that is likely unnecessary to most programmers, having a good sense of number is crucial. I've run into students who had no clue that four level deep nested for loops with 10 iterations each meant that the inside loop executed 10,000 times, or were completely unaware that their calculations were garbage because they overflowed large int.
There's something to be said for a course in formal logic too.
Programmers need math: They don't need the same math that physicists or engineers do.
I worked for YottaYotta for a bit over a year as a sysadmin. My boss wanted me to have a cell phone so I could be called on weekends to Fix Things That Broke.
I'd take a phone home. I'd watch the signal strength, and every place along the route where it was weak, I'd phone his office number and leave a message with my location. The next day I would give him a map with the spots I'd called from.
Two days later I'd get a new phone.
After five phones he admitted that this wasn't going to work. So I was allowed to keep my phone at work so they could find me in the building.
In general: If the boss wants me on a leash, he has to pay for it. I figure just being able to call me is worth at least minimum wage.
As to data on my cell phone. Sync to your computer. Get an app that wipes the phone on entry of the incorrect password.
While I can't see the FAA grounding planes that don't have GPS I can see them forbidding them access to certain airspace. A cub has no place playing with the big boys anyway.
A GPS enabled transponder doesn't have to be expensive, especially since this is an application that doesn't need 5 meter accuracy.
Given that a handheld GPS will run for 8 hours on a pair of double A batteries, they aren't exactly power pigs.
I would expect that any sensible plane design would be able to run most of the instrumentation off of battery for a reasonable period of time.
Running a radar is power intensive. Having a beacon that says 'here I am' every 5 - 15 seconds shouldn't be rocket science. For collision avoidance even a few watt signal would go miles and miles. And it doesn't have to do that every often. E.g. A signal receivable 10 miles away is 30 seconds closing time at jet speeds, roughly a minute between the cub and a jet at full speed, 3 minutes at terminal space speeds. Knowing the vector and location every 15 seconds would be lots.
More to the point:
What happens if a solar flare knocks out all the GPS satellites?
1. Drawing is hard on a computer. But it's hard with a pencil too. We took years to learn how to run a graphite stick in school.
2. For typing and for linear tasks in general keyboards work well. Adding a mouse, or learning all of vi/emacs move commands makes piecewise linear tasks (programming) faster. I would not want to try to use Adobe Illustrator with a keyboard interface. (Although I do use left hand on the keyboard, right hand on the wacom stylus. Left hand changes/mods tools.)
3. I once looked at paperless schools -- kids turn in their essays electronically, get marked by the teacher on screen, and returned for rewrites electronically. However I never found anything that allowed a teacher to work with that was as fast as a red pencil. The closest I ever saw was a NeXTStep application called, 'red pencil' that allowed you to use standard proof reader markup. Still wasn't as fast.
4. Screens aren't big enough. Even with dual monitors.
The screen of my dreams: It's a quarter cylinder laying across my desk on edge, with my eyes at the center of the arc. It runs from edge to edge of the desk. It has programmable hyperbolic geometry. As I move a window toward the edge, it shrinks. (but remains euclidian within the frame) This allows me to find stuff on my desktop.
5. The computer needs to become a better secretary. I need to talk to it. Right now to email a document I have to:
a. Select the gmail tab. b. Start an email to the recipient. c. Click attach file. d. Find the file in my file system. e. Say ok. f. Put a note in to give context g. Put a meaningfull subject line on it. h. Click send.
What we need to be able to do is say, "Thrall (or whatever you call your computer) send this file to Mike Wingate"
Yeah. I'm on 80 acres. I'm 6 miles from town. On that road to town there are 4 houses. I've got 4 more on the two miles the other side of me.
So 9 houses in 8 miles of road. I'm not holding my breath.
Recently the school I worked at had fibre trenched in. (The gov did it for all schools. Ironically the cable passes 1 mile from my house. I think the tech said that it cost $15 per meter to put it in. Road and driveway crossings extra because they stopped to tunnel under. Those I think were 10K each.
So 40 miles would bt a touch pricey.
I want to know why there aren't cheap laser links. Seems to me that an LED + rifle scope would work any day that wasn't foggy or *too* snowy.
I can buy dialup from phone company (They have the only modem pool that is not long distance) I can't do better than 38 Kbaud. $25/month.
Or I could tether my wife's iPhone and get the blistering speed of Edge network. (We aren't in range for 3G) Free with a 6GB cap per month. Not that we could ever get that much though it on Edge.
Or I could hook up to a satellite. $60 for 500K down, 125K up. In theory. In practice 1/3 to 2/3 of that.
Two different companies have wireless lan. One has a tower 3 miles away blocked by a 150 foot hill. The other has a tower 15 miles away blocked by a mile of forest. Of course if I put up a 150 foot mast, I could get either one.
***
Regarding Natural monopoly mentioned earlier.
Wi-Lan may be a partial answer. If you coupled low power wi-lan towers with extremely directional antennas you could probably bypass the wire system in residential areas. You would need a tower twice the height of the roof tops. Pringles can style antenna on the roof, it's mate on the tower. You would need to have enough channels so that antennas on the same radius wouldn't share a channel.
Let's see. Suppose a WiLan cell of 1 mile diameter. At 6 houses per acre, that's about 3000 house holds.
