Options to buy General Motors stock at today's price don't have any motivating value to the line worker who is unlikely to be able to raise the stock price of a company
It's an interesting question how the affect of options tails off with size. I suspect it is not linear though. By the time I left Yahoo it had 2000 people. You'd think that would be well past the point where ordinary employees thought they could affect the fortunes of the company, but it wasn't. We had the most gung-ho receptionists you can possibly imagine.
As you might guess from the fact that I estimated YHOO to be worth $12 when it was trading around $200, not a lot. I was well aware that the Bubble was a bubble at the time. Friends still imitate me saying "for God's sake, sell."
A lot of businesess are based on having stuff sold by resellers, which amounts to outsourcing your sales force. That's the kind you'd want to try the experiment in.
I didn't say that all essays in high school are written in that format. I just gave that example to take people's minds back to high school.
As for political vs. social history, the rest of the paragraph does implicitly explain why I prefer the latter. It's social history that is the "data." Political history is just he-said-she-said.
You're partly right about the Normans, I admit. That is kind of funny. On the other hand, that is the kind of political history that is practically social history, because it is about underlying social changes, and not just some general or pope who happened to be powerful at a certain time.
I *am* using the same approach with Arc as I do with essays. I talk over ideas with a small group of friends-- in fact, much the same group of friends.
I don't worry about other languages. Did all the ecosystem that grew up around Cobol make any difference?
For "other languages" to be a threat to Lisp, they'd have to become Lisp. I mean this quite literally. What use is a programming language without macros? And how could a language have macros without turning into Lisp? You'd have to invent the programming language equivalent of late Ptolemaic astronomy to do that.
There is a big connection to the Internet, if not to hackers per se. Roughly that blogs will lead to a renaissance in the essay.
I went back and forth about mentioning blogs by name, because not everything people publish on their own site is really "blogging" in the strict sense. "Blog" implies "log", which implies a time quantum of less than a day. Whereas it takes me weeks to write an essay.
Here's a footnote I commented out that made the connection to the Web more explicit:
When I first heard about blogs, I imagined they would be a complete waste of time. Blogging sounded like a long-play version of netnews. But I was mistaken; people care more about something that stays on their site, and the Web supplies a filter that's missing in newsgroups. The best writing online is not only better than netnews, but better than most print media.
I did talk about what I was up to with Arc at the Lisp conference last summer. I just didn't put that talk online. What I've been doing most recently is working on growing the language from the smallest set of axioms I can, and continuing all the way through stuff like data types and I/O instead of stopping where McCarthy did.
The main reason I don't talk much about Arc's status is because I don't want to feel like I have some kind of deadline I have to meet. The world has waited 45 years for a really good Lisp implementation. It's not going to make any difference if we have to wait 2 or even 10 more.
Re:Paul Graham is always right!
on
The Age of the Essay
·
· Score: 3, Interesting
Well, you have something here, but not what you think. If I'm not sure of something, I don't publish it online.
I spent about four days just researching the first English courses, and as I sit here writing this there is a nose-high stack of histories of various colleges on the desk next to me. So if I sound confident when writing about the topic, perhaps that's why.
If you disagree with something I said, you'd be more convincing if you could provide a counterexample.
if only an expert can write usable code -- not great, but merely usable -- the language is junk
Is that really true? In many fields there are tools that only experts can use properly. Why should it be different in programming? I think Michael Vanier makes a good point about LFSPs.
It's a bit misleading to say that I called Java uncool. I said in Revenge of the Nerds that hackers think Perl and Python and Ruby are cooler than Java, but I'm not claiming that that's how one should decide what language to use.
I dislike Java because (a) it lacks abstractions I need, (b) it is horribly verbose and bureaucratic, and (c) it is another Frankenstein in the C++ tradition of languages created by bolting features from dynamic languages onto a C substrate.
I'm often accused of being a language bigot because I violate the taboo against comparing one language to another. But if you think about it, anyone who works on language design *has* to do that, taboo or no. If you can't compare alternatives, how do you decide what direction to go in?
I've never been sure myself whether I was liberal or conservative. I think some things I wouldn't dare say out loud in front of a group of liberals, and others I wouldn't dare say out loud in from of a group of conservatives. It's a tossup which category of thoughts is bigger.
There's a footnote about this in "What You Can't Say." If you went back to visit, say, Victorian England, your opinions would probably shock Whigs and Tories about equally. If your goal is to be close to the truth, then you are going to seem like an alien to the people of your own time. It's like projecting a point onto a line segment that is very far away. Where you end up on it is almost random.
I'd guess about 30% of the text in the book is new. The essays that are already on the web have been rewritten too-- some quite extensively, some just tightened up a bit.
I did do some research, and it contradicts your anecdotal evidence.
