I had the same issue come up at my current contract. They wanted a fixed cost, knowing that the number of hours worked would be difficult to track (sysadmin support). Since the whole contract is pretty small, we agree to a flat rate for a limited contract time, and I insisted on a "nominal 40 hr work week" clause in the contract. This is still probably too vague.. but it was a decent compromise, and since the contract length is only 8 weeks I did not worry about abuse.
You have nothing to worry about from the IRS (as long as you pay them), but the employer does. The IRS has been taking a pretty tough stance on defining employees. Pretty much any individual paid by a company is an employee, and hence the employer is responsible for withholdings. The company must prove that you are not an employee (they must have an "arms length" relationship, which is up to the IRS to decide... but if you work at their location, are paid a flat rate, and are solely working for them, the IRS is likely to call you an employee). A contract is not enough. Which means if you don't pay your self employement withholdings, the company gets fined..
Also if you are a contractor, you should be paid about 1.5 to 2x what a comparable employee makes, otherwise you're likely getting screwed. The employer pays a lot of taxes that are not deducted from you check, and benefits are worth a few hundred bucks/month too. And how much value can you place of job security....
The real problem is organizations grip tightly to the idea that physical security exists.
The truth is that its only slighty harder for a attacker to get a physical connection to your network than for that same hacker to sit in your parking lot and wirelessly surf.
But, wait, we have id badges, and a security gurd at the door, no one can get to our cables: I once worked with a guy who was paid to do penetration testing, he spent a week wandering around inside the corporate headquarters, until the company IT director declared his attacks unsuccessful (they had no firewall logs of his intrusions, so he must have not got in.) The IT director was displeased with the final report, showing all the data he had accessed (some from the consoles of the "secure" machines) and with the CEO who had agreed that the testing included physical site security.
It becomes even easier when you accept that the vast majority of intrusions come from inside the company, from people who already likely have access to the network.
Sending confidential data in the clear on a wired or wireless network is not a good idea, period.
I am not familiar with Kroger's system, but from the article it looks like they designed it correctly.
Fingerprint scanners are lousy at identifying you from a large database of scans, too many false positives (hash collisions). What fingerprint scanners are good for are _confirming_ that the scan is you. My guess from the article would be Kroger's is using the scans like a pin code, the user presents their club card and the scanner is just used as a quick verification that the customer at the checkout is the person who signed up for the card (and linked it to some payment system).
Privacy concerns are pretty minor (assuming you already agreed to let them track your every purchase by getting the club card) Most of these scanners use a system that hashes key features of the fingerprint, it isn't possible to recreate the actual fingerprint from the scan (it is possible to copy the hash, just like a pin, making it usless in an untrusted env). So the police will not have much use for this db.
Its sound like a good idea, much better than a credit card, which relies on a bored checkout clerk verifying a signature.
And thousands of times better than Speedpass, which AFAIK sends a unique id number (the secret, essentially) in cleartext over a radio link!. Mobil was too cheap to implement a system where the secret wasn't sent in the clear (some simple challenge response would have done it) Tragically Mobil patented the use of RFID for payment, so forget about releasing a better version.
I've met Rob... even had the misfortune of trying to figure out a major project coded by him. Basically he is a nice guy, and a fairly smart one, but is also a bit of a blowhard, and prone to inflating his own importance to stratospheric levels. I would take anything written by him with a grain of salt.
Its also not feasible... it would have been possible back in the prop plane age... but the force of a jet landing would destroy most pavement. (Airport runways are made of a special higher strength concrete)
Oracle's licensing agreements have gotten far weirder in the last few years. Previously the license was based primarily on the platform you were running Oracle on, and the number of concurrent users. Easy to manage, regulate and understand.
Then came the web. Now several thousand people might be connecting to the web server, but the web server only has one connection to the DB.
Oracle tried per CPU licensing.. but that was too expensive and customer's got pissed. A customer running one accountant on a payroll app got charged the same as a multinational company running their entire HR system on the same size box.
