Actually, more cell towers than ever are using licensed wireless and microwave to connect back to other hub sites. AT&T and Verizon are the ones pushing this technology more than others.
The FAA has changed their rules in the last three years. They no longer have to work with another airline to get you to your final destination. There is now a timer that starts that if they don't get you there by any means within a comparable passage class within 2 hours of your scheduled destination, they owe you cash, based on some formula that is basically a refund of your ticket, plus possibly a penalty. If they know that they can't get you there within two hours, you fall prey to the machine and hope you find somebody who will work with you to make it right. The FAA no longer tracks airlines that bump passengers, and if you do get involuntary refused booking, you have to go to arbitration, and possibly court to get the full refund allowed by the FAA.
Having been one of the people that got bumped from a flight because they overbooked and they couldn't get enough volunteers, it does happen. First time I've flown United in about a decade (I usually fly Delta because they have the most coverage in the Midwest), so I didn't have any status. I was stuck in SF for three days. They put me up in a roach motel, and I ended up having to pay for my taxi to and from the airport each day to wait to catch the next flight that had an available seat. The compensation? $800, which half was used up in transportation, food, etc. It didn't even come close to covering the two days I missed at work.
They need to re-reimburse you up to about two times the cost of the leg of the flight they bump you, plus the average cost of the other legs you couldn't make. They sometimes will offer more to make you go away quietly. The amount they have to pay out is the same if you don't make it to the final destination 2 hours late as it is a week late. They have no incentive to get you there if they missed their mark. If you end up flying with multiple carriers on the same itinerary, then you can expect to be even more screwed since those other legs may not be covered.
Back in the 90's and early 00's, it was pretty rare to be completely screwed if they bumped you. Often times they would work with other airlines to get you to your destination as soon as possible. They also didn't overbook every flight going to common places, they would usually keep one flight a day at 95% with the others at 110%, so that if there was a failure somewhere along the way, they had seats to move people to. Now they pretty much every flight at 110%+ with little regard of what happens to the passengers.
You do realize that providing internet access to a device that is traveling at 300MPH+ is not exactly as simple as upgrading a WiFi access point... The WiFi system in the planes is not the problem -- it's the LMRS that either uses a point-to-point antenna or satellite system to provide internet access.
The nice thing about having those screens is that you don't have to have your laptop open all the time. Sometimes you want to just sit back and not have to juggle a laptop or ipad on your lap while everything else is going on around you.
Sure. Every inch of the world has free and open internet access. There isn't a single country that blocks websites, intercepts data, blocks VPNs or does anything else with commodity traffic that would stop this from working. Oh, and the internet is never disturbed when a country is in crisis. Ever.
I'm imagining a situation where this kid was the rock-star of the organization, which was pretty weak in the IT side. He wrote the systems that controlled everything, and he probably setup all the servers, etc. Since he was pretty young, he didn't use proper AAA systems, and probably created logins for himself on all the systems. He probably knew all the back-doors and other ways to get into the systems since he built it.
When he got fired, his employer probably removed his key-card access (if they used one), and most likely his email. They didn't/know/ about all the shadow systems and logins that exist because he didn't document them. And that is where the problem began...
But here is the kicker -- Senator A and Senator C both have towns in their districts that build floatys and arm-wings for the Navy... The Navy doesn't actually need them, and if they completely control their own spending, they will cancel the contract. Both Senators will have heavy job losses in their districts, spurning hard economic times, making it harder for them to get re-elected. If they write a bill that dictates that the Navy buys all these floatys and arm-wings for the Navy, then they saved jobs and are heroes... The same goes for army/navy bases, etc. Heck, it's reasons like that, that there is a fully stocked air-force armory in my town -- 75 miles away from the closest air-force base and 30 miles away from the closest airport that can even land a jet.
And that is why things don't chance. If we made these organizations more efficient and allowed them to spend in ways that are actually useful to them, lots of people lose their jobs (mostly people in congress, but you know)...
CLID and ANI are the same thing. The BTN is what you are really looking for (The Billing Telephone Number)... Field 0x71 on SS7, according to Telcordia...
The database that does the CLID -> Name lookup is owned by a company called NeuStar. They charge the telephone companies roughly a penny for each lookup they perform for each call. That's why there is a charge for pretty much every company to provide this data...
Sure, autodialers, by law are supposed to trust those tones. Guess what -- the people making these scam calls don't care, and often don't respect them.
