I've been hacking about with an interpreter for a little language I'm playing with. The end goals (as ever) as suitably vast, but for the moment I'm writing the dumbest interpreter possible. Here's a little snippet which actually runs:
var if = { #test x y# (ifte test x y) }
var cc = { return return }
var while = { #test body#
var k = cc
if (test) {
body return
k k
}
}
while { continue 1 } { #break# print }
Here are the rules:
And the explanation:
if is a simple wrapper around ifte which passed its arguments to it and calls the result. So you pass a value and two callables and one of the callables is run.
cc returns its own return continuation. In Scheme it would look like this:
(lambda (cc) (call/cc (lambda (k) (k k))))
Let's build the while loop up from parts. First, consider a function which is an infinite loop:
var loop = {
var k = cc
k k
}
k is the continuation of cc, so if we call it upon itself, cc returns again and returns the same value: k
Now we all a test callable. Each time round the loop we call the test to see if we should continue:
var loop-while = { #test#
var k = cc
if (test) {
k k
}
}
Then it's just a case of adding a body argument and calling it every time. The body gets a single argument - the return continuation of the while function. If the body calls this it acts the same as a break in a C-like language.
My frequency of posting has dropped to the point where one would be forgiven for wondering if I was still alive. Well, rest assured that I'm not typing this from beyond the grave, it's just that Google is keeping me busy 
.
None the less, I was motivated to post this:
“consumers concerned about the scam should avoid PIN-based retail transactions, and chose instead to make signature-based, credit-card-style transactions when making purchases with debit or check cards at stores.”
(from http://www.msnbc.msn.com/id/11731365/)
Because, chip and pin is so much safer, right? No one could have predicted that putting people's pin numbers everywhere would make it easy to steal them.
Firstly, anything which upsets so many people clear has to be worth looking at, right. Well, it would actually appear not - they're all pretty boring.
(p.s. from some of the descriptions of the specific cartoons on the news I'm not sure if they are actually on that page. Has anyone actually seen these cartoons?)
Micropayments, fungible (cash) or otherwise (hashcash), have been suggested sa the solution to spam for a long time. Clearly it's a very tough deployment problem (you need lots of people to suddenly start using it), but it would appear that Goodmail have persuaded AOL and Yahoo to sign up (thus might have solved the deployment problem).
The hope is that, by increasing the cost of sending email, one can make spam uneconomic. But it really doesn't appear that Goodmail is even trying to do that; unpaid email is treated the same as it always was. If the price is too high for spammers (and the quoted 1/4 cent probably is) then they won't pay and spam will be exactly the same as ever.
As an email sender, the only reason I would wish to pay this is because I have good reason for sending the email (ecommerce confirmations etc) and I don't want the hassel of dealing with customers who's spam filters eat my emails. That's pretty weak. These people have had to deal with it for a long time and rarely can customers not dig through their filters to find an email that they're expecting.
And since Goodmail are getting paid for each email, their motivation for not accepting emails from spammy senders isn't perfectly aligned with the interests of their customers. Clearly they don't want to squander any trust - but there's a strong temptation to see how far you can push customers for that extra bit of income.
And is spam really a problem for anyone these days? I get about 200/day and > 95% fall into the spam trap with very few false positives (that's with Gmail). The spams which do get through are random assortments of words usually.
So I predict that Goodmail will get few customers and make very little difference to anyone.
(and clueless quote from that article: "I still gets e-mails from lists I signed up for three years ago, but I haven't responded to a single one." - then unsubscribe you moron!
So Google have launched in China and are censoring results. This has made a lot of people very unhappy (we had protestors outside today voicing that unhappyness)
Just to be clear, Google doesn't have an option of running an uncensored version in China - it's censored or nothing. It's only a company and, as such, has to follow the laws of the countries it operates in. But Google could have taken a stand and refused to join in. I guess that's what all those people wanted us to do. But here's the thing...
I really don't think it matters
Years ago there was a great hope that the introduction of the Internet into China would open the eyes of all the people. Once they knew about the opression under which they toiled, how could they stand it? Well, they did. Turns out that people will put it with it. The Great Firewall of China isn't perfect; it's main function is just to remind people that Big Brother is watching. These people know the oppression under which they toil and accept it. We believed too much in the power of the Internet and we were wrong.