Assume 33% market. 1000 households. If the antenna beam can be kept to 3 degrees, and adjacent antenna don't use the same channel, then each channel handles 60 customers (360 degrees / 6 degrees/channel)
So for a thousand clients you'd need 16 channels.
On the mast, suppose that you have a 2' triangle cross section. Every foot you mount a ring on slight out riggers so any cables will end up between the ring and the the triangle. This gives a 3' diameter circle. Roughly 10 foot circumference. Leaving room for working, call it 8 antenna clamped on each ring.
1000 customers would mean 125 rings. At a foot apart that's a 160 foot tower.
Gotta figure a way to put more than one customer on a tower antenna -- An antenna has to be able to handle more than one channel.
Polarization allows you to double the channels. Time splitting allows a bunch of channels.
Finally you could do a divide and conquer. One person in each block gets free internet in exchange for having a 30' mast on his property, and supplying it with 100 W of power. His antenna rebroadcasts to the dozen people on his block.
As a small businessman I like cheques. I grow trees, and sell from the farm and from farmers markets. A point of sale terminal to accept plastic would cost me $40 per month, and on top of that I'd lose 2-3% of each sale to the banks.
With cheques, I take them in the next day, and deposit them.
Sure there's a risk of getting a bad cheque. So far that hasn't happened.
When I order trees from across the country I offer to pay by visa or cheque. The bigger outfits prefer VISA, the smaller ones cheques.
By looking at what the program does. * With partitioned application directories, any program that tries to write in another program's directory is suspect. * The number of apps on most computers that need network access is fairly small. Your AV program should know that Adobe installs a program that periodically check's adobe's site for updates. That's fine. But adobe's updater shouldn't be trying to send mail to controller.botnet.ru * The number of apps that need to run with other than user permissions is small. The AV program should know what these are.
Working with hydrogen under large pressures is tricky. Being a small molecule it leaks through lots of alloys, and can cause steel to get brittle.
That said: The concept is valid. What you want is an interuptable way to create methanol. One that you can turn on and off as power comes and goes. Methanol is a liquid at normal temp and pressure, and we have an infra-structure for moving liquids around.
Converting an IC engine to run on methanol should be fairly straight forward. It's also a suitable fuel for gas turbines.
MANY years ago, Scientific American had such a proposal. The expected costs at that point were less than purchasing a surface right of way for an interstate.
They were also looking at running them a substantial distance underground, so that gravity was used as an assist to accelerate and decelerate the train. My recollection was that the vacuum was a lot harder than 1-3 psi. I think they were talking about a few mmHg. Small enough that even running trains a 300 mph air resistance was minor.
The issue of failure modes to me is the sticky one. Get a train derailment inside a tunnel, and you have major problems. Just how do you clear the wreckage when you are 30 miles from the nearest end. Expecially if the wreckage is shorting out the power lines.
(Ok, ok. You cut the line near the wreck, pull out hte cars, Haul the bits out. Extend the power line a car. Repeat.
Imagine being on the train, and hearing the door gasket leaking as you go through the lock into the tunnel.
Be a cool way to move freight. Could be faster than truck, cheaper than air.
I would recommend imaging twice. For the first one, Get a print of graph paper with a 1" grid done on mylar film. Put the film over the image.
This gives you a control grid that you can use to correct for distortion of your imaging system. In particular if you end up imaging in chunks, each chunk will have distortion to correct for.
In creating your transformation, you have two things to consider: One is the projection used. Others have mentioned this. While knowing it will make it simpler, it should be possible with 9-16 well spaced control points to determine it.
The other factor is errors made by the surveyors. This one is more subtle, as there will be a 'circle of influence' around landmarks. A landmark that is highly visible, and well located (church spire on hill) will have a strong influence. A pond will have a much smaller influence just because the surveyors couldn't see it.
Surveyor errors may be random or systematic. E.g. If they are using a magnetic compass a declination error will have the effect of adding a shear transform to the image. A time piece that gains or loses time can stretch or contract the path. If it was done by a team, some of the team may be more careful than others. You may find one class of detail accurate, and another class less accurate.
A third source of error is the modern map. Unless *really* new, many maps are not orthorectified properly. The compensation for the position of a landmark due to the oblique angle of the aerial photograph depends on the elevation. And that in turn depends on both the quality of the stereoscope, the skill of the operator, and the attention to detail in the whole imaging process.
Recently I used a GPS and a 30 year old Canadian Topo map to create a cross country orienteering game. In our region the river has cut a broad valley about 150 to 200 feet below the plain. The location of the rim, as indicated by the contour lines was consistently about 50-80 meters off. East west coordinates were up to 150 meters off.
I use km/l all the time. I live in Canada. The odometer is in km. I pump l into my tank. AND it's a simple numeric conversion (2.75 roughly) to multiply mpg to get km/l
l/100k has it's merits too. I just haven't made that conversion in my mental processor yet. And converting back and forth in my head always requires division.
It's curious how I use different measuring systems in different domains. People's height: feet and inches. Weight: pounds. Fertilizer: kilograms. Construction: 2x4's and 4x8's. Plumbing: both. Weather: Centigrade and mm. Cooking Farenheit.
After my mom's first heart surgery, she said never again. She lived 3 more years. The last 6 months were in an extended care facility doing a balancing act between her failing lungs (emphysema, from 60 years of smoking) failing heart (overweight, diabetic) failing kidneys (diabetes, diaretics to keep her lungs from filling up.)