Actually, since digital switching began in the 60's and 70's, there have been three fields transmitted with every call (well, a lot more, but these are relevant) BTN = Bill To Number -- this is the number that the call is billed to. This is actually validated by the connecting carrier, and still is today. In most cases it will be the circuit number, SPID, or an account number for really large customers. CPN = Calling Party Number -- this is the number that the call is presenting itself as -- the Caller ID if you will. A long time ago, this was always validated by the phone company against the customer's record of DIDs. In the early 90's the LECs started charging companies to open up this field so that they could hide call center numbers, etc. and to make their phone number their brand. In the late 90's some LECs started offering this as a standard feature as a differentiation against other CLECs. RTN = Route To Number -- this is the number the call is destine to.
This biggest problem is that we started getting a lot of smaller CLECs that didn't understand the technology well enough and started giving everybody closer access to the PSTN (for example, by not watching the CPN they were sending). The problem was exacerbated when VoIP became a thing and CLECs started allowing anybody access to the PSTN with no restrictions and no regard to their physical location.
These scams are hard to track down. I'd venture to say that 80% of them are running on stolen credit cards, on AWS (or other cloud provider) EC2 instances, connected to some VoIP provider that is billing another stolen credit card. They connect their SIP phones from anywhere to the PBX in the cloud and they start. Labor is cheap in other places in the world and with everything being in the cloud they can be pretty much anywhere. If they get shut down, they just use another stolen credit card and launch another EC2 instance and they are back in business a few minutes later.
The problem is that most of these scam calls are originating from outside the United States. Our laws can't do much outside the US without a lot of legal paperwork -- and in most cases it won't be worth it.
One easy solution is to give consumers access to the BTN or Bill-To phone number. This is the number that is being billed for the call -- essentially pinning down the place where the call is being switched into the PSTN. If you get the BTN, you get the person behind the call -- regardless of what their Caller ID is. Unfortunately, right now, the only way to get access to the BTN is via the SS7 protocol (not available to consumers), or to compel your phone company to give it via a subpoena. Enough abuse from a single BTN -- cut them off until they can clean up their act.
You do realize that you still run all the Adobe software on your local computer -- the only thing that makes it "the cloud" is that it includes a document management and file storage piece. Oh, and I think/some/ settings are shared across installs of the software.
When they switched to this model, they also changed it so you rent the software instead of purchase it. The advantage is that they push updates more frequently, but you are paying a monthly fee for it.
Sure. But the hard things are :
- Ads, and tracking their placement, etc.
- Encrypted Content (most content producers don't want their content streamed unencrypted -- and that causes issues for vanilla browser deployments)
- Streaming (this becomes less trivial if you are looking to utilize existing infrastructure to stream to the browser).
And these are why HTML5 video is still slow to roll out. Once the HTML5 spec had a basic video player, everybody moved onto the next shiny object and left the rest so everybody had to come up with their own solution.
A lot of it has to do with the states dropping their support of higher-ed. In Michigan, as recently as 2000, 80% of the major Universities' operating budget came from the state. In 2015, it was down to 15%. Costs to educate each student (budget / number of students) has been flat, without considering inflation. Funding sources from outside the state have gone up, but not enough to offset the difference. Consequently, tuition used to cost $135/credit hour for in state, and now it costs $375/credit hour.
According to the IRS, Non-profits (501(c)3 organizations) are owned by the community and are for the greater good.
501(c)3's don't have owners or shareholders. They can have a board of directors (and should), and they can have employees. Those employees can be paid like in any other business.
When I helped run our Boy Scout troop when I was much younger, I learned some things about the United Way. In order to become eligible to receive a donation, you have to sign your organization up to volunteer for the United Way. In our case, we had to help run one of their call centers for I think three or four weekends. THEN, after you become eligible, you then have to buy their merch -- things like United Way flags, shirts, etc to promote them. Finally, the donation will usually come at an awards dinner -- which a few people from your org have to attend (and pay for).
They gave us a check for $1,200 for a new trailer. Not counting volunteer time, the troop ended up paying about $1,000 in fees, merch and expenses to participate. Sure, next year it would have been better but to tie up 8 guys for 4 weekends for $200, there were much better ways to get donations or raise money (selling popcorn would usually net us about $8,000 and selling Christmas Trees would net us closer to $10,000).
