Personally, I'd call an simplistic algorithm that gets 70% right brilliant, and one that has a tremendous amount of insight that also gets 70% correct overly complicated and prone to unpredictable failure.
g was good enough for me, but I bought an n anyway and was very pleased with the upgrade. Corners of the house that used to be spotty coverage became rock solid. The "yard range" went from near the house to almost 100' further into the yard. Sure, g was fast and reliable enough and covered what I needed, but n was a clear improvement and actually useful at times.
There's a local shop that hires a mix of experienced developers, kids still in college, and H1Bs... they have horrendous turnover (average tenure 1 year is because the H1Bs stay put. The college kids bolt at first opportunity, and the experienced ones seems to find better things fairly often too...
In this particular case, I'm not very impressed with a 70% prediction rate on a binary decision... you could get similar results by saying "Uphold" every time and ignoring all the data.
What would be more impressive is if the algorithm could predict (with greater accuracy) how theoretical courts would perform with new justices assigned to the bench. Say a seat is opening up and there are several candidates for the position, can the algorithm tell you what the outcome of an upcoming case will be with the various choices of new justices?
The manufacturers think they can do it safely. They even have multinational conferences where they get together and the 2 guys from every company who would rather travel than work sit around and agree with each other that they have put in enough safety checks to protect their customers.
The problem is, most people can't mentally scale risk up to millions of copies. The basic engineer's metric is: "I tried it on my test rig as many ways as I can think of and nothing ever failed." Put this guy in a "world class" test facility with all the best toys money can buy and he'll write you all kinds of analyses "proving" that their accelerated degradation models guarantee a trillion hour MTBF. Problem is: when you put a million imperfect copies of a thing into the real world, with a million different people operating them in thousands of different use cases in hundreds of different environments, the "world class" test facility becomes a myopic little ivory tower by comparison.
One of the answers is "post market surveillance" - but that's expensive, politically unpopular, and logistically difficult to implement, though it is getting cheaper and easier, I don't think it's getting any more politically acceptable. Personally, I feel that the commercial arm of the corporations have corrupted the good in onboard diagnostics, putting up a little "feed my mechanics' and dealers' families" light on your dashboard that comes on for every little problem, but still managing to let you get stranded by the side of the road with little to no warning Why would I ever trust such a system to "phone home" with data about my driving habits?
They've been playing at this since the 1970s. Scan code systems that sell for $50K. "Open" protocols that you have to be a member of the society to get a copy of, membership fee: $25K plus a reason they deem as valid to join. This was last century.
Just be glad that the OBD-III proposals with RFID communication requirements never got passed (or did they?) - with that, the same type of toll readers that are more and more common could as easily query your OBD port and read everything about your present vehicle condition - effectively making possible a "go directly to your mechanic and pay to fix your vehicle or get your license revoked" checkpoint anywhere desired, including across a 6 lane interstate where traffic moves at 80mph - yes, the protocol can query all the vehicles on the road simultaneously as they drive through a checkpoint.
Yeah, that nutso fruity computer company had absolutely no business getting into music players, or phones, did they?
If 1 out of 100 of Google's crazy ideas take off on large scale, they stand to profit overall, and the 99 so called failures can also been called learning experiences, helping that 1/100 to succeed.
The theory, from the investment bankers, is that for every 20 nutso projects, one will be a homerun and return more than 20:1 on the investment. 95% failure rate still = win.
Or, break down and download Notepad++ - it's what notepad should have been in Windows95.
Of course, I once tried to train a Nurse how to use edlin, because it was the only thing her boss provided her to work with on her new desk decoration (1985ish).
My programming work takes me into systems land, maybe only 4 hours a month, but it happens - and being able to handle it myself means getting that work done in 4 hours, instead of spending 4 hours lobbying for the political capital to get the project approved in the IT department and then waiting a week or two (or more) for it to have it implemented, poorly.