I said my good-byes and left her with my sister.
****
When I was 14 my father had open heart surgery. they replaced 3 valves that had been damaged by rheumatic fever when he was a young man. The hours on the heart lung machine, coupled with vastly improved heart action after, caused a series of micro strokes. The strong vibrant man who was my father was replaced with this hollow shell that could no longer help me with my math, no longer could follow the plot of a TV show. Worse, he knew what he had been and no longer was.
****
My grandmother Zelia was in a car accident at age 75. She recovered consciousness in the hospital, her kids had time to see her. Most of the grand kids came by, and some of the great grand kids. She died that night. Good way to go.
****
My other grandmother broke her hip in the old folks home she was in, and never got out of bed again. It took 6 months for her to finish dying.
****
The father of a friend was a heavy smoker. Early in his 60's he was diagnosed with lung cancer. He refused the lung transplant offered, opting for paliative care. He lived at home with his wife. The kids and grandkids came by frequently, and he would play with them, less and less actively as lung function deteriorated. I think only the last 2-3 weeks were in a hospice. He had courage.
My wife and I have living wills. Neither of us are to keep the other alive once there is little chance of recovery.
My sister has deliberately chosen to live on an island. It's half an hour to the ferry terminal, an hour plus ferry ride, and another hour drive to the nearest hospital.
***
My best answers:
* Once I go to bed for the last time, try to keep me alive until I've said good-bye to my people. After that, just paliative care.
Our culture has become morbidly fearful about death. As a practitioner and teacher of backcountry travel I have come close to death a few times. Events such that I was certain I was not going to live more than another minute or two.
I discovered that death is not to be feared. I had regrets of things not done, did not like the pain I was in, but death itself was just there.
Do not project your own fears onto other people. Have courage. Face your death when it's time.
2. At present salaries are about 85% of a schools operating costs. In many places schools are funded by property taxes. Tell people that their property taxes are going to double, and you will have riots.
3. In Alberta a teacher starts at about 35K/year. After 10 years they are at about 55K. If they pick up a masters degree they get about 70K. This is in a province where the average FAMILY income is about 50K. For comparison when I worked for the U of A as a unix sysadmin, I got $48K with 15 years experience. Teaching is a middle of the road professional job here.
A big part of the problem is that teachers aren't taught well. I worked in a private school without a teaching certificate for 10 years. Later I went back to school and got a B.Ed. Two years of sand piling.
My headmaster would NOT hire someone who only had a teaching degree. Claimed that a border collie could do a better job, and the kids would like her better. He preferred young people with a 'hard' degree -- one that required either brains or scholarship -- figuring he could teach them how to teach in a year.
Rather than create a darwinian survival of the fittest, I think it would better to create a positive feed back loop.
Try this: Most school districts have some form of peer selected teaching execellence awards. Take the N runner ups. Secund them away from the school, arm them with clipboards, and let them loose in the Education college at the local university.
Let them rewrite the entire curriculum. Chuck the sandpiling courses.
Move the education system away from the universities and into the Community colleges, and the Institutes of technology. Run it more like the trade program rather than college.
Here's how the teacher training system would work.
1. You pick up a 3 or 4 year degree in some subject. For high school teachers it's related to their specialty. For elementary teachers it's some combination of childhood psychology, and general level courses in everything.
2. You go to 'Normal School' (What teacher's schools used to be called.) This is your typical trade school program. 12 weeks of training, followed by a minimum of 24 weeks of work experience in the field. You get paid for work, just not as much as a journeyman does.
3. You get out, and join the fray.
4. After N years experience you can apply to become a teacher at the Normal school. N should be short initially 5 years or so, to maximize feedback.
***
The second aspect that needs to change is the expectations of parents and kids. Teachers are expected to put up with behaviours from both that are off the wall:
* My sister-in-law had a grade 2 student that was acting out. She kept her after class to talk to her at the start of lunch hour. Possibly kept her for 5 minutes. The girls mom stormed in, said that Libby had no right to keep her kid late at all without written notice. Grabbed her kid, as didn't bring her back that day. Mom complained to the principal, and Libby was required to apologise to the mom.
* I worked for a while in a school where almost none of the kids wanted to be there. Most kids, when push comes to shove recognize that they need an education. They may not like it, but they tolerate it. Add to the social group, and school isn't as boring as just being at home. In this school I had one bright young blade come up and inform me that he wasn't coming back. "Why not?" "I want to go to a school where there are otehr kids in my class who want to learn"
* Merging the seriously learning disabled into regular classes is a mistake. A teacher teaching 24 students has about 2 minutes per kid in the course of a class. Typically 1/3 of the kids get it from being shown once. They learn from your talk at the beginning. Another third will learn from a few seconds of 1 on 1 showing an example or two. The final third need several examples and coaching. This gets really difficult if you have a kid who is noisy or otherwise disruptive. It gets difficult if they have such special needs that the regular curriculum doesn't work. If they are quiet, try hard, and don't need a separate curriculum, then merge them in.
Slashdot Mirror
Voice recognition is hard. Image comparison is hard. Working with video is hard.
Kodak has a digital camera back for view cameras that has 50 mega pixels 16 bits each per channel. That's 300 Mbytes per photoshop layer. Todays computers cannot render that fast enough. Most people don't use view cameras. Most of those don't use 20,000 buck digital backs. Those that do can put fractional terabytes of RAM in their computers.