In Michigan it's not easy to become a dealership (not a franchise -- they can't sell cars).
You first have to join the dealer trade association. Then you have to apply to become a franchise with the state. The Secretary of State decides your initial market area that you are allowed to cover. If you don't belong to the dealer trade association, you immediately are not given a territory, and most likely won't be approved by the State. In order to get your dealership license, you also have to have a setup for service, dollars spent, setup with the SOS for tab registration, etc. Essentially you have to dump a ton of money into the location right away -- in essence a huge barrier to entry for anybody starting up.
This actually started with Wallmart. When Wallmart started carrying electronics, they were demanding to be cheaper than their competitors. They used their strength as a retailer to force the vendors to make custom versions of their products to accomplish this. On printers, they would ship without ink, for computers they would have 30 or 60 day warranty instead of 1 or 2 years. For TVs, they may have had certain featured disabled (for example, DVI ports were there, but you couldn't tune to them).
MOST electronics stores carried the same SKU at the time, and Wallmart/Sams Club had their own. Then other retailers started demanding custom packages -- so they got their own SKUs too. It's not that bad, but you will typically see 3 or 4 different SKUs... a discount retailer (Wallmart), a big box retailer (BBY/Micro Center), and online retailer (Amazon) and sometimes a SKU for premium retailers.
And I think that is the key. They were upfront about it, and allowed you to make an informed decision. The owners of this wrkriot company lied and deceived their employees about what was going on.
If there are rough waters ahead, be open and honest about it. Most people will understand and try and help out. In this case, they lied about what their situation was to most of the employees they brought on and continued lying when they ran out of runway and couldn't pay them anymore.
The crazy thing was they leveraged themselves so much, they had no way to control their spending. Private jet service? Hiring dozens of employees when you are on the edge of the cliff? Just speaks volumes of their willingness to live reality vs. living the dream.
People who build real big data centers are usually too smart to buy Cisco. For the last 10 years, people buy Cisco for the name, not for the product. Cisco does not perform as well, they cost more, and they are harder to use than products from their competitors. They also like to do things their own ways -- which makes compatibility issues a real deal when you try to work with other vendors.
Cisco's deal was to talk to the CEOs, sell them over a nice steak and not consider tech specs. For a SMB or medium sized business, that was easy to do. Large technology firms, like data centers, actually pay attention to what they are buying.
The problem with your example of Amazon is that Amazon invests every penny they earn back into the business. Companies like Delta don't. So when there are bad times, Amazon will be much better poised to do well because they've diversified and built up their business to handle it. All it takes is a generator to malfunction and Delta could be out of business forever (yes, a bit of a stretch, but still).
Delta has a virtual monopoly for a large swath of the nation. You have no choice to fly delta in the midwest and large portions of the south. In recent history, they were never/that/ bad off. In the last market crash, they decided to burn their cash on buying NorthWest instead of modernizing their own systems or investing in their own infrastructure. In the last three years where they've been posting record profits they continued to do cost-cutting measures in all portions of their business, and move money out of the business by paying investors and the execs.
From the reports I've been seeing, it wasn't the ATS that failed, but rather a generator that caught on fire -- and in order to extinguish the fire safely, they had to cut commercial power.
Freak accidents like that happen. But what also happens is that companies that big invest in redundant systems in geo-redundant locations. What happens if a tornado, sharknado or other natural disaster happens and takes out the physical servers? Does Delta just cancel flights for the next month while they rebuild?
If you compare any of the APIs from Google, Microsoft and Amazon, you will clearly see why the different groups are in the place they are.
Google's API set is probably one of the crappiest I've ever seen. It's impossible to do anything with their service unless you use their pre-baked SDK. Sure, it's a REST api, but you can't authenticate against it, because they won't really tell you how -- only why you wouldn't want to do it. They have no docs on how to use their APIs with just CURL.
Microsoft's is better. Their APIs are a pain (mostly because they keep changing), but at least they are pretty well documented and done in a way that you can actually use if you want to. They offer a really rich set of features.... but they do keep changing them on the fly and don't really version stuff like you would expect.
Amazon knows how to API enable their stuff. Their own services and tools use their own published API to do things. They give lots of examples in a bunch of different languages. If you write against it, it will pretty much work forever unless you change your own setup.