The driving force for change in China is capitalism, not free expression. All the recent improvements in the lot of the newly rich chinese can be traced to good old fashioned material want. And change in the future is going to come when the bulk of the, still very poor, people start to wake up to this fact. It's happening already and it's only going to get better.
It might have been nice to make a stand for freedom in this case - but pointless. I really hope Google would never turn someone in like Yahoo did, that's just plain bad, but so far it's not worth the fuss.
The UK's answer to the EFF is now up and running and accepting donations. I was at the 'birth' of it in Hammersmith last year when the pledge was setup to fund it.
Well, they've reached 1000 members and it's time to pay up if you are one of them . I don't really want to use PayPal so they'll have to wait until I'm back in the country and I'll mail them a cheque (but I'll be back in a few weeks time; I think they can manage till then).
Things one should read:
Things one should listen to:
(both are BBC and so both need RealPlayer. Sorry. You can get them working with mplayer if you try.)
Turns out that char isn't short for signed char, it's a distinct type:
% cat tmp.c
void foo(void)
{
signed char *ps = "signed?";
unsigned char *pu = "unsigned?";
}
% gcc -c tmp.c
tmp.c: In function 'foo':
tmp.c:3: warning: pointer targets in initialization differ in signedness
tmp.c:4: warning: pointer targets in initialization differ in signedness
From this thread
So it's been a while since I've written anything here. In fact I probably haven't had a gap this long since the last time I worked at Google.
A number of people have emailed to ask how I'm doing and I've had to batch up all the replies for the last week. Serving dinner at work is great from a healthy eating point of view, but I really don't feel like doing anything much when I get home at 8, 9, 10 o'clock after dining there. You all deserve personal replies, but it's Sunday evening, just before bed, as I write this and I have to admit that it's not going to happen.
Over the weekend I moved into the forth place I've had since I've been here. I'm averaging three weeks in any one place so far but this place is going to last for a year - really it is. It's a really nice house in Palo Alto with four other people. Thanks to IKEA I now have some furniture but not, yet, any Internet connection here.
And the climate is crazy. It's mid December and I can quite happily walk home at one in the morning in only a t-shirt. Of course the natives are complaining about the bitter cold and wrapping up in three or four layers, but such is the world.
It's also the time of Company Christmas parties. Google's was last weekend and was suitably huge, taking up two San Francisco pier buildings. Turns out that the cheapest way to get to Palo Alto from SF at 3am is to hire a stretch limo. Odd but true, and a cool way to round off the evening.
I'll not be home for Christmas, I've only been here a few months and my family shan't be there anyway. I suspect Christmas will probably be pretty quiet so, you never know, I might get round to writing another blog post by then .
Read the Google weblog post if you want, but the operative information is that you goto Google Print and use the date operator to restrict searches to public domain books.
If you are in the US you can use date:1500-1923 and, if you are outside the US, you can use date:1500-1846 (or proxy through a US host).
Update: fixed link - thanks Aaron.
Blink is really a collection of short stories (I've just finished it). They are well weaved together and they are all related - but I'm not sure that they really deserve to be in the same book.
Usually a work like this would be presented as a more old fashioned argument but Blink gets by without any real structure at all. And it's a good read which sells well so I can hardly say that that's a bad idea. Yet, by the end, I was left wondering what exactly I should be taking away from this. I made a list of all the stories, grouped them together and came up with titles for each of the groups:
So it seems to me that the book is arguing six different things. Or, at least, it would be if it were arguing anything at all.
Most of them are actually pretty interesting results, even if the point is pretty bland (e.g. the second group). I've heard about the system about divorces before, but the doctors getting sued was new to me.
If I were forced to draw it all together I would have to say that it's a pretty damming attack on the notion of The Ration Being. That poor being has been under attack from lots of directions (esp neuroimaging) for years now and it does feel like the ideal of the rational, scientific mind is going the way of Newton's clockwork universe in the face of physiological quantum theory.
(Just to recap, that means that my rational mind is flagging the fact that I don't seem to have a rational basis in believing that I'm rational. The irony is eye-watering)
It's a fun read, but I couldn't help looking at the price on the inside cover and thinking that it's not that good a read.