People have gotten used to mediocrity in computers, and they don't expect much from them, and they are so used to their limitations that they are unaware of them. It will take some 'killer app' to move that forward.
I mentioned the phone market mostly to emphasize that reasonable network speed is NOT a given, and that even service of any kind is far from ubiquitous outside metro areas.
To a certain point you are right. The market is no longer one of, "replace my computer because it can't keep up" and more of one of "replace my computer because it's worn out." The number sold each year is declining to replacement units rather than new units.
Re: desktop. I want the desktop to BE the computer. 5 x 3 feet rolled into a quarter cylinder, touch sensitive, hyperbolic scaling. (If I move a window toward the edge of the screen it shrinks.) This is not a commodity item yet.
Would you replace your laptop if doing so it would weigh less than a pound, be as thin as a clipboard, be usable in daylight, was waterproof, and had enough battery to allow you to fly to Perth, Australia watching Blu-Ray movies all the way?
I disagree about cheap commodity computing being the death knell of hardware.
1. I still don't have real time voice recognition on my computer. I can't say, "Max: email Heather at Tree Time: Start message: Heather can you increase my order for aspen to 200 number 5 pots." End Message. Show. Send. Go to sleep.
2. One of the biggest problems with the 'paperless office' is the lack of desk size monitors: a monitor system that has the information capacity of a desk. 5' x 3' x 200 pixels/inch (at least on the centre blotter size area.
3. I still don't have an application that I can tell, "Find me the near duplicates of this image, despite crops, and edits in the contrast, brightness and hue.
4. I live in Canada. 3G communication may nominally be 6 mbit/s but my iPhone takes 30-60 seconds to load a page. I barely get cell phone coverage at home. Edge Network (2G?)
covers about 75% of the paved roads outside the cities. I don't want my life to depend on cell phone coverage.
Home broadband? Currently my options are satellite and dialup. I pay $85/month for 1 Mbit download speed with a 50 MByte per hour cap. If I exceed the cap, I'm reduced to dialup speeds until there is essentially NO activity for an hour. I'm looking at a non-satellite option. It will require that I build a 60-75 foot tower to get line of sight to the provider's tower.
Tablets may be a partial answer, but there's portable and there's portable.
For me, when I'm in the field portable means it MUST fit in my pocket. I use two hands at a time most of the day. I can't have something that requires that I grab it and carry it. The iPhone is too small. I think I want something that just fits a shirt pocket that opens like a paperback.
I can't depend on networks, so it has to be self contained.
From a security/military point of view, the key is what they are capable of. Intentions are hard to determine, and can change. Capabilities may be hard to determine, but overestimated capabilities leaves you still safe.
However, the ability to travel in large groups between stars implies a level of energy control that is daunting. Any efforts to resist need to be based on making it just too much work.
If the aliens can use a blasted planet, then we're hooped. However if they want to have some form of 'economic' dealings with us, there is some space to work. (We have economic dealings with cows: We exchange steady food, disease prevention, good water for steak.)
The first level short of blasting could involve taking out most of the world's electronics with EMP pulses. At one point I read an estimate that a 100 MT bomb exploded 120 miles over Nebraska would take out most of the electronics in North America. This would essentially stop our economy, probably without killing more than 5% of the population directly.
I’m sorry. I don’t get really excited about microbes and brine shrimp. If you do NOTHING, they will become crunchy.
Dike off a square mile or two as a preserve.
In general solutions that pay for themselves work faster. Bringing water in from either the Red Sea or the Mediterranean can generate power twice — once by the elevation difference, and once by the dilution process. Sure, you’ll end up with a salinity gradient — less salty where the resalinization plant comes in.
Done right, it can be a source of water and power in a region that needs both. (There's a trade off - more fresh water produced, less power)
The only way we as a species can have zero impact is to all commit suicide. I suspect that many people would object. So we have to choose at any given time between workable alternatives.
30 years of watching catastrophists predict the End of Life As We Know It has convinced me that eco-systems are much more robust than previously thought:
Most impacts from serious oil spills vanish in a couple years. A decade brings fish back in the Aral Sea. Lake Erie is coming back to life. Polar bears that lived on baby seals are now eating bird eggs,
The big disruptions are ones when we move a pest from one habitat to another, and it takes a while for a new balance to be struck. We still finding the balance from having earthworms in North America, honey bees in North America. Dutch elm disease. White Pine Blister Rust.
Much of the time natural selection for 5-10 generations works out a resistant type. E.g. We’re starting to see resistant elms, resistant chestnuts.
Critters with short life cycles — microbes and brine shrimp and their ilk, will adapt to changing salinity fairly quickly.
Study the Dead Sea options carefully, yes. But this looks like a viable soltuion to me.
You've got three problems:
1. Physical.
1a. Location. Could be a room number, a ceiling space in a hall. (leaf switches tend to live here in many buildings) If it's a server room you need rack and level information.
1b. What it is. Size, maker, asset tag number, Stuff you can't find from an ssh login.
2. Topological:
2a. Network connectivity. What wires to what. Some of this can be mapped automatically. Ping, followed by arp gives you the MAC/IP mapping. Most managed switches can give you a MAC/Port mapping. This is limited if done passively, as it won't discover redundent links, and it may be tricky to pry out the info on trunked lines.