Slashdot Mirror
Actually, more cell towers than ever are using licensed wireless and microwave to connect back to other hub sites. AT&T and Verizon are the ones pushing this technology more than others.
The FAA has changed their rules in the last three years. They no longer have to work with another airline to get you to your final destination. There is now a timer that starts that if they don't get you there by any means within a comparable passage class within 2 hours of your scheduled destination, they owe you cash, based on some formula that is basically a refund of your ticket, plus possibly a penalty. If they know that they can't get you there within two hours, you fall prey to the machine and hope you find somebody who will work with you to make it right. The FAA no longer tracks airlines that bump passengers, and if you do get involuntary refused booking, you have to go to arbitration, and possibly court to get the full refund allowed by the FAA.
Having been one of the people that got bumped from a flight because they overbooked and they couldn't get enough volunteers, it does happen. First time I've flown United in about a decade (I usually fly Delta because they have the most coverage in the Midwest), so I didn't have any status. I was stuck in SF for three days. They put me up in a roach motel, and I ended up having to pay for my taxi to and from the airport each day to wait to catch the next flight that had an available seat. The compensation? $800, which half was used up in transportation, food, etc. It didn't even come close to covering the two days I missed at work.
They need to re-reimburse you up to about two times the cost of the leg of the flight they bump you, plus the average cost of the other legs you couldn't make. They sometimes will offer more to make you go away quietly. The amount they have to pay out is the same if you don't make it to the final destination 2 hours late as it is a week late. They have no incentive to get you there if they missed their mark. If you end up flying with multiple carriers on the same itinerary, then you can expect to be even more screwed since those other legs may not be covered.
Back in the 90's and early 00's, it was pretty rare to be completely screwed if they bumped you. Often times they would work with other airlines to get you to your destination as soon as possible. They also didn't overbook every flight going to common places, they would usually keep one flight a day at 95% with the others at 110%, so that if there was a failure somewhere along the way, they had seats to move people to. Now they pretty much every flight at 110%+ with little regard of what happens to the passengers.
You do realize that providing internet access to a device that is traveling at 300MPH+ is not exactly as simple as upgrading a WiFi access point... The WiFi system in the planes is not the problem -- it's the LMRS that either uses a point-to-point antenna or satellite system to provide internet access.
The nice thing about having those screens is that you don't have to have your laptop open all the time. Sometimes you want to just sit back and not have to juggle a laptop or ipad on your lap while everything else is going on around you.
I guess the positive side of VMWare is that whey they pick a name, they usually only stick with it for about 6 months or so.
Sure. Every inch of the world has free and open internet access. There isn't a single country that blocks websites, intercepts data, blocks VPNs or does anything else with commodity traffic that would stop this from working. Oh, and the internet is never disturbed when a country is in crisis. Ever.
I'm imagining a situation where this kid was the rock-star of the organization, which was pretty weak in the IT side. He wrote the systems that controlled everything, and he probably setup all the servers, etc. Since he was pretty young, he didn't use proper AAA systems, and probably created logins for himself on all the systems. He probably knew all the back-doors and other ways to get into the systems since he built it.
When he got fired, his employer probably removed his key-card access (if they used one), and most likely his email. They didn't /know/ about all the shadow systems and logins that exist because he didn't document them. And that is where the problem began...
But here is the kicker -- Senator A and Senator C both have towns in their districts that build floatys and arm-wings for the Navy... The Navy doesn't actually need them, and if they completely control their own spending, they will cancel the contract. Both Senators will have heavy job losses in their districts, spurning hard economic times, making it harder for them to get re-elected. If they write a bill that dictates that the Navy buys all these floatys and arm-wings for the Navy, then they saved jobs and are heroes... The same goes for army/navy bases, etc. Heck, it's reasons like that, that there is a fully stocked air-force armory in my town -- 75 miles away from the closest air-force base and 30 miles away from the closest airport that can even land a jet.
And that is why things don't chance. If we made these organizations more efficient and allowed them to spend in ways that are actually useful to them, lots of people lose their jobs (mostly people in congress, but you know)...
CLID and ANI are the same thing. The BTN is what you are really looking for (The Billing Telephone Number)... Field 0x71 on SS7, according to Telcordia...
The database that does the CLID -> Name lookup is owned by a company called NeuStar. They charge the telephone companies roughly a penny for each lookup they perform for each call. That's why there is a charge for pretty much every company to provide this data...
