Musings

This is a slightly convoluted tale of spam and the troubles it causes.

SMTP AUTH is the standard for Authenticated SMTP. Back in the “old days,” email delivery was a cooperative process, with each mail server relaying mail onward to its destination. Then came the spammers, who exploited the “open” nature of SMTP to have other people’s mail servers deliver their spam. So mail servers had to be reconfigured to relay only for specific hosts, or for clients which properly authenticated themselves. Hence the need for SMTP AUTH.

When I first got my iBook, I wanted to set it up to use SMTP AUTH, with golem as the outgoing mail server. At the time, Mail.app’s SMTP AUTH support was broken. But, since MacOSX is Unix, that was no problem. I just told Mail.app that the mail server was localhost and configured sendmail on my iBook as a nullclient, using SMTP AUTH to forward all mail to golem.

Eventually, Apple fixed Mail.app’s SMTP AUTH support. But I kept using the above-mentioned system because it was more convenient. I could “send” emails while offline, and have sendmail queue them up for delivery as soon as I got back online. Worked great, and never gave me a lick of trouble.

Imagine my shock on this trip when I discovered that all of the emails I had “sent” out over the course of two days were still queued on the iBook. None had gone out, despite having a working DSL connection! The reason, it turned out, was that Covad DSL filters all outgoing traffic on port 25 (SMTP). Sendmail on the iBook kept trying to contact golem, but the connection kept timing out.

Why is Covad doing such a seemingly dumb thing? Because spamming technology has evolved. Another favourite spammer technique is to use a program which directly contacts the recipient’s mail server, bypassing the spammer’s ISP’s mail server (which might, say, limit the number of outgoing messages per second!). Instead of going after the spammer for violating Covad’s AUP, they decided to simply block all outgoing SMTP traffic. Spammers can’t send their “direct-to-MX” spam, but “roving” users like myself can’t contact their home mail server either.

What to do? Turns out it’s time to ditch sendmail, and reconfigure Mail.app to contact golem, not on port 25 (SMTP), which is blocked, but on port 587 (MSP) which, mercifully, Covad has left unblocked.

I post this hint because, even if you are doing everything right (using authentication, not running an open mail relay, …) you will eventually get screwed — if only indirectly — by the spammers. And I doubt that more than one person in a thousand knows that, in addition to listening as an MTA on port 25, sendmail also listens as an MSA on port 587.

P.S.: This post was composed with the latest version of Kung-Log, which uses the recently-released Webcore rendering engine (i.e. the same one used in Safari) for previewing. Sweet!

That’s one of the long list of prohibited items in your checked luggage. With self-service check-in now standard, my conversation with the clerk at the American Airlines counter consisted of some exchanged hilarities about the laminated sheet of prohibited items (complete with little pictures).

I’m off to Palo Alto (on family business), and there’s no more high-tech experience that flying the Nerd Bird from Austin to San Jose. The woman next to me has a laptop, a cell phone, a Palm Pilot and a digital camera with a stack of SmartMedia and a USB cable — all operating simultaneously. To complete the look, she’s wearing a high-tech surgical mask.

Last night, I was using Google to do a little background research on MAO Inhibitors (not for me, thank God!). They inhibit the breakdown of tyramine, an amino acid present varying levels in a weird assortment of foods. Too much tyramine in your bloodstream can lead to a surge in blood pressure, brain hemorrhage and death. So folks using MAOIs have to be very careful. Interestingly, I found that the top-listed hits (and much of the most useful information) came from web sites selling (or extolling) shrooms. I’ll leave the organic chemistry involved in that little association as an exercise for the reader.

Watched Steve’s WWDC Keynote and, after the RDF wore off a bit, here are my thoughts:

1. The G5 is a great relief. Thanks to IBM, Apple finally has a next-generation 64bit CPU to replace the aging G4 (without a hint of irony, I am writing this post on a G3-based iBook). And they’ve built a kick-ass system around it. What surprised me was the absence of a rack-mount version. They badly need this for the XServe II.
2. NextStep redux. After all these years, the “FAX” button in the Print dialog box finally makes its triumphant reappearance. More amusing still, was the malarky surrounding the new Finder (“User-centric,” not “Computer-centric!”). I should reserve judgement until I can see 10.3 in action, but from the screen shots, it’s clear that the left-most column of the new Finder is a born-again version of the Shelf. Now, the Jaguar implementation of the same idea (as a toolbar) is next-to-useless, so this must surely be an improvement. But, since I was never much-enamored of Workspace.app (the NeXT name for the Finder), I am somewhat underwhelmed.
3. iChat AV and the new iSight camera look really cool. I don’t do IM, but ad-hoc video-conferencing, now that’s a bandwidth killer I could get into.
4. Otherwise, Pather (MacOSX 10.3) looks to be chock full of good features:
   • Fast User Switching (I know XP has it first, but this is a slick implementation and the rest of the OS doesn’t suck.)
   • On-the-fly directory encryption
   • IPSec, FreeBSD 5.0, X11,…

They just keep churning out new features while the “rest” of the world waits patiently for Longhorn (which prompts one of the more amusing clips in the Keynote).