2b. Subsystem connectivity. Which machines are connected to which disk arrays. Hopefully the array partitioning software can help you here.
3. Services
3a. Which servers run which services. At the lowest level, NMap is your friend. It can tell you at least what is running what service. When you get to a scale where you have multiple instances of services (E.g. you have 4 web servers) it starts getting trickier to do automatically.
3b. Failover boxes are extremely tricky this way. If you're confident in your setup, you can shut down a service on one machine and see if it gets picked up properly, but there are a lot of ways this can go wrong.
I solved #2 and part of 3 in one job with a bunch of perl scripts. It created stanzas much like AIX uses for a lot of config information.
192.168.1.1
FQDN: gateway.foo.edu
CNAMES: kerberos
MAC: 00:01:02:03:04:05
OS: OpenBSD 3.8
SWITCH: R1:p17 (port 17 on riser switch 1)
ONLINE: 17-Sep-2003 10:36:10
Ports: ssh
Location:
Owner:
192.168.1.2 ...
Bunch of other stuff too. The master script for this could either pull data (run ping, then arp, run NMap) or push data (write dhcp files, NIS files, arp tables) Location and stuff like asset tags had to be added by hand. But it meant that my info was in one file. This worked fairly well in an academic department with 300 computers running 11 different versions of Unix.
Alberta is a province with laws similar to California. If you get into the intersection on yellow, you can proceed.
1. Left turns. Lots of times two cars will proceed part way into the intersection on green. When the light goes yellow, they wait another 2 seconds for the last car to clear, then they proceed. Usually both are still in the intersection on red.
2. Winter driving. Lots of times you can be 50 meters from an intersection, moving at 40 km/hr, have the light go yellow, and you cannot stop before the intersection or before the light goes red. Most people don't try very hard. The converse is true for the cross traffic. They spin and slide trying to get started.
The combination of the two effects is such that most lights now don't go green for several seconds after the otehr way has gone red. This allows time to clear the intersection.
I'm not opposed to all geoengineering. But it's certainly a lot better to work with things that are fairly easily reversed.
Example: There is a proposal to inject SO2 into the stratosphere. Mount Pinitubo did the original experiment for us.
Best thing: Residence time of SO2 in the upper atmosphere is only about 2 years. So if we goof, we stop, and things (hopefully) go back.
But wait, there's more.
Turns out that SO2 while reflective, also absorbs some energy, heating up the upper atmosphere. The decreases the lapse rate, making it harder for a cloud to rise enough to cool enough to precipitate. so while it makes it cooler, it also makes it dryer.
Ooops.
The problem with bubbles: This requires a major change int he chemistry of the ocean. A chemical that works in 1 part per million requires a HUGE amount of chemical to be effective when working on an oceanic scale.
Remember that the ocean averages 12,000 feet deep. a 30 foot layer of water over the planet has the same mass as the entire atmosphere. Sicne the ocean covers only 70% of the planet, we'll say that it takes 50 feet of ocean to equal one atmospheric mass. So the ocean has a mass of 240 times that of the atmosphere. It's taken us over a century to run the CO2 content of the atmosphere up by 100 ppm. That same mass is less than 0.5 ppm of the ocean. And someone is advocating deliberate polution ^H^H^H^H^H^H alteration of the ocean on the scale of ppms?
Bubbles are essentially soap.
Quick possible consequences:
* slime washes off of fish.
* sea birds lose waterproofing oils
* shore bugs that depend on surface tension sink.
Geoengineering may have solutions in it, but THIS is not one of them.
Make a distinction between MATH and ARITHMETIC.
All kids end up learning to count and do simple arithmetic just so they can handle the loose change that gravitates to their grubby paws.
Lot of merit in giving even young kids a solid foundation in arithmetic. Balance the checkbook, check the visa statement, figure out the materials list for the new deck. Do your income taxes. Estimate what 20% off a price really means. Calculate interest, figure out fuel economy, compare prices.
Young brains are better at certain things. Up to about 8 or so, young brains are language sponges. In kindergarten and early years kids should learn at least 2 other languages.
Up to age 12 kids are factoid memorizers. Rote stuff. They are good at learning it.
Around age 12 kids are ready to manipulate those facts and learn logic. They take great delight in showing up a contradiction in what you say.
Around age 15 kids start getting passionate about causes. (Yeah, other things too.) This is when you teach them to debate, persuade, sell. It's also when you teach them to recognize all the tricks used.
At the high school level, tying math into physics would be a win.
I had a colleague, a math major, who also taught physics at the high school I worked at.
I asked him why he didn't teach calculus and physics in the same course? He looked at me strangely.
"Look, velocity is the derivitive of position with respect to time. Acceleration is the derivative of velocity."
He had never thought of it that way.
Lots of high school physics is a lot easier with calculus.
In general lots of kids have real problems with abstract ideas. Even in college you run into students who learn better from getting the concrete exampels first, then the theory.
As to programming: While there is a bunch of abstract math that is likely unnecessary to most programmers, having a good sense of number is crucial. I've run into students who had no clue that four level deep nested for loops with 10 iterations each meant that the inside loop executed 10,000 times, or were completely unaware that their calculations were garbage because they overflowed large int.
There's something to be said for a course in formal logic too.
Programmers need math:
They don't need the same math that physicists or engineers do.