Now Oracle is moving back to a per seat licensing system, but their latest contracts include weird provisions that try to count how many end users are actually using the system. Two hundred people using the web server or another piece of middleware require 200 seats of Oracle, regardless of the actual number of connections to the db that are made by the middleware layer. A good idea, except.. once again.. its expensive.. just harder to figure out how expensive. (It gets really convoluted for companies that have middleware that talks to many databases, since each database has to have seats for every person using the middleware, even if some of then never query one of the databases.)
The end result.... Oracle is expensive, and slowly more companies are realizing that they really don't need all the nifty features that come with Oracle. Maybe they'll consider running Postgres.. but then, nobody gets fired for buying Oracle.
I agree, you've already got the experience on your resume that proves you know how to program. If you need the paper to advance pick up a management degree.. these are usually pretty easy (It doesn't take much knowledge to be a manager) and are taught as night courses/distance ed. One of the more successful consultants I know had the same problem. The CEO told him he needed a degree to advance.. so he went and finished his degree in Music.
OTOH.. you're only 24, a good CS degree might help you earn some respect and you're young enough to enjoy the university "experience" (I learned far more goofing off in a lab with some that I ever did in class). You'll want to go to an excellent University and take the time to do it right (3 years if you load up on summer classes and overload credits) If you're as talented as you imply it will set you up for an excellent long term career.
Many car manufacturers are moving to using industry standard buses and protocols (CAN and a couple of SAE standards) for internal communication. There is a lot of "wiggle" room for people to come up with Gizmos that attach to this hardware and do things the manufacturer never intended. For example.. many people love to have a tach, but many dashs lack them. The tach information is available on the computer/diagnostic bus, how about designing a simple PIC circuit to read the RPM message and display it on a LED display.
Lets face it.. hardware hacking in all its forms has gotton harder and harder for the last 20 years as more custom PLC and ASIC devices appear and Surface mount becuase the standard.
If you really want a challenge, convert your old gas powered car to electric. You'll end up with an extremely simple system you can work on yourself (only one moving part in an electric motor and no need for complex computers and emission controls) as well as a car that will get you to work in the worst weather, without the need to warm it up. (Just jump in and go, heat is electric and instantanous) and DC motors can really hual ass. Oh.. and its non-polluting, so you can feel smug about never having to get a smog check again.
Any good auth system (according to Bruce Schneier ) should use two things from the following list:
Something you have. (Smartcard, token card)
Something you know. (Pin, password)
Something you are. (Biometrics, fingerprint iris scan etc
A smartcard + pin solution would be far better than a system that only used on form of auth. A smartcard can be stolen, but without the pin: no access. A password can be evesdropped, but you'd need to swipe the card too.
I have no idea about his qualifications to be a writer, but I do know Doc personally from local Users Group meetings. Certainly he is knowledgable about the subject, mostly becuase of a strong Unix background. I don't think he has contributed anything substantial to Linux, but has has been using and teaching it for better than 3 years. Like Joe Barr (another Austinite who appears in a number of online publications) he is just a strong user and advocate, and not a developer. (But Doc has a much more plesent personality)
In answer to a question further down the page, hydrogen fuel cells are better than batteries because of the rate they can deliver energy. It's difficult to make an electric car that can make a decent top speed. Hydrogen fuel cells pack the punch to give you a good boost.
This is incorrect, while hydrogen fuel cells can in theory develop more Watt/hr per kg than batteries none of the units produced have been able to do so. Most fuel cell powered vehicles will need battery or flywheel systems to store energy for peak loads.
The energy required to accelerate a car quickly is incredible. For example the Electric drag racers require peak current of over 1000amps at 300VDC (300kw, enough to power 75 homes) to run 12 seconds times on the 1/4 mile track. Most average cars Electric cars require 600amps of current. No hydrogen fuel cell on the planet puts out those current levels and can still fit in a car. For those current levels only batteries can deliver energy quickly enough.