Safari 1.0 is out. Pretty darn good for a 1.0 release. It is not without CSS bugs though. Three happen to affect this blog:

1. overflow:auto doesn’t work right. The scrollbar obscures the text.
2. list-style-position:inside doesn’t work. The contents of the <li> overlap the numbers.
3. border-collapse:collapse for tables doesn’t work.

Here’s a little reduction to illustrate the first two problems. For the third, just look at the table in the previous blog entry (ignore Safari’s lack of MathML support).

I’m hoping Dave Hyatt is hard at work on Safari 1.1.

One of the most celebrated connections between math and physics is the relation between knot theory and 3D Chern-Simons theory. If you consider a Chern-Simons gauge theory for a compact gauge group, $G$, the natural observables are the expectation-values of Wilson loops in unitary representations, $R_i$ of $G$. These compute the Jones polynomial (for $G=SU(2)$) and its generalizations.

3D gravity can also be cast in the form of a Chern-Simons theory,

At least, with Euclidean signature, you might ask whether there’s some relation between 3D quantum gravity and knot theory. Recently, Gukov has proposed that, with negative cosmological constant, 3D quantum gravity is related to the A-polynomial of knot-theory

A while back, I blogged about the RIAA’s support of efforts to develop Trojan Horses, DoS Attacks and other nasties with which to attack P2P file-sharers. I speculated that we would eventually see a renewed effort to pass a “Berman II” bill, exempting them from anti-hacking laws.

Some scoffed, but a little over a month later, the RIAA seems to have enlisted the support of no less than the Chairman of the Senate Judiciary Committee:

During a hearing that Hatch convened Tuesday on the “national security risks” of P2P networks, he asked a witness, “Can you destroy their set in their home?” referring to a home PC.

Randy Saaf of MediaDefender, a secretive Los Angeles company that works with the recording industry to disrupt P2P networks, replied by saying “nobody” is interested in that approach.

“I’m interested in doing that. That may be the only way you can teach someone about copyright…That would be the ultimate way of making sure” no more copyright is infringed.

Hatch suggested that Congress would have to amend laws restricting computer intrusions. “If it’s the only way you can do it,” Hatch said, “then I’m all for destroying their machines…but you’d have to pass legislation permitting that, it seems to me, before someone could really do that with any degree of assurance that they’re doing something that might be proper.”

I guess nobody’s laughing now …

Sometimes, it just doesn’t work out, and you find yourself, at midlife, needing to retool. Maybe there’s a company out there which needs your talents. Or maybe there are other options.

(Okay, I know it’s a little old, but I miss him so …)

I spent my last afternoon in Israel at the Tel Aviv Museum of Art. They had Retrospectives of works by Liliane Klapisch and the late Ari Aroch. The catalogue for the latter wins the prize for best one-sentence biography:

“Born the son of a Zionist Jew in 1908 in the city of Kharkhov in the Ukraine, he spent his life half in Jerusalem, half in the world’s capitals, and half in art, half in diplomacy, and died a modern legend at the age of sixty-six.”

Why go on, eh?

Klapisch, who lived and received her artistic training in Paris, immigrating to Jerusalem in 1969, was, in an odd way, more interesting. Aroch is, arguably, much more original. But she brings to bear a more profound technical vocabulary (with influences from Cézanne, through the cubists, through the post-war French abstractionists) on the visual landscape of her adopted land. And I can’t help but have been influenced by the fact that her exhibition was much better-staged.

Anyway, the permanent collection of modern art is also worth a look. It contains, among other gems, a healthy sampling of the delightful post-cubist constructions of Alexander Archipenko.

This blog received its first spam comments today. Three comments were posted from

24.125.118.3
c-24-125-118-3.va.client2.attbi.com

which now has the dubious distinction of being the first IP address on my blog’s blocking list.

Two of the comments involved the hack:

<body onload="window.location='http://www.goatse.cx';"/>

using different Comment TextFilters (in the forlorn hope that the choice of TextFilter would make a difference). One simply left http://www.ak47.il/ as a URL.

All three were filtered successfully, and neither the evil Javascript redirect (which would have sent IE users elsewhere upon loading the Individual Entry page), nor the above (bogus) URL, made it into the published comments.

I decided not to delete these comments this time, as further forensics may be interesting.

It’s clear this was just some pimply-faced kid experimenting, rather than a serious spammer. Automated comment-spam attacks could be seriously unpleasant. And I don’t know of an easy solution any more than I know of an easy solution to email spam.

Speaking of email spam, there’s a string theorist in New Jersey who

• uses comcast.net as a his/her ISP
• runs Windoze on his/her home computer
• has either not noticed, or doesn’t care that his/her machine has been infected by some Microsoft Outlook email virus for the past two months.