... at Google today (and we got free copies of “Blink” - don't you love Google.) It looks like his next book will probably have one of the themes that he talked on today.
The first was conceptual innovation vs experimental innovation. Conceptual innovation is the eureka moment - a new idea which is just very good. Picasso is his example. Picasso made most of his ground breaking stuff when he was young, he planned it, it was a new idea and he faded out as he got older.
Experimental innovation is the kind which takes a very long time to develop. First example: Cézanne. As opposed to Picasso he was unremarkable for decades. If you study his paintings from when he was 40 (Gladwell says) you would not predict that he would be world class in his 60s.
Second example: Fleetwood Mac. Their first hit album was Rumours and it was their 16th album. Their long suffering record company supported them through 15 duff albums before getting some money back. (Can you imagine that happening today?)
And that last comment seems to be Gladwell's hook for the book - we're missing too much experimental innovation because, as a society, we're geared towards conceptual breakthroughs
Next topic; targeting elite kids. Gladwell, he claims, was one of the top three junior runners in Canada at age 14. You wouldn't know it to look at him and, although the Canadian government sent him to special running training etc, he didn't turn out to be a great runner at age 21 etc.
Gladwell says that selecting at a young age is a terrible thing to do because performance in a given area (physical or mental) is a terrible predictor of success in that field when they get older. He seems to have a lot of studies to back this up. Education, he believes, should be more egalitarian.
One interesting study he quoted was the relation between going to Harvard and earning more later in life. Harvard claims that going there is great for future income (and it would have to be because Harvard costs a lot) but it turns out that the biggest predictor for earning lots is applying to Harvard. You don't have to get in, you just have to be the sort of person who believes that they can and are willing to try it.
Well there you go. I await the next book, Malcolm.
Update: There's a recent text, by Gladwell, from the New Yorker about Harvard's entry system.
I don't usually comment on anything Google related here since I started, but I'll make an exception this time:
Once upon a time there was a website, a kind of proto wiki, and, at the bottom of each page, was a link titled "Delete this page". The Google crawler did its crawl and that was the end of that website.
(The reason it came to light is because the webmaster of that site emailed Google "asking for his website back" (or words to that effect) and I believe that we dug them out of the crawl data for him. But this is besides the point.)
The point is that no one does that sort of thing any more. GET links on a website must not be mutable or you can be sure that one of a number of crawlers will mutate it.
But people didn't really learn, they just retreated behind login pages and such and made all the same mistakes. Now 37signals, no less, a group which carried my respect (until today) is getting very upset that crawlers are getting behind the login pages all over again.
I won't even comment that they seem to think that GWA has been pulled because of their blog post, nor about the commentators who are making a big deal of the MUST NOT vs SHOULD NOT wording in the spec. Here's the end result
It's a bad idea - anywhere.
(I'm not on the GWA team. This is not from them and involves no non-public knowledge of that project)
This is a big book, and not just because the font is really huge; yet I can't help but feel that it would have been much better had it been a lot smaller.
It's broken into three main parts. The first consists of a lot of graphs to really hammer home the message that exponential growth is happening all around us. And it's all very convincing in the areas which he chooses. Certainly everyone has come to live with and expect transistor counts to double every 18 months or so, but I will admit that I didn't know that productivity per hour of a US worker is also rising exponentially. What's neat is exactly how many of these graphs are really good straight lines.
That goes on for quite a while and then the main part of the book is a huge long list of all the cool things that are happening right now across a wide range of subjects. All these developments are introduced to support the idea of the singularity happening and, to give credit where it's due, Kurzweil does make solid predictions about when things will start the happen. The amount of research that has gone into this book is impressive and it reads like a 10,000 overview of most of the interesting work in science and engineering today. That's also the problem with it. There's page after page of the stuff and none of it ever goes into enough detail to really be interesting. As soon as you want to know more you're whisked away to the next wonderful development.
There is a large section of notes referencing everything, and this is good. But it's very hard to say that the book as a whole is very interesting reading. There is a third section, his responses to critics, but I don't feel that I actually want to bother reading it having slogged through the first two sections.