Hydrogen's only advantage over battery power is you can refuel quicker.. it might only take 10 minutes to pump enough hydrogen into the car for another 100 miles of driving. Batteries might take 15 minutes to several hours (charge rates are mostly limited on how much electricity is available, most homes do not have sufficent power available to quick charge an EV).
Hydrogen's biggest drawback... its a bitch to store. It leaks out of almost any container you put it in. Its hard to store it as a liquid (have to keep it too cold) and as a gas you can't store enough of it. (hydrogen powered cars average about 100 miles before refueling, only slighty better than batteries)
How long does the data need to be stored for? Tape is good if indefinate storeage is not a requirement. (Tape degrades fast.. but is reusable)
Terabyte tape libraries are fairly common. Check out any of the major datacenter manufacturers. Sun and HP both have a unit of about 7TB. But you're talking several 100k$ for a fully automated unit.
Cheapest route would be to go back to the dark ages. Buy a bunch of 100GB tape drives and lots of tape (70 tapes a day ain't bad). Hire a few minimun wage tape monkeys to change tapes on command. Setup a LED display or a big monitor for the computer to flash tape change commands on. (Old IBM trick)
I can't think of much data that would be sensitive enough to warrent this level of protection. AFAIK even the government feels good encryption and frequent (weekly) inspection of the fiber is good enough to protect critical military operational data.
If your worried about someone with government level resources cracking the data and the amount of data trafficing the pipe is not huge your best bet is a pad cypher (generally considered to be unbreakable). Generate completely random data (atmospheric noise is a good source), burn it to two CDROMs. Encrypt and decrypt the stream of data on each end. You can use a small embedded PC on each end if the data stream is non-standard. Never reuse the same stream of random data.
I'm planning on using a small (15 button) remote for often used functions (Tivo menu keys and volume). And a cheap, small PHP programmabletouchscreen PC and LIRC.
Ham radios would probably be out. And, because we have 900mhz radio communications, relatively unnecessary.
Well.. how fast do you need to go? You can run 56k or so on a 900mhz system. If you have any free channels on your comm system you could probably purchase radio data modems. The base then just needs an "always on" or dial on demand internet connection.
Or if you don't want to pay excessive prices from your radio supplier you can buy a couple of 9600bps TNC's from a HAM radio outlet.
Most of the inexpensive 2way satellite services require tricky aiming to get them working, probably beyond what your users are capable of. Automatic or aim free systems are expensive in equipment costs or services.
I found old Pentium laptops to make excellent firewalls. They are a little more pricey than the old PC but they have a few advantages:
Built in battery backup
Low power consumption
Few (if any) noisy fans
Small, and fit nicely in a rack shelf
Built in collapsible console
Look around and you can find one for about the same price as the small NAT routers. The only real shame is they only have typically two PCMCIA slots, so you can't have a DMZ or wireless net interface seperate from the internal and external interfaces.
We are having similar problems, most of the (unnamed) agency's documents are in Word and Wordperfect format, neither of which is terrible portable or searchable. Lots of documentation is rotting on random harddrives, or being lost of machines are surplused.
The biggest problem is the dual government requirements to post most of these documents on the web, and at the same time make those webpages ADA compliant (which requires seperating content from presentation, and simplfying documents so they can be read by screenreader.) Neither Word or Wordperfect is capable of creating standards compliant HTML so the exporters are useless.
Currently our "documentation management" system appears to consist of a few web developers who cut and paste ASCII text from documents and HTML format it.
I offered to start an Open Source project that would be an XML based document database (possibly using the Star Office DTD as a base) figuring that more than a few organizations could use a tool like this. I received the "but if you leave we won't have a vendor to turn to! Where can we buy a support contract?" and "We don't do internal development, we outsource programming tasks" and "We can't afford to retrain everyone." (Ignoring the fact that they are retraining everyone from Wordperfect to Word as we speak.)
Really? My copy of Word 2k doesn't seem to save native in XML, nor can I find any options concerning it. We would love to see this feature, since most of our technical documents are in word and the inability to search the documents is killing us. In fact XML is the primary reason we are examining OpenOffice.