I worked for YottaYotta for a bit over a year as a sysadmin. My boss wanted me to have a cell phone so I could be called on weekends to Fix Things That Broke.
I'd take a phone home. I'd watch the signal strength, and every place along the route where it was weak, I'd phone his office number and leave a message with my location. The next day I would give him a map with the spots I'd called from.
Two days later I'd get a new phone.
After five phones he admitted that this wasn't going to work. So I was allowed to keep my phone at work so they could find me in the building.
In general: If the boss wants me on a leash, he has to pay for it. I figure just being able to call me is worth at least minimum wage.
As to data on my cell phone. Sync to your computer. Get an app that wipes the phone on entry of the incorrect password.
While I can't see the FAA grounding planes that don't have GPS I can see them forbidding them access to certain airspace. A cub has no place playing with the big boys anyway.
A GPS enabled transponder doesn't have to be expensive, especially since this is an application that doesn't need 5 meter accuracy.
Given that a handheld GPS will run for 8 hours on a pair of double A batteries, they aren't exactly power pigs.
I would expect that any sensible plane design would be able to run most of the instrumentation off of battery for a reasonable period of time.
Running a radar is power intensive. Having a beacon that says 'here I am' every 5 - 15 seconds shouldn't be rocket science. For collision avoidance even a few watt signal would go miles and miles. And it doesn't have to do that every often. E.g. A signal receivable 10 miles away is 30 seconds closing time at jet speeds, roughly a minute between the cub and a jet at full speed, 3 minutes at terminal space speeds. Knowing the vector and location every 15 seconds would be lots.
More to the point:
What happens if a solar flare knocks out all the GPS satellites?
If you make a large format camera smaller, is it still a large format camera? Hmm.
Right now you can get digital backs for large format camerers The Phase 1 P45 is a 39 Megapixel 16 bit, 6x9 frame.
1. Drawing is hard on a computer. But it's hard with a pencil too. We took years to learn how to run a graphite stick in school.
2. For typing and for linear tasks in general keyboards work well. Adding a mouse, or learning all of vi/emacs move commands makes piecewise linear tasks (programming) faster. I would not want to try to use Adobe Illustrator with a keyboard interface. (Although I do use left hand on the keyboard, right hand on the wacom stylus. Left hand changes/mods tools.)
3. I once looked at paperless schools -- kids turn in their essays electronically, get marked by the teacher on screen, and returned for rewrites electronically. However I never found anything that allowed a teacher to work with that was as fast as a red pencil. The closest I ever saw was a NeXTStep application called, 'red pencil' that allowed you to use standard proof reader markup. Still wasn't as fast.
4. Screens aren't big enough. Even with dual monitors.
The screen of my dreams: It's a quarter cylinder laying across my desk on edge, with my eyes at the center of the arc. It runs from edge to edge of the desk. It has programmable hyperbolic geometry. As I move a window toward the edge, it shrinks. (but remains euclidian within the frame) This allows me to find stuff on my desktop.
5. The computer needs to become a better secretary. I need to talk to it. Right now to email a document I have to:
a. Select the gmail tab.
b. Start an email to the recipient.
c. Click attach file.
d. Find the file in my file system.
e. Say ok.
f. Put a note in to give context
g. Put a meaningfull subject line on it.
h. Click send.
What we need to be able to do is say, "Thrall (or whatever you call your computer) send this file to Mike Wingate"
Yeah. I'm on 80 acres. I'm 6 miles from town.
On that road to town there are 4 houses. I've got 4 more on the two miles the other side of me.
So 9 houses in 8 miles of road. I'm not holding my breath.
Recently the school I worked at had fibre trenched in. (The gov did it for all schools. Ironically the cable passes 1 mile from my house. I think the tech said that it cost $15 per meter to put it in. Road and driveway crossings extra because they stopped to tunnel under. Those I think were 10K each.
So 40 miles would bt a touch pricey.
I want to know why there aren't cheap laser links. Seems to me that an LED + rifle scope would work any day that wasn't foggy or *too* snowy.
Life under the Antarctic shelf has external inputs: Water circulates there from the open ocean, bringing with it all sorts of gourmet delicacies.
Life on Europa doesn't have an adjacent sunlit ocean providing energy inputs.
The colonies of critters living around hydrothermal vents is a better model of potential Europa life than life under the edge of the shelf.
I've got competition.
Really!
I can buy dialup from phone company (They have the only modem pool that is not long distance) I can't do better than 38 Kbaud. $25/month.
Or I could tether my wife's iPhone and get the blistering speed of Edge network. (We aren't in range for 3G) Free with a 6GB cap per month. Not that we could ever get that much though it on Edge.
Or I could hook up to a satellite. $60 for 500K down, 125K up. In theory. In practice 1/3 to 2/3 of that.
Two different companies have wireless lan. One has a tower 3 miles away blocked by a 150 foot hill. The other has a tower 15 miles away blocked by a mile of forest. Of course if I put up a 150 foot mast, I could get either one.
***
Regarding Natural monopoly mentioned earlier.
Wi-Lan may be a partial answer. If you coupled low power wi-lan towers with extremely directional antennas you could probably bypass the wire system in residential areas. You would need a tower twice the height of the roof tops. Pringles can style antenna on the roof, it's mate on the tower. You would need to have enough channels so that antennas on the same radius wouldn't share a channel.