In the end, even if I'm convinced that it's all going to happen, just as he says, how is it useful information? Knowing that the washing machine is broken is useful information because it allows me to make better choices (e.g. to not bother trying to do any washing today). But knowing that the future is going to be wonderful is nice - but I can't see how it helps me yet. If I'm going to live for hundreds of years then maybe I should save more? But, if Kurzweil is right, then we will all be fantastically (materially) wealthy anyway, so it doesn't really matter.
How ever you spin it, there's a lot of hard work between here and there, so
get back to work and you might have a technological utopia in a few decades if
you're lucky
(I'm not sure if you get positive or negative points for knowing why I chose that title )
So this weekend saw the first (maybe of many) Startup Schools, run by Y Combintator, which is Paul Graham's hacker starter helping company. I'm sure there is lots of stuff being written about it, but I can't give you any links right now because I'm sitting in Boston Airport and the WiFi costs about $8.
Y Combinator have a couple of offices, one is walking distance from where I'm living in Mt View, the other is in Boston - the other side of the country. Guess where it was held? Nevermind, it gave me a chance to visit the east coast of the US, which I've never managed to do so far. The heavy rain certainly makes a difference from the montonous metrological monotone which is Bay Area weather. It's damp and cold here; it could almost be home. Photos on Flickr when I upload them.
The event consisted of a party on Friday night, all day talks Saturday and, I think, something today, but I've to catch a flight. The speakers list was very impressive and uniformly the talks were excellent and very well received. Of course, Google was there with a recruitment talk which (In the opinion of several people more independent than I), kicked the arse of the Yahoo talk.
It was very good meeting up with people; new people, people for the first physically, people unexpectedly and famous people. The latter category includes Stephen Wolfram (who is easer to talk to than to read), Paul Graham and Michael Mandel, the economics editor of Business Week.
I don't know how many good startups will come of it. I'm certainly not going to be one of them (yet), comforable as I am in the generous embrace of Google. But Y Combinator is doing an excellent job.
Still, I've a bike upside down on my kitchen floor with a punchture and a wheel nut which is too tight to get off with a spanner, a driving test to sort out, a better apartment to find and work tomorrow. Back to reality and, hopefully, back to posting a little more often than I have been.
I'll be at Paul Graham's Startup School this October 15th in case anyone who reads this will be there too. It's not that I'm looking to get out of Google before I've even started, but all good things...
Looks like I left it a little late to book the hotel though. All the near by ones are sold out on the Saturday and I've ended up in a Holiday Inn some way away. Nevermind, there are always taxis.
And I really need to replace this old laptop as my main system. I'm sure that it was good in it's day, but it was a hand-down from someone else many years ago and it just can't keep up with Firefox. Thankfully, Opera have now released their browser for gratis, and without advertising. I guess that they've decided that they aren't going to win the desktop wars now and they should concentrate on their mobile offerings.
It's certainly lighter than firefox (my laptop goes swap crazy a lot less) and it properly threads the page rendering so that loading (say) Bloglines doesn't freeze the whole browser for many seconds.
It's Javascript/general AJAXy support isn't so good. Gmail works though. And the tabs act slightly wrong when you close them - it switches to the last viewed tab, which is almost never what I want.
Still, if you too have a system with only 64MB of memory then give it a shot.
Am now living in Mountain View, CA. In a hotel at the moment, but with a week before I start work and Craigslist to hand, hopefully somewhere more permeant soon. I understand that most people start by living in Mt View and then, over time, move into San Francisco city proper as they realise how dull the nightlife is here.
By booking a seat as far forward as I could and with a brisk walk off the plane, I managed to be one of the first people into border control and cut my Time to Clear Border Control to 10 minutes (from some two hours last time). Most of that time was taken up with the official complaining that I hadn't filled out one side of the I94 form. Perhaps I should have stood up for foreigners everywhere and pointed out that the words "For Government Use Only" were written in large, red letters across the top. In reality I said sorry in fear of being sent to the back of a queue of several hundred people which I had put so much effort into being at the front of. So I'm a coward, but a less tired and pissed off coward.
The flip side of being fast through border control is that they can't clear the luggage off the plane that quickly. So, in future, it may be possible to look less like you're trying to get ahead of everyone while maintaining the same TTCA (Total Time to Clear Airport).