We're faced with a problem of a scarce but valuable resource. As usual the government and the corporations that control it are loudly contemplating how much money its worth, but everyone forgets that the government holds this spectrum in the public trust. The government remembers this occasionally, which is why there were minority clauses in the last spectrum auctions, allowing disadvantaged organizations to buy spectrum at a reduced price.. a dismal failure since it turned out small organizations didn't have enough money to build giant centralized systems using that spectrum.
It turns out the idea that spectrum must be parcelled out to monopolies in order to avoid interferce is largely a lie. New technologies like spread spectrum make it possible to cram far more signals into the same spectrum and do so in a decentralized way. Take the unlicensed 2.4GHz band for example, this bit of free for all spectrum suffers from some interference, but at the same time wireless devices utilizing it (cordless phones, 802.11) are dirt cheap, and widely available.
The best (for the public) way to parcel out 3G spectrum is to make it unlicensed, and force everyone one to the same playing field.
At this point studios want CPU cycles cheap, and they are already comfortable writing toolchains on Unix.
Linux combines the best of both worlds, cheap fast PC hardware and Unix. One studio said they could afford to replace their Linux cluster twice as often as the SGI renderfarm (since it cost half as much) so they could keep themselves closer to the state of the art in processing power.
SGI used to offer awesome custom graphics acceleration hardware but custom hardware limits choice, and costs more than general purpose stuff. And the general purpose stuff is nearly as fast.
As much as I hate to make a me too post.. I would agree. Nothing is more frustrating than watching the downloads count up and not getting any feedback. Good, bad or ugly, at least the programmer knows people are using it. Even bug reports... Anything to feed a starving ego.
Look for people with a broad knowledge of many technologies, even if they aren't experts. You're looking for people who might not know everything, but have a large enough framework knowledge and the willingness to learn anything new.
The easist way to weed out the "Quick Study Course" MCSEs is to ask them about thier experiences/knowledge on Unix/Linux (even if they don't activily use Linux any competent sysadmin has read about it).
If the position is going to be inside a team, and the interviewee seems pretty comfortable, declare the interview over with. Then take the interviewee to the breakroom/lunch and arraige for the other team members to drift over. (Don't go to someone's office to say hello.. this puts the interviewee on unfamilar turf) Maybe have one of the team members toss out a problem they're working on or give a status report. See if you can get the interviewee to interject ideas or solutions. They'll be pretty nervous, so don't hold it against them for being quiet but the really good ones will love talking shop and may even give some free advice. Plus this makes the team members feel more involved in the process.
The biggest thing I can say.. is look for experiences outside the workplace. If someone did something for fun, odds are good that they learned more about it than they ever could of on the job or in a classroom.
With some PHP and a little thought you could do the a similar thing using my Linux based ETC software and a $50 surplus touchscreen computer. Monochrome only, unfortunetly.
If there is demand for this I might be able to wire the touchscreen inline between the keyboard and PC (for compatibility with Windows and Linux).
Until more business' make a switch from Wintel, its really just not that valuable a skill
A common misconception. In truth it is dumb to think you can teach high schoolers (or even college students) the one Dominant Word Processor business uses and they'll be prepared. By the time they exit school business will be using a different version or even a different product.
A better policy is to show the students several applications and stress the common interface. Every Word processor from nano/pico, to Office XP has certain things in common. The only differences is how many features they have, and where they appear on the menus. Teach what kind of tools are available, from simple things like spell checking, to more advanced functions like mail merge and macros. Stress how to use help and reference materials to find solutions (problem solving skills). Let them learn that there is seldom one "right" way to get a task done, and to explore their options (nice area to introduce free software).
You want students to walk away not with time sensitive reference knowledge (they can purchase dummies books when they need a reference) but an understanding of how to learn each new software package as it is introduced.
But teaching this way requires extremely knowledgable and up to date teachers and doesn't fit well with the teaching striaght from a book methodology that is so popular.