The article examined 5 different editors - I don't think users of vi or emacs have enough spare neurons to be able to learn and objectively compare that many alternatives. Not saying the author is "right" to exclude them, just that if he really knows enough to evaluate these other 5, he probably can't work his way around vi or emacs well enough to give them a fair evaluation.
Nano isn't bad, I like jed a little better - not quite as lame, and almost as universally available.
If you're doing serious code composition, then, yes, use a well honed tool for the job that has helpful auto-complete, highlighting, etc. But if you're hacking through twisted network links - X usually isn't available and something lighter weight is a very good to be able to use.
Personally, I only "hack the net" about 1% of the time, so I don't think it's worth using a text based editor as my primary tool, but knowing about a simple, easily accessible, editor can really help out during that 1% time.
What the graphic editors miss is the advantage of being able to run in any environment, including lame ssh shell sessions... I think that's what really keeps vi and emacs going.
Personally, I prefer jed, and yes, it's clunky and feature poor, but it's good enough to get the job done and I'm not devoting 12cc of brain volume to remember how to use it.
Too lazy to RTFA, I take the meaning of the summary this way:
Like a society can be judged by how it treats its elderly, infirm, and more fragile members, a coding project (open source or privately funded) can be judged by how it treats its least well regarded developer.
Are you Nazi Germany, do you show people the door based on the color of their eyes/hair, how tall they are, their GRE scores, or how they perform on some arbitrary admission test before you give a 15 minute in-person interview?
Are you Genghis Khan's Mongolia, do you abuse and then fire anyone who isn't running at the front of the pack? Rank and yank does not generally improve morale.
Are you the European Middle Ages, do you just ignore your weaker team members and let them be consumed by plague rats / drown in their own stinking code while you isolate the shipping product in the ivory tower?
Are you a more modern quasi-socialist society where you educate your weaker team members as best you can and enable them to contribute as they are capable?
There are cases to be made for the advantages and efficiencies of all approaches, but, generally, you need to be a strong development team to carry and build up the weaker team members - if everybody is too focused on product and producing to care about helping their fellow team members to improve, your team is overtaxed (too weak for the job at hand) and probably not able to perform well (provide a reliable living wage for the developers while producing and maintaining the product) in the long term.
But people get off on unenforceable judgements. The real force of the judgement is that if the photos ever become public after the ruling, then the photographer can be found in contempt - which is a whole other golden opportunity for the ex to leak the photos and make additional hell for the guy....
If you study the eastern philosophies, you will find that hot pizza is more important to happiness than whether or not someone else knows what you are doing.
It would be the height of conceit to believe that what you do in your living room is interesting enough for anyone important or in-power to care about. I worked at a company that had "listening bug" phones on every desk in every office - we still talked openly in front of the phones, openly disparaging the leadership, their policies, their personal habits, etc. and, somehow, when the layoffs came around, we, the brazen flaunters of the surveilance state, were not the ones let go. It's not because we were too valuable or otherwise endeared to the leadership, it was because they simply didn't care enough to listen - even though they had the capability.
If you really have something to hide, then hide it, and know that your fancy new television _could_ spy on you. If you live an unremarkable life - as most of us do, nobody will ever bother to activate the bug in your television, or set up an IR laser reflection listening device on your windows, or tap your phone line, or any of the other hundreds of methods that exist - and mostly have existed for centuries - to find out what you are up to. Conspiracy comes down to who you communicate with, and most acts of terror come down to collection and assembly of dangerous materials / devices. You don't need a smart TV to figure this out.
As to embedded 3G antennae, I can barely get a signal out on my cell phone through a metal roof, seems like a tinfoil hat would be quite effective here.
Slashdot Mirror
Personally, I'd call an simplistic algorithm that gets 70% right brilliant, and one that has a tremendous amount of insight that also gets 70% correct overly complicated and prone to unpredictable failure.
g was good enough for me, but I bought an n anyway and was very pleased with the upgrade. Corners of the house that used to be spotty coverage became rock solid. The "yard range" went from near the house to almost 100' further into the yard. Sure, g was fast and reliable enough and covered what I needed, but n was a clear improvement and actually useful at times.
