Let's say I pay $x / month for this service - what's to stop Jo Schmoe next door using my "frequency" for nothing. Experience with 802.11b, or whatever, is slowly teaching us that wireless is not as secure as fibre / cable.
There is no "your" frequency. Everyone shares a wide band of spectrum. It works like Ethernet, except it can be faster if the frequency band is wider. A lot of data is transmitted in short pulses (short in time) and wide in frequencies. Since each transmission is very short (nano-seconds) there is little chance of interference.
Security is an independent issue.
For infrastructure you can imagine a network that was formed by our computers talking to each other and forwarding packets (google "mesh grids"). So, in theory at least, no infractructue is needed, other than our own computers. Just think of FIDOnet, except at Fast Ethernet speeds...
Do you really think there is a comparison between the two? Do you want the government responsible for creating and setting up these kind of high-tech networks?
The roads are a shared resource like the spectrum. The goverment doesn't have to build the spectrum, it just should not allocate it to private companies, if the resource can be shared.
The goverment should sponsor/define open standards that anyone who wants to use the spectrum must follow, and then anyone can build the equipment.
Just like anyone can build and drive a truck, as long as they conform to the federal guidelines..
How the fuck is this new spectrum supposed to get put to use without private investment?
The point is that the spectrum does not have to belong to a private company to be utilized. For example, interstate highways belong to the public, but everyone can use them.
Read about Ultra Wide Band to learn how bandwidth can be shared, just like the highways.
In my opinion bandwidth caps are ok as long as they're agreed upon when you signup for service (i.e. you ask for 500kbps down, and thats what you get). Per byte charges are historically disfavored for home users even though businesses like the idea.
I agree with you that caps are fine, as long as you understand when you sign up. Per byte charges will be the death of streaming video service (as though it's not dead already).
I can just see customers paying for a movie online and then finding out that they used up their monthly bandwidth and must pay extra for web surfing and email!
Broadband has caught on to a large degree, but not as large as anyone thought it would, and certainly not enough for the huge streaming video boom that was supposed to happen.
Streaming movies is just a stupid idea. It's a poor use of servers and bandwith. NetFlix, for example,
makes a much better use of bandwith - they snail-mail DVDs...
I have seen a trend when it comes to layoffs that is echoed in the experience I had -- for some oddball reason, it seems the management likes to trim the knowledge base at the wrong points. It stands to reason that, when letting go a very knowledgeable person, someone else must be trained up to fill the shoes of that person.
There is a very good discussion of this problem in the book PeopleWare. Basically the authors agree with you and they provide some interesting cost analysis.
Here are two radical possibilities for a different career:
Become a patent lawyer. We really need laweyers who understand software and there is great need for them.
Become a high school math and science teacher. Pay not as good, but much improved quality of life, you'll be doing the society a great service, as we all need people who understand science and math.
I agree that there is a need for experienced technical managers to lead. But I also believe that there is a need for really good, seasoned programmers/DBAs/sys admins/etc to work with the "youngsters", lead by example and pass along some of the accumulated wisdom.
I agree that there is a need for technical mentors (I'm available:-)). This reminds me of an old riddle:
Q: What's the difference between a leader and a manager?
Collaborative filtering sucks. Your idea is valid, just that's not the right recommendation technology to use.
I agree with you that the automatic filtering sucks. But what I had in mind is people in some smaller group do the filtering.
So, for example I read the Jazz Guitar news group. Occasionally someone will post a recomendation for a CD for an artist I haven't heard yet. Others may add their comments. Based on those I can judge pretty well whether I'd be interested.
Similarly I post my own opinions that others can use.
I'm more likely to follow a jazz recomendation of someone from the jazz guitar news group, rather than from Slashdot. The members of the newsgroup are self-selected to be interested in the same thing.
Come to think of it, Usenet is a sort of distributed P2P system...hmmm...
The problem with ratings systems is that they get corrupted by fans of the band. If band members have very large families (perhaps they're all rednecks and have family trees that don't branch:-), their family members alone might be enough to get them unfairly pushed to the front of the list. Then there's the 'bot problem. How do we tell that a rating is coming from a human being, and not from a 'bot that is repeatedly disconnecting and then reconnecting in order to get a different IP address at the ISP?
But what about a closed group of people who share ratings. You'd have to join, get an ID etc, etc (sort of like/.). Since the groups shares taste, it's unlikely that ratings will get pumped for other reasons. "Collaboration" means working together...
One of the functions that big media companies serve is to act as the consumers' agents in discovering good talented artists for them. In theory it's more efficient for consumers to pay the media companies a fee so that consumers do not need to spend the time, energy, money required to discover talent themselves
But what about some colaborative filtering , P2P system? Imagine hooking it up with Gnutella so that you can not only download music, but add in your ratings.
Then groups of people with simlar taste find the good stuff they all like.
This sort of happens with your friends, and via USENET groups. I'm much more likely to listen to a recomendation of a fellow news group reader, than a commercial. Of course the stuff I listen to(jazz guitar for example) is hardly ever heard on the radio or MTV....