Sure, autodialers, by law are supposed to trust those tones. Guess what -- the people making these scam calls don't care, and often don't respect them.
Actually, since digital switching began in the 60's and 70's, there have been three fields transmitted with every call (well, a lot more, but these are relevant)
BTN = Bill To Number -- this is the number that the call is billed to. This is actually validated by the connecting carrier, and still is today. In most cases it will be the circuit number, SPID, or an account number for really large customers.
CPN = Calling Party Number -- this is the number that the call is presenting itself as -- the Caller ID if you will. A long time ago, this was always validated by the phone company against the customer's record of DIDs. In the early 90's the LECs started charging companies to open up this field so that they could hide call center numbers, etc. and to make their phone number their brand. In the late 90's some LECs started offering this as a standard feature as a differentiation against other CLECs.
RTN = Route To Number -- this is the number the call is destine to.
This biggest problem is that we started getting a lot of smaller CLECs that didn't understand the technology well enough and started giving everybody closer access to the PSTN (for example, by not watching the CPN they were sending). The problem was exacerbated when VoIP became a thing and CLECs started allowing anybody access to the PSTN with no restrictions and no regard to their physical location.
These scams are hard to track down. I'd venture to say that 80% of them are running on stolen credit cards, on AWS (or other cloud provider) EC2 instances, connected to some VoIP provider that is billing another stolen credit card. They connect their SIP phones from anywhere to the PBX in the cloud and they start. Labor is cheap in other places in the world and with everything being in the cloud they can be pretty much anywhere. If they get shut down, they just use another stolen credit card and launch another EC2 instance and they are back in business a few minutes later.
The problem is that most of these scam calls are originating from outside the United States. Our laws can't do much outside the US without a lot of legal paperwork -- and in most cases it won't be worth it.
One easy solution is to give consumers access to the BTN or Bill-To phone number. This is the number that is being billed for the call -- essentially pinning down the place where the call is being switched into the PSTN. If you get the BTN, you get the person behind the call -- regardless of what their Caller ID is. Unfortunately, right now, the only way to get access to the BTN is via the SS7 protocol (not available to consumers), or to compel your phone company to give it via a subpoena. Enough abuse from a single BTN -- cut them off until they can clean up their act.
You do realize that you still run all the Adobe software on your local computer -- the only thing that makes it "the cloud" is that it includes a document management and file storage piece. Oh, and I think /some/ settings are shared across installs of the software.
When they switched to this model, they also changed it so you rent the software instead of purchase it. The advantage is that they push updates more frequently, but you are paying a monthly fee for it.
Sure. But the hard things are :
- Ads, and tracking their placement, etc.
- Encrypted Content (most content producers don't want their content streamed unencrypted -- and that causes issues for vanilla browser deployments)
- Streaming (this becomes less trivial if you are looking to utilize existing infrastructure to stream to the browser).
And these are why HTML5 video is still slow to roll out. Once the HTML5 spec had a basic video player, everybody moved onto the next shiny object and left the rest so everybody had to come up with their own solution.
A lot of it has to do with the states dropping their support of higher-ed. In Michigan, as recently as 2000, 80% of the major Universities' operating budget came from the state. In 2015, it was down to 15%. Costs to educate each student (budget / number of students) has been flat, without considering inflation. Funding sources from outside the state have gone up, but not enough to offset the difference. Consequently, tuition used to cost $135/credit hour for in state, and now it costs $375/credit hour.
According to the IRS, Non-profits (501(c)3 organizations) are owned by the community and are for the greater good.
501(c)3's don't have owners or shareholders. They can have a board of directors (and should), and they can have employees. Those employees can be paid like in any other business.
When I helped run our Boy Scout troop when I was much younger, I learned some things about the United Way. In order to become eligible to receive a donation, you have to sign your organization up to volunteer for the United Way. In our case, we had to help run one of their call centers for I think three or four weekends. THEN, after you become eligible, you then have to buy their merch -- things like United Way flags, shirts, etc to promote them. Finally, the donation will usually come at an awards dinner -- which a few people from your org have to attend (and pay for).
They gave us a check for $1,200 for a new trailer. Not counting volunteer time, the troop ended up paying about $1,000 in fees, merch and expenses to participate. Sure, next year it would have been better but to tie up 8 guys for 4 weekends for $200, there were much better ways to get donations or raise money (selling popcorn would usually net us about $8,000 and selling Christmas Trees would net us closer to $10,000).