Slashdot Mirror
I had the same issue come up at my current contract. They wanted a fixed cost, knowing that the number of hours worked would be difficult to track (sysadmin support). Since the whole contract is pretty small, we agree to a flat rate for a limited contract time, and I insisted on a "nominal 40 hr work week" clause in the contract. This is still probably too vague.. but it was a decent compromise, and since the contract length is only 8 weeks I did not worry about abuse.
You have nothing to worry about from the IRS (as long as you pay them), but the employer does. The IRS has been taking a pretty tough stance on defining employees. Pretty much any individual paid by a company is an employee, and hence the employer is responsible for withholdings. The company must prove that you are not an employee (they must have an "arms length" relationship, which is up to the IRS to decide... but if you work at their location, are paid a flat rate, and are solely working for them, the IRS is likely to call you an employee). A contract is not enough. Which means if you don't pay your self employement withholdings, the company gets fined..
Also if you are a contractor, you should be paid about 1.5 to 2x what a comparable employee makes, otherwise you're likely getting screwed. The employer pays a lot of taxes that are not deducted from you check, and benefits are worth a few hundred bucks/month too. And how much value can you place of job security....
Fun, eh?
The real problem is organizations grip tightly to the idea that physical security exists.
The truth is that its only slighty harder for a attacker to get a physical connection to your network than for that same hacker to sit in your parking lot and wirelessly surf.
But, wait, we have id badges, and a security gurd at the door, no one can get to our cables: I once worked with a guy who was paid to do penetration testing, he spent a week wandering around inside the corporate headquarters, until the company IT director declared his attacks unsuccessful (they had no firewall logs of his intrusions, so he must have not got in.) The IT director was displeased with the final report, showing all the data he had accessed (some from the consoles of the "secure" machines) and with the CEO who had agreed that the testing included physical site security.
It becomes even easier when you accept that the vast majority of intrusions come from inside the company, from people who already likely have access to the network.
Sending confidential data in the clear on a wired or wireless network is not a good idea, period.
I am not familiar with Kroger's system, but from the article it looks like they designed it correctly.
Fingerprint scanners are lousy at identifying you from a large database of scans, too many false positives (hash collisions). What fingerprint scanners are good for are _confirming_ that the scan is you. My guess from the article would be Kroger's is using the scans like a pin code, the user presents their club card and the scanner is just used as a quick verification that the customer at the checkout is the person who signed up for the card (and linked it to some payment system).
Privacy concerns are pretty minor (assuming you already agreed to let them track your every purchase by getting the club card) Most of these scanners use a system that hashes key features of the fingerprint, it isn't possible to recreate the actual fingerprint from the scan (it is possible to copy the hash, just like a pin, making it usless in an untrusted env). So the police will not have much use for this db.
Its sound like a good idea, much better than a credit card, which relies on a bored checkout clerk verifying a signature.
And thousands of times better than Speedpass, which AFAIK sends a unique id number (the secret, essentially) in cleartext over a radio link!. Mobil was too cheap to implement a system where the secret wasn't sent in the clear (some simple challenge response would have done it) Tragically Mobil patented the use of RFID for payment, so forget about releasing a better version.
I've met Rob... even had the misfortune of trying to figure out a major project coded by him. Basically he is a nice guy, and a fairly smart one, but is also a bit of a blowhard, and prone to inflating his own importance to stratospheric levels. I would take anything written by him with a grain of salt.
Its also not feasible... it would have been possible back in the prop plane age... but the force of a jet landing would destroy most pavement. (Airport runways are made of a special higher strength concrete)
Then came the web. Now several thousand people might be connecting to the web server, but the web server only has one connection to the DB.
Oracle tried per CPU licensing.. but that was too expensive and customer's got pissed. A customer running one accountant on a payroll app got charged the same as a multinational company running their entire HR system on the same size box.