Re:Beware of overusing patterns.
on
Design Patterns
·
· Score: 2
Which brings me to a question. I have always viewed design patterns to be something that are used during the planning phase of a project.
Are you a coder or a planner?
In the "real world" (tm) there is no clean separation between the "design" and the "planning" phase - because things change.
In real projects things happen iteratively - you design and code a little, see how things are going, then amend the plan, and go back to coding. In between you ask the users to see if they thought of new requirements.
[...]But the economics don't, and so-called community ISPs are going to kill the goose that laid the golden egg, if they're not careful.
But if you can keep the heavy traffic on the CLAN, and limit the traffic on the DSL, things may work much better. What about doing heavy caching within the CLAN and encouraging P2P amoung the CLAN members?
More nitpicking - Struts is a framework that uses JSPs (actually TLDs). JPetstore connects to a database, so it has to use JDBC. The only thing that's missing is EJBs (maybe there is a good reason for this;-))
Re:Suit and Tie do not make the programmer.
on
Suit Up Or Ship Out?
·
· Score: 2
Do not underestimate the power of a good, motivated team. And also recognize that you can often do good code when you are fresh, and then do good debugging of that code as you get more tired. This helps maintain the quality of the code.
Oh, I don't. I think a motivated/gelled team can perform near miracles, even during a 40 hour week.
I usually do my best debugging in a shower. When you have a nasty problem you can't figure out, it's best to step away from it, and more often than not the solution will pop into your head when you are doing something else.
Life outside work is often a great source for new ideas for the actual problems you are trying to solve at work. You should try not to loose the sight of the forrest while down in the bushes...
Finally, there are some things more important than work - like taking your daughter trick-or-treating on Halloween...:-)
Re:Suit and Tie do not make the programmer.
on
Suit Up Or Ship Out?
·
· Score: 2
I don't know what you guys were doing, but I have done lots of quality systems level work (like building custom OLTP monitors, device drivers, com protocols, etc) on very long hours. I think it depends very much on the environment and on attitudes. If you get on a team that is really moving, you can work 18 hours a day, enjoy it, and do it for months.
Maybe if you are a 20 year old with no life outside work you could do that. However, I've built similar system working 40/50 hour weeks. Some of these have been running in production for over 15 years.
Read about Heroic
Programming. I feel sorry for the guys who had to
maintain your code...
3. The selection is very limited or non-existent (see emusic, musicnet, liquidaudio, etc, etc.)
4. They don't bother to show you what songs they actually have unless you sign up (I'm just using emusic and musicnet as an
example)
I use www.emusic.com. Emusic lets you sign up for free and download 50 files. You can then search and see what they have.
For me Emusic has been a treasure trove of great jazz recordings. I've downloaded several hundered files by now of classic jazz guitar and bunch of other great jazz stuff.
So, for me at least the selection is great. I don't want the crap that RIAA is trying to sell.
We should all get more serious about building wireless grid networks. With wireless cards and routers you could build a network that could cover the entire world.
Imagine that in a small community (eg. a college) you could P2P over the air with UWB, without the need to involve any other company network.
Transmission should be encrypted and the bandwidth is virtually unlimited...
Who needs the cable companies, let's turn our computer into routers...
Perhaps because successive generations are iterated at a much faster rate? Compare seconds to centuries: do you see the difference?
But evolution is massively parallel. So even if an individual organism take a while to compute the next step, there are billions and billions organisms doing it...
You may not be aware of them, but there are several techniques for solving a problem on a computer when you have no idea how to solve the problem yourself. Neural networks and genetic algorithms for example.
OK. I'm aware of GA and neural networks. And to a point you are correct. But can you name two significant problems that were solved by GA, and nothing else?
The other point (which I read about in a book by Stanilaw Lem) is that evolution can be looked at as a massively parallel GA-style computation. Look how long it took to create conciousness. Why do you think that a GA written by humans will do any better (i.e. take 30 years, instead of 4 billion).
Some SF books that explore this idea
on
Downloading The Mind
·
· Score: 4, Interesting
There are couple of SF books that explore this idea. I think they are worth checking out:
Software - Rudy Rucker. Exactly on this topic - transferring human minds to a computer (too bad the process destroys the brain).
Golem XIV - Stanislaw Lem. A supercomputer becomes intelligent, but the intelligence is completely not like human mind, but something quite different. After all, human biology influences how the mind works. The book is a "transcription" of lectures by the computer on the nature of inteligence.
Slashdot Mirror
There is no "your" frequency. Everyone shares a wide band of spectrum. It works like Ethernet, except it can be faster if the frequency band is wider. A lot of data is transmitted in short pulses (short in time) and wide in frequencies. Since each transmission is very short (nano-seconds) there is little chance of interference.
Security is an independent issue.
For infrastructure you can imagine a network that was formed by our computers talking to each other and forwarding packets (google "mesh grids"). So, in theory at least, no infractructue is needed, other than our own computers. Just think of FIDOnet, except at Fast Ethernet speeds...