Ships travel in fleets - updating the relative antenna positions in heaving seas, that's where the fun is...
There's a local shop that hires a mix of experienced developers, kids still in college, and H1Bs... they have horrendous turnover (average tenure 1 year is because the H1Bs stay put. The college kids bolt at first opportunity, and the experienced ones seems to find better things fairly often too...
The Clayton Act only applies when someone applies it. If you were wronged by these people, bring suit under the Clayton Act and have at them.
Unfortunately, if you're just a bystander, or the statute of limitations has run out, or you have accepted other settlement in the matter, you can't.
Bear in mind, the model only gets it right 70% of the time, and a red-black roulette spin would get it right nearly 50% of the time.
I was going to say: media preference datapoint: NPR vs FOX, might be the strongest predictor.
In this particular case, I'm not very impressed with a 70% prediction rate on a binary decision... you could get similar results by saying "Uphold" every time and ignoring all the data.
What would be more impressive is if the algorithm could predict (with greater accuracy) how theoretical courts would perform with new justices assigned to the bench. Say a seat is opening up and there are several candidates for the position, can the algorithm tell you what the outcome of an upcoming case will be with the various choices of new justices?
The manufacturers think they can do it safely. They even have multinational conferences where they get together and the 2 guys from every company who would rather travel than work sit around and agree with each other that they have put in enough safety checks to protect their customers.
The problem is, most people can't mentally scale risk up to millions of copies. The basic engineer's metric is: "I tried it on my test rig as many ways as I can think of and nothing ever failed." Put this guy in a "world class" test facility with all the best toys money can buy and he'll write you all kinds of analyses "proving" that their accelerated degradation models guarantee a trillion hour MTBF. Problem is: when you put a million imperfect copies of a thing into the real world, with a million different people operating them in thousands of different use cases in hundreds of different environments, the "world class" test facility becomes a myopic little ivory tower by comparison.
One of the answers is "post market surveillance" - but that's expensive, politically unpopular, and logistically difficult to implement, though it is getting cheaper and easier, I don't think it's getting any more politically acceptable. Personally, I feel that the commercial arm of the corporations have corrupted the good in onboard diagnostics, putting up a little "feed my mechanics' and dealers' families" light on your dashboard that comes on for every little problem, but still managing to let you get stranded by the side of the road with little to no warning Why would I ever trust such a system to "phone home" with data about my driving habits?
They've been playing at this since the 1970s. Scan code systems that sell for $50K. "Open" protocols that you have to be a member of the society to get a copy of, membership fee: $25K plus a reason they deem as valid to join. This was last century.
Just be glad that the OBD-III proposals with RFID communication requirements never got passed (or did they?) - with that, the same type of toll readers that are more and more common could as easily query your OBD port and read everything about your present vehicle condition - effectively making possible a "go directly to your mechanic and pay to fix your vehicle or get your license revoked" checkpoint anywhere desired, including across a 6 lane interstate where traffic moves at 80mph - yes, the protocol can query all the vehicles on the road simultaneously as they drive through a checkpoint.
Yeah, that nutso fruity computer company had absolutely no business getting into music players, or phones, did they?
If 1 out of 100 of Google's crazy ideas take off on large scale, they stand to profit overall, and the 99 so called failures can also been called learning experiences, helping that 1/100 to succeed.
The theory, from the investment bankers, is that for every 20 nutso projects, one will be a homerun and return more than 20:1 on the investment. 95% failure rate still = win.
Big comfort for the dead ratepayers after the 20 year decomissioning project is over. Still better than going to executive perks.
Is there still a nude beach nearby?
Or, break down and download Notepad++ - it's what notepad should have been in Windows95.
Of course, I once tried to train a Nurse how to use edlin, because it was the only thing her boss provided her to work with on her new desk decoration (1985ish).
My programming work takes me into systems land, maybe only 4 hours a month, but it happens - and being able to handle it myself means getting that work done in 4 hours, instead of spending 4 hours lobbying for the political capital to get the project approved in the IT department and then waiting a week or two (or more) for it to have it implemented, poorly.