Let's see. Suppose a WiLan cell of 1 mile diameter.
At 6 houses per acre, that's about 3000 house holds.
Assume 33% market. 1000 households. If the antenna beam can be kept to 3 degrees, and adjacent antenna don't use the same channel, then each channel handles 60 customers (360 degrees / 6 degrees/channel)
So for a thousand clients you'd need 16 channels.
On the mast, suppose that you have a 2' triangle cross section. Every foot you mount a ring on slight out riggers so any cables will end up between the ring and the the triangle. This gives a 3' diameter circle. Roughly 10 foot circumference. Leaving room for working, call it 8 antenna clamped on each ring.
1000 customers would mean 125 rings. At a foot apart that's a 160 foot tower.
Gotta figure a way to put more than one customer on a tower antenna -- An antenna has to be able to handle more than one channel.
Polarization allows you to double the channels. Time splitting allows a bunch of channels.
Finally you could do a divide and conquer. One person in each block gets free internet in exchange for having a 30' mast on his property, and supplying it with 100 W of power. His antenna rebroadcasts to the dozen people on his block.
As a small businessman I like cheques. I grow trees, and sell from the farm and from farmers markets. A point of sale terminal to accept plastic would cost me $40 per month, and on top of that I'd lose 2-3% of each sale to the banks.
With cheques, I take them in the next day, and deposit them.
Sure there's a risk of getting a bad cheque. So far that hasn't happened.
When I order trees from across the country I offer to pay by visa or cheque. The bigger outfits prefer VISA, the smaller ones cheques.
How to identify a new virus?
By looking at what the program does.
* With partitioned application directories, any program that tries to write in another program's directory is suspect.
* The number of apps on most computers that need network access is fairly small. Your AV program should know that Adobe installs a program that periodically check's adobe's site for updates. That's fine. But adobe's updater shouldn't be trying to send mail to controller.botnet.ru
* The number of apps that need to run with other than user permissions is small. The AV program should know what these are.
Working with hydrogen under large pressures is tricky. Being a small molecule it leaks through lots of alloys, and can cause steel to get brittle.
That said: The concept is valid. What you want is an interuptable way to create methanol. One that you can turn on and off as power comes and goes. Methanol is a liquid at normal temp and pressure, and we have an infra-structure for moving liquids around.
Converting an IC engine to run on methanol should be fairly straight forward. It's also a suitable fuel for gas turbines.
MANY years ago, Scientific American had such a proposal. The expected costs at that point were less than purchasing a surface right of way for an interstate.
They were also looking at running them a substantial distance underground, so that gravity was used as an assist to accelerate and decelerate the train. My recollection was that the vacuum was a lot harder than 1-3 psi. I think they were talking about a few mmHg. Small enough that even running trains a 300 mph air resistance was minor.
The issue of failure modes to me is the sticky one.
Get a train derailment inside a tunnel, and you have major problems. Just how do you clear the wreckage when you are 30 miles from the nearest end. Expecially if the wreckage is shorting out the power lines.
(Ok, ok. You cut the line near the wreck, pull out hte cars,
Haul the bits out. Extend the power line a car. Repeat.
Imagine being on the train, and hearing the door gasket leaking as you go through the lock into the tunnel.
Be a cool way to move freight. Could be faster than truck, cheaper than air.
Lots of good stuff here.
I would recommend imaging twice. For the first one, Get a print of graph paper with a 1" grid done on mylar film. Put the film over the image.
This gives you a control grid that you can use to correct for distortion of your imaging system. In particular if you end up imaging in chunks, each chunk will have distortion to correct for.
In creating your transformation, you have two things to consider: One is the projection used. Others have mentioned this. While knowing it will make it simpler, it should be possible with 9-16 well spaced control points to determine it.
The other factor is errors made by the surveyors. This one is more subtle, as there will be a 'circle of influence' around landmarks. A landmark that is highly visible, and well located (church spire on hill) will have a strong influence. A pond will have a much smaller influence just because the surveyors couldn't see it.
Surveyor errors may be random or systematic. E.g. If they are using a magnetic compass a declination error will have the effect of adding a shear transform to the image. A time piece that gains or loses time can stretch or contract the path. If it was done by a team, some of the team may be more careful than others. You may find one class of detail accurate, and another class less accurate.
A third source of error is the modern map. Unless *really* new, many maps are not orthorectified properly. The compensation for the position of a landmark due to the oblique angle of the aerial photograph depends on the elevation. And that in turn depends on both the quality of the stereoscope, the skill of the operator, and the attention to detail in the whole imaging process.
Recently I used a GPS and a 30 year old Canadian Topo map to create a cross country orienteering game. In our region the river has cut a broad valley about 150 to 200 feet below the plain. The location of the rim, as indicated by the contour lines was consistently about 50-80 meters off. East west coordinates were up to 150 meters off.
Good luck on your project.
I use km/l all the time. I live in Canada. The odometer is in km. I pump l into my tank. AND it's a simple numeric conversion (2.75 roughly) to multiply mpg to get km/l
l/100k has it's merits too. I just haven't made that conversion in my mental processor yet. And converting back and forth in my head always requires division.
It's curious how I use different measuring systems in different domains. People's height: feet and inches. Weight: pounds. Fertilizer: kilograms. Construction: 2x4's and 4x8's. Plumbing: both. Weather: Centigrade and mm. Cooking Farenheit.