Still the hotel is good and I've just got myself a bike to get around on. I need to change rooms, however, because I'm out of range of the wireless network for the hotel. Sitting in the lobby with a laptop is a pain. Oh, and ALSA is very upset about my laptop and that's foiled my plan to use SkypeOut. It's listing no soundcards and yet it's playing quite happily. Sadly recording is right out.
I also see that Aaron Swartz has decided to keep with his startup company and not return to Stanford. And I was going to look him up now that I'm here too! Dropping out of college to start a company ... how's it feel to be a stereotype, Aaron?
So today I brought the new KT Tunstall album because it's winning lots of awards and sounded like it could actually be pretty good. I'm going to skip over why a pressed CD needed so much error correction work from cdparanoia on the first couple of tracks and skip straight to the mastering. Here's part of the (raw, PCM) waveform from the forth track (which I picked at random):
There's clearly some headroom there, but it's been really aggressivly compressed - and it sounds like it. I've never heard a kick drum sound like the skin was a hot water bottle.
Which is a terrible shame because there's good music under there somewhere.
Update: just to compare I thought I'd show a bit of Dire Straits at the same scale:
New pyGnuTLS release thanks to a patch from Johan Rydberg
12 days to go... A few things which I've been doing recently before I loose them:
Adding libevent support to Gambit Scheme: [patch]
Adding edge triggered support to libevent: [patch]. You need this for the Gambit patch. Also, does anyone know if Neils (the libevent maintainer) is still alive?
A small libevent based async DNS library designed to be embedded into applications rather than shipped as a .so file. This acutally has a known, rare bug in it but I'll have a fix and a real release of this soon: [eventdns.c eventdns.h]
| / | Root |
| Alternate | The Weird and Wonderful |
| Backlinks | What are backlinks |
| John Gilmore | What's Wrong with Copy Protection |
| Archives | Blog Archives |
| One | Archive 1 |
| Two | Archive 2 |
| Three | Archive 3 |
| Four | Archive 4 |
| Five | Archive 5 |
| Six | Archive 6 |
| Seven | Archive 7 |
| Eight | Archive 8 |
| Nine | Archive 9 |
| Ten | Archive 10 |
| Eleven | Archive 11 |
| Twelve | Archive 12 |
| Thirteen | Archive 13 |
| Fourteen | Archive 14 |
| Fifteen | Archive 15 |
| Sixteen | Archive 16 |
| Seventeen | Archive 17 |
| Eighteen | Archive 18 |
| Nineteen | Archive 19 |
| Twenty | Archive 20 |
| Twenty One | Archive 21 |
| Twenty Two | Archive 22 |
| Twenty Three | Archive 23 |
| Twenty Four | Archive 24 |
| Twenty Five | Archive 25 |
| Twenty Six | Archive 26 |
| Twenty Seven | Archive 27 |
| Twenty Eight | Archive 28 |
| Twenty Nine | Archive 29 |
| Thirty | Archive 30 |
| Photos | Poor People Caught on Film |
| Jack and the Beanstalk | Jack and the Beanstalk |
| RIP Scan | Results of a Stage Scan Fire |
| Yosemite | Yosemite National Park |
| Projects | Incomplete things from the lab |
| Seagull's Bane | Linux Automounter |
| bttrackd | BitTorrent Tracker |
| CAPTCHA | CAPTCHA CGI script |
| Conserv | Console Serving |
| Deerpark | Using Tor with Firefox/1.1 (Deerpark) |
| DNSFix | Fixing DNS |
| Xovers | XTA Crossover Control |
| IAFS | Archive Org Storage |
| JBIG2 | JBIG2 Encoder |
| Verify | PGP Key Verifier |
| MaxFlow | Maximal Flow in Python |
| PyBloom | Bloom Filters in Python |
| pyGnuTLS | Python wrapping of GnuTLS |
| Sxmap | Apache SuEXEC Map |
| Hellard | Union Server Notes |
| Recordings | Free recordings |
| ICSM Choir | St Paul's Church |
| School | Ancient School Stuff |
| Writings | Who knows |
| Cap Systems | Capability Systems |
| Intro | Introduction to me |
| Suprema | JMC2 Group Project |
| MP Letters | Letters I've written to my MP |
| Sound | Sound With Dramsoc |
| SyncThreading | The wonders of user-land threads |