Now Oracle is moving back to a per seat licensing system, but their latest contracts include weird provisions that try to count how many end users are actually using the system. Two hundred people using the web server or another piece of middleware require 200 seats of Oracle, regardless of the actual number of connections to the db that are made by the middleware layer. A good idea, except.. once again.. its expensive.. just harder to figure out how expensive. (It gets really convoluted for companies that have middleware that talks to many databases, since each database has to have seats for every person using the middleware, even if some of then never query one of the databases.)
The end result.... Oracle is expensive, and slowly more companies are realizing that they really don't need all the nifty features that come with Oracle. Maybe they'll consider running Postgres.. but then, nobody gets fired for buying Oracle.
I agree, you've already got the experience on your resume that proves you know how to program. If you need the paper to advance pick up a management degree.. these are usually pretty easy (It doesn't take much knowledge to be a manager) and are taught as night courses/distance ed. One of the more successful consultants I know had the same problem. The CEO told him he needed a degree to advance.. so he went and finished his degree in Music.
OTOH.. you're only 24, a good CS degree might help you earn some respect and you're young enough to enjoy the university "experience" (I learned far more goofing off in a lab with some that I ever did in class). You'll want to go to an excellent University and take the time to do it right (3 years if you load up on summer classes and overload credits) If you're as talented as you imply it will set you up for an excellent long term career.
This guy has pinouts and protocols for Kenwood, Panasonic and Pioneed changers. Might give you some ideas.
m
http://www-user.tu-chemnitz.de/~harn/mp3_cdc.ht
Mark
Many car manufacturers are moving to using industry standard buses and protocols (CAN and a couple of SAE standards) for internal communication. There is a lot of "wiggle" room for people to come up with Gizmos that attach to this hardware and do things the manufacturer never intended. For example.. many people love to have a tach, but many dashs lack them. The tach information is available on the computer/diagnostic bus, how about designing a simple PIC circuit to read the RPM message and display it on a LED display.
Lets face it.. hardware hacking in all its forms has gotton harder and harder for the last 20 years as more custom PLC and ASIC devices appear and Surface mount becuase the standard.
If you really want a challenge, convert your old gas powered car to electric. You'll end up with an extremely simple system you can work on yourself (only one moving part in an electric motor and no need for complex computers and emission controls) as well as a car that will get you to work in the worst weather, without the need to warm it up. (Just jump in and go, heat is electric and instantanous) and DC motors can really hual ass. Oh.. and its non-polluting, so you can feel smug about never having to get a smog check again.
- Something you have. (Smartcard, token card)
- Something you know. (Pin, password)
- Something you are. (Biometrics, fingerprint iris scan etc
A smartcard + pin solution would be far better than a system that only used on form of auth. A smartcard can be stolen, but without the pin: no access. A password can be evesdropped, but you'd need to swipe the card too.The best security is a layered defense...
I have no idea about his qualifications to be a writer, but I do know Doc personally from local Users Group meetings. Certainly he is knowledgable about the subject, mostly becuase of a strong Unix background. I don't think he has contributed anything substantial to Linux, but has has been using and teaching it for better than 3 years. Like Joe Barr (another Austinite who appears in a number of online publications) he is just a strong user and advocate, and not a developer. (But Doc has a much more plesent personality)
In answer to a question further down the page, hydrogen fuel cells are better than batteries because of the rate they can deliver energy. It's difficult to make an electric car that can make a decent top speed. Hydrogen fuel cells pack the punch to give you a good boost.
This is incorrect, while hydrogen fuel cells can in theory develop more Watt/hr per kg than batteries none of the units produced have been able to do so. Most fuel cell powered vehicles will need battery or flywheel systems to store energy for peak loads.
The energy required to accelerate a car quickly is incredible. For example the Electric drag racers require peak current of over 1000amps at 300VDC (300kw, enough to power 75 homes) to run 12 seconds times on the 1/4 mile track. Most average cars Electric cars require 600amps of current. No hydrogen fuel cell on the planet puts out those current levels and can still fit in a car. For those current levels only batteries can deliver energy quickly enough.