I disagree.
After my mom's first heart surgery, she said never again.
She lived 3 more years. The last 6 months were in an extended care facility doing a balancing act between her failing lungs (emphysema, from 60 years of smoking) failing heart (overweight, diabetic) failing kidneys (diabetes, diaretics to keep her lungs from filling up.)
I said my good-byes and left her with my sister.
****
When I was 14 my father had open heart surgery. they replaced 3 valves that had been damaged by rheumatic fever when he was a young man. The hours on the heart lung machine, coupled with vastly improved heart action after, caused a series of micro strokes. The strong vibrant man who was my father was replaced with this hollow shell that could no longer help me with my math, no longer could follow the plot of a TV show. Worse, he knew what he had been and no longer was.
****
My grandmother Zelia was in a car accident at age 75. She recovered consciousness in the hospital, her kids had time to see her. Most of the grand kids came by, and some of the great grand kids. She died that night. Good way to go.
****
My other grandmother broke her hip in the old folks home she was in, and never got out of bed again. It took 6 months for her to finish dying.
****
The father of a friend was a heavy smoker. Early in his 60's he was diagnosed with lung cancer. He refused the lung transplant offered, opting for paliative care. He lived at home with his wife. The kids and grandkids came by frequently, and he would play with them, less and less actively as lung function deteriorated. I think only the last 2-3 weeks were in a hospice. He had courage.
My wife and I have living wills. Neither of us are to keep the other alive once there is little chance of recovery.
My sister has deliberately chosen to live on an island. It's half an hour to the ferry terminal, an hour plus ferry ride, and another hour drive to the nearest hospital.
***
My best answers:
* Once I go to bed for the last time, try to keep me alive until I've said good-bye to my people. After that, just paliative care.
Our culture has become morbidly fearful about death. As a practitioner and teacher of backcountry travel I have come close to death a few times. Events such that I was certain I was not going to live more than another minute or two.
I discovered that death is not to be feared. I had regrets of things not done, did not like the pain I was in, but death itself was just there.
Do not project your own fears onto other people. Have courage. Face your death when it's time.
1. Because some people like working with kids.
2. At present salaries are about 85% of a schools operating costs. In many places schools are funded by property taxes. Tell people that their property taxes are going to double, and you will have riots.
3. In Alberta a teacher starts at about 35K/year. After 10 years they are at about 55K. If they pick up a masters degree they get about 70K. This is in a province where the average FAMILY income is about 50K. For comparison when I worked for the U of A as a unix sysadmin, I got $48K with 15 years experience. Teaching is a middle of the road professional job here.
A big part of the problem is that teachers aren't taught well. I worked in a private school without a teaching certificate for 10 years. Later I went back to school and got a B.Ed. Two years of sand piling.
My headmaster would NOT hire someone who only had a teaching degree. Claimed that a border collie could do a better job, and the kids would like her better. He preferred young people with a 'hard' degree -- one that required either brains or scholarship -- figuring he could teach them how to teach in a year.
Rather than create a darwinian survival of the fittest, I think it would better to create a positive feed back loop.
Try this: Most school districts have some form of peer selected teaching execellence awards. Take the N runner ups. Secund them away from the school, arm them with clipboards, and let them loose in the Education college at the local university.
Let them rewrite the entire curriculum. Chuck the sandpiling courses.
Move the education system away from the universities and into the Community colleges, and the Institutes of technology. Run it more like the trade program rather than college.
Here's how the teacher training system would work.
1. You pick up a 3 or 4 year degree in some subject. For high school teachers it's related to their specialty. For elementary teachers it's some combination of childhood psychology, and general level courses in everything.
2. You go to 'Normal School' (What teacher's schools used to be called.) This is your typical trade school program. 12 weeks of training, followed by a minimum of 24 weeks of work experience in the field. You get paid for work, just not as much as a journeyman does.
3. You get out, and join the fray.
4. After N years experience you can apply to become a teacher at the Normal school. N should be short initially 5 years or so, to maximize feedback.
***
The second aspect that needs to change is the expectations of parents and kids. Teachers are expected to put up with behaviours from both that are off the wall:
* My sister-in-law had a grade 2 student that was acting out. She kept her after class to talk to her at the start of lunch hour. Possibly kept her for 5 minutes. The girls mom stormed in, said that Libby had no right to keep her kid late at all without written notice. Grabbed her kid, as didn't bring her back that day. Mom complained to the principal, and Libby was required to apologise to the mom.
* I worked for a while in a school where almost none of the kids wanted to be there. Most kids, when push comes to shove recognize that they need an education. They may not like it, but they tolerate it. Add to the social group, and school isn't as boring as just being at home. In this school I had one bright young blade come up and inform me that he wasn't coming back. "Why not?" "I want to go to a school where there are otehr kids in my class who want to learn"
* Merging the seriously learning disabled into regular classes is a mistake. A teacher teaching 24 students has about 2 minutes per kid in the course of a class. Typically 1/3 of the kids get it from being shown once. They learn from your talk at the beginning. Another third will learn from a few seconds of 1 on 1 showing an example or two. The final third need several examples and coaching. This gets really difficult if you have a kid who is noisy or otherwise disruptive. It gets difficult if they have such special needs that the regular curriculum doesn't work. If they are quiet, try hard, and don't need a separate curriculum, then merge them in.