If the Council of Europe has their way, they'll be banning a wide swath of ideas as "hate speech." This ban would of course apply in Holland too. But it is not simply something they're getting stuck with against their will. The Dutch government seems to be out in front on this one.
I think individual random words in the subject lines are intended to spoof signature-based filters, not Bayesian filters. Presumably signature-based filters like Brightmail eventually realized that they should discard strings like xhh97890fsvcs at the end of subj lines before calculating the checksum. A real word is just as random and harder to recognize and discard.
I'm working on it right now. Drop by ILC on Oct 14 and you can see for yourself.
http://www.international-lisp-conference.org/
As for preventing people from creating new dialects of Lisp, as far as I know I've done nothing to slow down the guys working on Perl, Python, and Ruby;-)
It's amusing that the posting purporting to the conference carried as a payload (in fact mostly consisted of) a mini-editorial about port-25 blocking. Seems like Slashdot needs to do some filtering too.
No one who read the conference announcement should be surprised that a lot of the talks were about filtering. We said four times in the first three paragraphs that the conference was about spam filtering. There are of course other solutions to spam, and I'm all in favor of them, if only because they make filtering easier too. (For example, the fact that many spammers feel constrained by various laws to include even fake unsubscribe links is a great help to filters.)
A good spam solution will have to work even if the spammers know how it works. I believe that Bayesian filtering, which is what a lot of the speakers at the conference will be talking about, is such a solution. Spammers can't outweight the incriminating words they need to use in their sales pitches with innocent words, because the very innocent words (names of friends, terms used in one's work, etc.) are unique to each user.
Even when they do get paid a flat rate by outsiders, they can't get paid a lot more than the spam is generating in revenue. If we cut the revenue to be made from spam by a factor of 20, then ultimately (in fact, quite soon) customers are only going to be willing to pay spammers 1/20th of what they would now.
Actually, filters may turn out to be the ultimate block. If we can filter out even 95% of spam, which is pretty easy, we decrease the response rate for spam by a factor of 20. Spammers send spam because it makes money. How many would still do it for 1/20th as much as they make now? Not many, I suspect.
When you attack response rates, you attack the thing spammers care about most. They would much rather get their account cancelled (they have a hot-swappable backup account ready and waiting) than have their response rates cut.
Cut response rates enough, and you effectively introduce a block as far upstream as you can get, within the spammer's head, as he realizes "this doesn't make money any more."
Slashdot Mirror
Options to buy General Motors stock at today's price don't have any motivating value to the line worker who is unlikely to be able to raise the stock price of a company
It's an interesting question how the affect of options tails off with size. I suspect it is not linear though. By the time I left Yahoo it had 2000 people. You'd think that would be well past the point where ordinary employees thought they could affect the fortunes of the company, but it wasn't. We had the most gung-ho receptionists you can possibly imagine.
I don't know how much money he lost
As you might guess from the fact that I estimated YHOO to be worth $12 when it was trading around $200, not a lot. I was well aware that the Bubble was a bubble at the time. Friends still imitate me saying "for God's sake, sell."
A lot of businesess are based on having stuff sold by resellers, which amounts to outsourcing your sales force. That's the kind you'd want to try the experiment in.
I didn't say that all essays in high school are written in that format. I just gave that example to take people's minds back to high school.
As for political vs. social history, the rest of the paragraph does implicitly explain why I prefer the latter. It's social history that is the "data." Political history is just he-said-she-said.
You're partly right about the Normans, I admit. That is kind of funny. On the other hand, that is the kind of political history that is practically social history, because it is about underlying social changes, and not just some general or pope who happened to be powerful at a certain time.
I *am* using the same approach with Arc as I do with essays. I talk over ideas with a small group of friends-- in fact, much the same group of friends.
I don't worry about other languages. Did all the ecosystem that grew up around Cobol make any difference?
For "other languages" to be a threat to Lisp, they'd have to become Lisp. I mean this quite literally. What use is a programming language without macros? And how could a language have macros without turning into Lisp? You'd have to invent the programming language equivalent of late Ptolemaic astronomy to do that.
I went back and forth about mentioning blogs by name, because not everything people publish on their own site is really "blogging" in the strict sense. "Blog" implies "log", which implies a time quantum of less than a day. Whereas it takes me weeks to write an essay.
Here's a footnote I commented out that made the connection to the Web more explicit:
A language doesn't have to be Arc to be the new Lisp. There are a bunch of new Lisp dialects being developed at the moment, most notably Goo.
I did talk about what I was up to with Arc at the Lisp conference last summer. I just didn't put that talk online. What I've been doing most recently is working on growing the language from the smallest set of axioms I can, and continuing all the way through stuff like data types and I/O instead of stopping where McCarthy did.