Hydrogen's only advantage over battery power is you can refuel quicker.. it might only take 10 minutes to pump enough hydrogen into the car for another 100 miles of driving. Batteries might take 15 minutes to several hours (charge rates are mostly limited on how much electricity is available, most homes do not have sufficent power available to quick charge an EV).
Hydrogen's biggest drawback... its a bitch to store. It leaks out of almost any container you put it in. Its hard to store it as a liquid (have to keep it too cold) and as a gas you can't store enough of it. (hydrogen powered cars average about 100 miles before refueling, only slighty better than batteries)
Gonna be expensive,
How long does the data need to be stored for? Tape is good if indefinate storeage is not a requirement. (Tape degrades fast.. but is reusable)
Terabyte tape libraries are fairly common. Check out any of the major datacenter manufacturers. Sun and HP both have a unit of about 7TB. But you're talking several 100k$ for a fully automated unit.
Cheapest route would be to go back to the dark ages. Buy a bunch of 100GB tape drives and lots of tape (70 tapes a day ain't bad). Hire a few minimun wage tape monkeys to change tapes on command. Setup a LED display or a big monitor for the computer to flash tape change commands on. (Old IBM trick)
Mark
I can't think of much data that would be sensitive enough to warrent this level of protection. AFAIK even the government feels good encryption and frequent (weekly) inspection of the fiber is good enough to protect critical military operational data.
If your worried about someone with government level resources cracking the data and the amount of data trafficing the pipe is not huge your best bet is a pad cypher (generally considered to be unbreakable). Generate completely random data (atmospheric noise is a good source), burn it to two CDROMs. Encrypt and decrypt the stream of data on each end. You can use a small embedded PC on each end if the data stream is non-standard. Never reuse the same stream of random data.
I'm planning on using a small (15 button) remote for often used functions (Tivo menu keys and volume). And a cheap, small PHP programmabletouchscreen PC and LIRC.
Well.. how fast do you need to go? You can run 56k or so on a 900mhz system. If you have any free channels on your comm system you could probably purchase radio data modems. The base then just needs an "always on" or dial on demand internet connection.
Or if you don't want to pay excessive prices from your radio supplier you can buy a couple of 9600bps TNC's from a HAM radio outlet.
Most of the inexpensive 2way satellite services require tricky aiming to get them working, probably beyond what your users are capable of. Automatic or aim free systems are expensive in equipment costs or services.
I found old Pentium laptops to make excellent firewalls. They are a little more pricey than the old PC but they have a few advantages:
Built in battery backup
Low power consumption
Few (if any) noisy fans
Small, and fit nicely in a rack shelf
Built in collapsible console
Look around and you can find one for about the same price as the small NAT routers. The only real shame is they only have typically two PCMCIA slots, so you can't have a DMZ or wireless net interface seperate from the internal and external interfaces.
We are having similar problems, most of the (unnamed) agency's documents are in Word and Wordperfect format, neither of which is terrible portable or searchable. Lots of documentation is rotting on random harddrives, or being lost of machines are surplused.
The biggest problem is the dual government requirements to post most of these documents on the web, and at the same time make those webpages ADA compliant (which requires seperating content from presentation, and simplfying documents so they can be read by screenreader.) Neither Word or Wordperfect is capable of creating standards compliant HTML so the exporters are useless.
Currently our "documentation management" system appears to consist of a few web developers who cut and paste ASCII text from documents and HTML format it.
I offered to start an Open Source project that would be an XML based document database (possibly using the Star Office DTD as a base) figuring that more than a few organizations could use a tool like this. I received the "but if you leave we won't have a vendor to turn to! Where can we buy a support contract?" and "We don't do internal development, we outsource programming tasks" and "We can't afford to retrain everyone." (Ignoring the fact that they are retraining everyone from Wordperfect to Word as we speak.)
I gave up.
Really? My copy of Word 2k doesn't seem to save native in XML, nor can I find any options concerning it. We would love to see this feature, since most of our technical documents are in word and the inability to search the documents is killing us. In fact XML is the primary reason we are examining OpenOffice.