The IP address of the offending computer has recently been

68.46.139.218
pcp155581pcs.lambrv01.nj.comcast.net

but, being dynamically assigned, is probably subject to change over time.

Up until I started blocking mail from comcast.net (a drastic step, I know), this one individual alone was the source of a half-dozen bogus messages a day to my account. I’d love to unblock comcast.net, so, mystery person, let me know when you’ve fixed your computer …

I hate saying , “Hmmm. Maybe I’ll take a taxi to Tel Aviv University, instead of hopping on the train.”

But lunatics are loose in the land, firing rockets from helicopter gunships and blowing up buses. Suddenly, public transport doesn’t sound so appealing.

President Bush looked really pissed-off last night on CNN. I’m not ordinarily much of a fan. But, in this case, perhaps a good brow-beating from his “friend” is just what Mr. Sharon needs right now.

The OmniGroup have release a public beta version of Omniweb 4.5. The new Omniweb uses the same WebCore KHTML-based rendering engine as Safari. That’s good, because OmniWeb’s previous rendering engine had crappy Standards-support. But it also means that, by default, OmniWeb’s new USER_AGENT string contains the word “Gecko”, just like Safari. (In keeping with its rebel roots on the NeXT platform, OmniWeb lets you customize the USER_AGENT string it sends.)

To accommodate the new release, we need to tweak our mod_rewrite rules to send it text/html:

RewriteCond  %{HTTP_USER_AGENT} Safari

becomes

RewriteCond  %{HTTP_USER_AGENT} Safari|OmniWeb

The postdocs, Drukker, Fiol and Simon, here at Weizmann have put out a paper on a possible stringy resolution of the Gödel geometry.

String theory has a variety of mechanisms for resolving the pathologies of general relativity. The one most relevant here is the enhançon mechanism, in which a collection of D-branes puff out in the transverse directions, effectively excising the neighbourhood of the singularity, replacing it by a smooth interior geometry.

Drukker et al look at the Gödel universe with fluxes (a configuration of type IIA preserving 8 supercharges)

where $A=- c r^2 d\phi$. For $r\gt 1/c$, this geometry has closed timelike curves (the circles $\phi\in [0,\pi]$, with the other coordinates constant are spacelike for $r\lt 1/c$, but timelike for $r\gt 1/c$.

In a sort of inside-out version of the enhançon, they argue that there’s a supertube configuration whose interior geometry is Gödel, but whose exterior is smooth, with no CTC’s.

Now, the nice thing about the enhançon is that it’s a local effect. If you had some physical process which would lead to a repulson singularity, you expect the nucleation of some D-branes which would pop out from the origin and repair the singularity. This is a local process; the asymptotic behaviour of the metric is unchanged.

In the Gödel geometry, it is the asymptotic region that is bad. Drukker et al’s supertubes nucleate, as it were, at infinity, and come in to repair the metric. Normally, we would reject such a scenario as bananas. It requires a nonlocal conspiracy of the quantum fields at arbitrarily large separations. On the other hand, this region of the Gödel geometry has closed timelike curves, so there is infinite scope for such a conspiracy to assemble itself. Maybe not so bananas …

I don’t want to advocate this picture too literally. The physical question is whether there is some initial configuration which would, in GR, evolve into a geometry with closed timelike curves and whether it is saved from this fate in string theory. The Gödel geometry does not address the former question, so any “resolution” of Gödel does not address the latter.

(5/31/2003) So I have five hours to kill in the Continental Terminal at Newark Airport. You’d think that this spanking new Terminal would be replete with all the latest creature comforts. You’d be half right. There’s a “Meditation Room”, and Sushi for sale at the food court. But no WiFi. Not even the Starbucks had a hotspot.

Grumble …

I’ve been thinking a lot about the theft of my bag and what to do in the future to safeguard my notes.

Some people conscientiously TeX up all their calculations. If I did that, I’d have a copy on golem and a copy on my iBook, in addition to (or, perhaps, instead of) the original hand-written copy in my notebook. It’d be handy when it came time to write things up (something I’m insufficiently assiduous about). On the other hand, I ‘m not sure I’m ever going to be conscientious enough to pull it off.

A lower-tech alternative is to xerox all my note and keep one copy at home and one copy at work. Less stuff to schlep back and forth. [Interesting side note: the cocoaAspell spell-checker agrees that “schlep” is an English word.] But bad for the trees.

We’ll see…

(6/1/2003) My forebearance of the primitive conditions at Newark was rewarded by an uneventful flight. Surreal conversation one row over on the relative threat of suicide bombers vs SARS as disincentives to travel. Those involved were evidently unfazed by the former; they couldn’t quite agree on the latter.

(6/2/2003) I’m settling in at the Weizmann Institute for a short stay. By late afternoon, I finally have network connectivity.