The main reason I don't talk much about Arc's status is because I don't want to feel like I have some kind of deadline I have to meet. The world has waited 45 years for a really good Lisp implementation. It's not going to make any difference if we have to wait 2 or even 10 more.
Well, you have something here, but not what you think. If I'm not sure of something, I don't publish it online.
I spent about four days just researching the first English courses, and as I sit here writing this there is a nose-high stack of histories of various colleges on the desk next to me. So if I sound confident when writing about the topic, perhaps that's why.
If you disagree with something I said, you'd be more convincing if you could provide a counterexample.
Is that really true? In many fields there are tools that only experts can use properly. Why should it be different in programming? I think Michael Vanier makes a good point about LFSPs.
It's a bit misleading to say that I called Java uncool. I said in Revenge of the Nerds that hackers think Perl and Python and Ruby are cooler than Java, but I'm not claiming that that's how one should decide what language to use.
I dislike Java because (a) it lacks abstractions I need, (b) it is horribly verbose and bureaucratic, and (c) it is another Frankenstein in the C++ tradition of languages created by bolting features from dynamic languages onto a C substrate.
I'm often accused of being a language bigot because I violate the taboo against comparing one language to another. But if you think about it, anyone who works on language design *has* to do that, taboo or no. If you can't compare alternatives, how do you decide what direction to go in?
Actually I advocate Python because I consider it a gateway drug.
Ok, deal.
I've never been sure myself whether I was liberal or conservative. I think some things I wouldn't dare say out loud in front of a group of liberals, and others I wouldn't dare say out loud in from of a group of conservatives. It's a tossup which category of thoughts is bigger.
There's a footnote about this in "What You Can't Say." If you went back to visit, say, Victorian England, your opinions would probably shock Whigs and Tories about equally. If your goal is to be close to the truth, then you are going to seem like an alien to the people of your own time. It's like projecting a point onto a line segment that is very far away. Where you end up on it is almost random.
I'd guess about 30% of the text in the book is new. The essays that are already on the web have been rewritten too-- some quite extensively, some just tightened up a bit.
If the Council of Europe has their way, they'll be banning a wide swath of ideas as "hate speech." This ban would of course apply in Holland too. But it is not simply something they're getting stuck with against their will. The Dutch government seems to be out in front on this one.
I think individual random words in the subject lines are intended to spoof signature-based filters, not Bayesian filters. Presumably signature-based filters like Brightmail eventually realized that they should discard strings like xhh97890fsvcs at the end of subj lines before calculating the checksum. A real word is just as random and harder to recognize and discard.
I'm working on it right now. Drop by ILC on Oct 14 and you can see for yourself.
;-)
http://www.international-lisp-conference.org/
As for preventing people from creating new dialects of Lisp, as far as I know I've done nothing to slow down the guys working on Perl, Python, and Ruby
Text is unreadable if there are more than about 65 characters in a line. Try it yourself and see.
Wouldn't work. The algorithm only cares about the most statistically significant 15 words. TEENS easily beat yams.
It's amusing that the posting purporting to the conference carried as a payload (in fact mostly consisted of) a mini-editorial about port-25 blocking. Seems like Slashdot needs to do some filtering too.
No one who read the conference announcement should be surprised that a lot of the talks were about filtering. We said four times in the first three paragraphs that the conference was about spam filtering. There are of course other solutions to spam, and I'm all in favor of them, if only because they make filtering easier too. (For example, the fact that many spammers feel constrained by various laws to include even fake unsubscribe links is a great help to filters.)
A good spam solution will have to work even if the spammers know how it works. I believe that Bayesian filtering, which is what a lot of the speakers at the conference will be talking about, is such a solution. Spammers can't outweight the incriminating words they need to use in their sales pitches with innocent words, because the very innocent words (names of friends, terms used in one's work, etc.) are unique to each user.
Even when they do get paid a flat rate by outsiders, they can't get paid a lot more than the spam is generating in revenue. If we cut the revenue to be made from spam by a factor of 20, then ultimately (in fact, quite soon) customers are only going to be willing to pay spammers 1/20th of what they would now.
I don't think they could spew out 20x as much spam. They already operate at capacity.
Actually, filters may turn out to be the ultimate block. If we can filter out even 95% of spam, which is pretty easy, we decrease the response rate for spam by a factor of 20. Spammers send spam because it makes money. How many would still do it for 1/20th as much as they make now? Not many, I suspect.
When you attack response rates, you attack the thing spammers care about most. They would much rather get their account cancelled (they have a hot-swappable backup account ready and waiting) than have their response rates cut.
Cut response rates enough, and you effectively introduce a block as far upstream as you can get, within the spammer's head, as he realizes "this doesn't make money any more."