We're faced with a problem of a scarce but valuable resource. As usual the government and the corporations that control it are loudly contemplating how much money its worth, but everyone forgets that the government holds this spectrum in the public trust. The government remembers this occasionally, which is why there were minority clauses in the last spectrum auctions, allowing disadvantaged organizations to buy spectrum at a reduced price.. a dismal failure since it turned out small organizations didn't have enough money to build giant centralized systems using that spectrum.
It turns out the idea that spectrum must be parcelled out to monopolies in order to avoid interferce is largely a lie. New technologies like spread spectrum make it possible to cram far more signals into the same spectrum and do so in a decentralized way. Take the unlicensed 2.4GHz band for example, this bit of free for all spectrum suffers from some interference, but at the same time wireless devices utilizing it (cordless phones, 802.11) are dirt cheap, and widely available.
The best (for the public) way to parcel out 3G spectrum is to make it unlicensed, and force everyone one to the same playing field.
At this point studios want CPU cycles cheap, and they are already comfortable writing toolchains on Unix.
Linux combines the best of both worlds, cheap fast PC hardware and Unix. One studio said they could afford to replace their Linux cluster twice as often as the SGI renderfarm (since it cost half as much) so they could keep themselves closer to the state of the art in processing power.
SGI used to offer awesome custom graphics acceleration hardware but custom hardware limits choice, and costs more than general purpose stuff. And the general purpose stuff is nearly as fast.
As much as I hate to make a me too post.. I would agree. Nothing is more frustrating than watching the downloads count up and not getting any feedback. Good, bad or ugly, at least the programmer knows people are using it. Even bug reports... Anything to feed a starving ego.
Want GPL'd touchscreen home control: ETC, the Extensible Touchscreen Controller.
Look for people with a broad knowledge of many technologies, even if they aren't experts. You're looking for people who might not know everything, but have a large enough framework knowledge and the willingness to learn anything new.
The easist way to weed out the "Quick Study Course" MCSEs is to ask them about thier experiences/knowledge on Unix/Linux (even if they don't activily use Linux any competent sysadmin has read about it).
If the position is going to be inside a team, and the interviewee seems pretty comfortable, declare the interview over with. Then take the interviewee to the breakroom/lunch and arraige for the other team members to drift over. (Don't go to someone's office to say hello.. this puts the interviewee on unfamilar turf) Maybe have one of the team members toss out a problem they're working on or give a status report. See if you can get the interviewee to interject ideas or solutions. They'll be pretty nervous, so don't hold it against them for being quiet but the really good ones will love talking shop and may even give some free advice. Plus this makes the team members feel more involved in the process.
The biggest thing I can say.. is look for experiences outside the workplace. If someone did something for fun, odds are good that they learned more about it than they ever could of on the job or in a classroom.
With some PHP and a little thought you could do the a similar thing using my Linux based ETC software and a $50 surplus touchscreen computer. Monochrome only, unfortunetly.
If there is demand for this I might be able to wire the touchscreen inline between the keyboard and PC (for compatibility with Windows and Linux).
A common misconception. In truth it is dumb to think you can teach high schoolers (or even college students) the one Dominant Word Processor business uses and they'll be prepared. By the time they exit school business will be using a different version or even a different product.
A better policy is to show the students several applications and stress the common interface. Every Word processor from nano/pico, to Office XP has certain things in common. The only differences is how many features they have, and where they appear on the menus. Teach what kind of tools are available, from simple things like spell checking, to more advanced functions like mail merge and macros. Stress how to use help and reference materials to find solutions (problem solving skills). Let them learn that there is seldom one "right" way to get a task done, and to explore their options (nice area to introduce free software).
You want students to walk away not with time sensitive reference knowledge (they can purchase dummies books when they need a reference) but an understanding of how to learn each new software package as it is introduced.
But teaching this way requires extremely knowledgable and up to date teachers and doesn't fit well with the teaching striaght from a book methodology that is so popular.