The article examined 5 different editors - I don't think users of vi or emacs have enough spare neurons to be able to learn and objectively compare that many alternatives. Not saying the author is "right" to exclude them, just that if he really knows enough to evaluate these other 5, he probably can't work his way around vi or emacs well enough to give them a fair evaluation.
Nano isn't bad, I like jed a little better - not quite as lame, and almost as universally available.
If you're doing serious code composition, then, yes, use a well honed tool for the job that has helpful auto-complete, highlighting, etc. But if you're hacking through twisted network links - X usually isn't available and something lighter weight is a very good to be able to use.
Personally, I only "hack the net" about 1% of the time, so I don't think it's worth using a text based editor as my primary tool, but knowing about a simple, easily accessible, editor can really help out during that 1% time.
What the graphic editors miss is the advantage of being able to run in any environment, including lame ssh shell sessions... I think that's what really keeps vi and emacs going.
Personally, I prefer jed, and yes, it's clunky and feature poor, but it's good enough to get the job done and I'm not devoting 12cc of brain volume to remember how to use it.
Dr Chun,
What area of AI development is currently making the most progress? In other words, where are the next big advances most likely to come from?
Too lazy to RTFA, I take the meaning of the summary this way:
Like a society can be judged by how it treats its elderly, infirm, and more fragile members, a coding project (open source or privately funded) can be judged by how it treats its least well regarded developer.
Are you Nazi Germany, do you show people the door based on the color of their eyes/hair, how tall they are, their GRE scores, or how they perform on some arbitrary admission test before you give a 15 minute in-person interview?
Are you Genghis Khan's Mongolia, do you abuse and then fire anyone who isn't running at the front of the pack? Rank and yank does not generally improve morale.
Are you the European Middle Ages, do you just ignore your weaker team members and let them be consumed by plague rats / drown in their own stinking code while you isolate the shipping product in the ivory tower?
Are you a more modern quasi-socialist society where you educate your weaker team members as best you can and enable them to contribute as they are capable?
There are cases to be made for the advantages and efficiencies of all approaches, but, generally, you need to be a strong development team to carry and build up the weaker team members - if everybody is too focused on product and producing to care about helping their fellow team members to improve, your team is overtaxed (too weak for the job at hand) and probably not able to perform well (provide a reliable living wage for the developers while producing and maintaining the product) in the long term.
But people get off on unenforceable judgements. The real force of the judgement is that if the photos ever become public after the ruling, then the photographer can be found in contempt - which is a whole other golden opportunity for the ex to leak the photos and make additional hell for the guy....
We have 3 mobile phones in our house, and only 2 adults, though, if we didn't also have 2 kids, we might not feel the need for a "backup" phone.
If you study the eastern philosophies, you will find that hot pizza is more important to happiness than whether or not someone else knows what you are doing.
It would be the height of conceit to believe that what you do in your living room is interesting enough for anyone important or in-power to care about. I worked at a company that had "listening bug" phones on every desk in every office - we still talked openly in front of the phones, openly disparaging the leadership, their policies, their personal habits, etc. and, somehow, when the layoffs came around, we, the brazen flaunters of the surveilance state, were not the ones let go. It's not because we were too valuable or otherwise endeared to the leadership, it was because they simply didn't care enough to listen - even though they had the capability.
If you really have something to hide, then hide it, and know that your fancy new television _could_ spy on you. If you live an unremarkable life - as most of us do, nobody will ever bother to activate the bug in your television, or set up an IR laser reflection listening device on your windows, or tap your phone line, or any of the other hundreds of methods that exist - and mostly have existed for centuries - to find out what you are up to. Conspiracy comes down to who you communicate with, and most acts of terror come down to collection and assembly of dangerous materials / devices. You don't need a smart TV to figure this out.
As to embedded 3G antennae, I can barely get a signal out on my cell phone through a metal roof, seems like a tinfoil hat would be quite effective here.