In Michigan it's not easy to become a dealership (not a franchise -- they can't sell cars).
You first have to join the dealer trade association. Then you have to apply to become a franchise with the state. The Secretary of State decides your initial market area that you are allowed to cover. If you don't belong to the dealer trade association, you immediately are not given a territory, and most likely won't be approved by the State. In order to get your dealership license, you also have to have a setup for service, dollars spent, setup with the SOS for tab registration, etc. Essentially you have to dump a ton of money into the location right away -- in essence a huge barrier to entry for anybody starting up.
This actually started with Wallmart. When Wallmart started carrying electronics, they were demanding to be cheaper than their competitors. They used their strength as a retailer to force the vendors to make custom versions of their products to accomplish this. On printers, they would ship without ink, for computers they would have 30 or 60 day warranty instead of 1 or 2 years. For TVs, they may have had certain featured disabled (for example, DVI ports were there, but you couldn't tune to them).
MOST electronics stores carried the same SKU at the time, and Wallmart/Sams Club had their own. Then other retailers started demanding custom packages -- so they got their own SKUs too. It's not that bad, but you will typically see 3 or 4 different SKUs... a discount retailer (Wallmart), a big box retailer (BBY/Micro Center), and online retailer (Amazon) and sometimes a SKU for premium retailers.
And I think that is the key. They were upfront about it, and allowed you to make an informed decision. The owners of this wrkriot company lied and deceived their employees about what was going on.
If there are rough waters ahead, be open and honest about it. Most people will understand and try and help out. In this case, they lied about what their situation was to most of the employees they brought on and continued lying when they ran out of runway and couldn't pay them anymore.
The crazy thing was they leveraged themselves so much, they had no way to control their spending. Private jet service? Hiring dozens of employees when you are on the edge of the cliff? Just speaks volumes of their willingness to live reality vs. living the dream.
People who build real big data centers are usually too smart to buy Cisco. For the last 10 years, people buy Cisco for the name, not for the product. Cisco does not perform as well, they cost more, and they are harder to use than products from their competitors. They also like to do things their own ways -- which makes compatibility issues a real deal when you try to work with other vendors.
Cisco's deal was to talk to the CEOs, sell them over a nice steak and not consider tech specs. For a SMB or medium sized business, that was easy to do. Large technology firms, like data centers, actually pay attention to what they are buying.
The problem with your example of Amazon is that Amazon invests every penny they earn back into the business. Companies like Delta don't. So when there are bad times, Amazon will be much better poised to do well because they've diversified and built up their business to handle it. All it takes is a generator to malfunction and Delta could be out of business forever (yes, a bit of a stretch, but still).
Delta has a virtual monopoly for a large swath of the nation. You have no choice to fly delta in the midwest and large portions of the south. In recent history, they were never /that/ bad off. In the last market crash, they decided to burn their cash on buying NorthWest instead of modernizing their own systems or investing in their own infrastructure. In the last three years where they've been posting record profits they continued to do cost-cutting measures in all portions of their business, and move money out of the business by paying investors and the execs.
From the reports I've been seeing, it wasn't the ATS that failed, but rather a generator that caught on fire -- and in order to extinguish the fire safely, they had to cut commercial power.
Freak accidents like that happen. But what also happens is that companies that big invest in redundant systems in geo-redundant locations. What happens if a tornado, sharknado or other natural disaster happens and takes out the physical servers? Does Delta just cancel flights for the next month while they rebuild?
If you compare any of the APIs from Google, Microsoft and Amazon, you will clearly see why the different groups are in the place they are.
Google's API set is probably one of the crappiest I've ever seen. It's impossible to do anything with their service unless you use their pre-baked SDK. Sure, it's a REST api, but you can't authenticate against it, because they won't really tell you how -- only why you wouldn't want to do it. They have no docs on how to use their APIs with just CURL.
Microsoft's is better. Their APIs are a pain (mostly because they keep changing), but at least they are pretty well documented and done in a way that you can actually use if you want to. They offer a really rich set of features.... but they do keep changing them on the fly and don't really version stuff like you would expect.
Amazon knows how to API enable their stuff. Their own services and tools use their own published API to do things. They give lots of examples in a bunch of different languages. If you write against it, it will pretty much work forever unless you change your own setup.