The roads are a shared resource like the spectrum. The goverment doesn't have to build the spectrum, it just should not allocate it to private companies, if the resource can be shared.
The goverment should sponsor/define open standards that anyone who wants to use the spectrum must follow, and then anyone can build the equipment. Just like anyone can build and drive a truck, as long as they conform to the federal guidelines..
The point is that the spectrum does not have to belong to a private company to be utilized. For example, interstate highways belong to the public, but everyone can use them.
Read about Ultra Wide Band to learn how bandwidth can be shared, just like the highways.
The real funny part is that "Steamboat Willie" was a parody of then current Buster Keaton movie.
I agree with you that caps are fine, as long as you understand when you sign up. Per byte charges will be the death of streaming video service (as though it's not dead already).
I can just see customers paying for a movie online and then finding out that they used up their monthly bandwidth and must pay extra for web surfing and email!
Streaming movies is just a stupid idea. It's a poor use of servers and bandwith. NetFlix, for example, makes a much better use of bandwith - they snail-mail DVDs...
There is a very good discussion of this problem in the book PeopleWare. Basically the authors agree with you and they provide some interesting cost analysis.
I agree that there is a need for technical mentors (I'm available :-)). This reminds me of an old riddle:
Q: What's the difference between a leader and a manager?
A: A leader has followers.
I agree with you that the automatic filtering sucks. But what I had in mind is people in some smaller group do the filtering.
So, for example I read the Jazz Guitar news group. Occasionally someone will post a recomendation for a CD for an artist I haven't heard yet. Others may add their comments. Based on those I can judge pretty well whether I'd be interested.
Similarly I post my own opinions that others can use.
I'm more likely to follow a jazz recomendation of someone from the jazz guitar news group, rather than from Slashdot. The members of the newsgroup are self-selected to be interested in the same thing.
Come to think of it, Usenet is a sort of distributed P2P system...hmmm...
But what about a closed group of people who share ratings. You'd have to join, get an ID etc, etc (sort of like /.). Since the groups shares taste, it's unlikely that ratings will get pumped for other reasons. "Collaboration" means working together...
But what about some colaborative filtering , P2P system? Imagine hooking it up with Gnutella so that you can not only download music, but add in your ratings.
Then groups of people with simlar taste find the good stuff they all like.
This sort of happens with your friends, and via USENET groups. I'm much more likely to listen to a recomendation of a fellow news group reader, than a commercial. Of course the stuff I listen to(jazz guitar for example) is hardly ever heard on the radio or MTV....
Are you a coder or a planner?
In the "real world" (tm) there is no clean separation between the "design" and the "planning" phase - because things change.
In real projects things happen iteratively - you design and code a little, see how things are going, then amend the plan, and go back to coding. In between you ask the users to see if they thought of new requirements.
Google "Extreme Programming".
But if you can keep the heavy traffic on the CLAN, and limit the traffic on the DSL, things may work much better. What about doing heavy caching within the CLAN and encouraging P2P amoung the CLAN members?
But then Red Hat will have to sign an NDA and will not be allowed to release any source code. So how does this help us?
Oh, I don't. I think a motivated/gelled team can perform near miracles, even during a 40 hour week.
I usually do my best debugging in a shower. When you have a nasty problem you can't figure out, it's best to step away from it, and more often than not the solution will pop into your head when you are doing something else.
Life outside work is often a great source for new ideas for the actual problems you are trying to solve at work. You should try not to loose the sight of the forrest while down in the bushes...
Finally, there are some things more important than work - like taking your daughter trick-or-treating on Halloween... :-)
Maybe if you are a 20 year old with no life outside work you could do that. However, I've built similar system working 40/50 hour weeks. Some of these have been running in production for over 15 years.
Read about Heroic Programming. I feel sorry for the guys who had to maintain your code...
4. They don't bother to show you what songs they actually have unless you sign up (I'm just using emusic and musicnet as an example)
I use www.emusic.com. Emusic lets you sign up for free and download 50 files. You can then search and see what they have.
For me Emusic has been a treasure trove of great jazz recordings. I've downloaded several hundered files by now of classic jazz guitar and bunch of other great jazz stuff.
So, for me at least the selection is great. I don't want the crap that RIAA is trying to sell.
HIH
Imagine that in a small community (eg. a college) you could P2P over the air with UWB, without the need to involve any other company network.
Transmission should be encrypted and the bandwidth is virtually unlimited...
Who needs the cable companies, let's turn our computer into routers...
I suppose the writer may have meant all the "Easter Eggs" that are found in various software packages.
But evolution is massively parallel. So even if an individual organism take a while to compute the next step, there are billions and billions organisms doing it...
OK. I'm aware of GA and neural networks. And to a point you are correct. But can you name two significant problems that were solved by GA, and nothing else?
The other point (which I read about in a book by Stanilaw Lem) is that evolution can be looked at as a massively parallel GA-style computation. Look how long it took to create conciousness. Why do you think that a GA written by humans will do any better (i.e. take 30 years, instead of 4 billion).