3 May 2002 raph   » (Master)


I think it was Zooko who pointed me to Kademlia. It's good. In particular, it's a lot less fragile than Chord because of their XOR metric.

Of course, it's not attack resistant yet. I'm sending mail to them to point to Chapter 7 of my thesis, which is actually very similar in a lot of ways to what they're doing. My problem is that I haven't gotten my ideas published yet.


I see more and more recognition that email is becoming unusable because of spam. The recognition is a very good thing, largely because it funds and motivates research into spam-resistant messaging. There's now a ton of work on the problem, exploring many different directions. Also encouraging is the fact that both free software and commercial products and services are well represented.

Making email resistant to spam is a hard problem. I believe that attack resistant trust is a very good approach. In particular, I believe that the stamp trading network in Chapter 7 of my thesis is likely to do a good job filtering out the spam without trimming the legitimate messages. Of course, the only way to test this hypothesis is to build a prototype. I don't have the time right now to do so myself, and so far I haven't been able to interest anyone else in it either.

Until quite recently, I'd believed that email would have to be completely replaced by a better system. The existing email infrastructure, of course, not only lacks any kind of attack resistance, but also lacks any form of authentication, which is needed in the next level down. Thus, I reasoned, to solve these problems, you have to build something new.

But a few days ago, in a conversation with Jeff Darcy, I realized that this is not necessarily the case. Highly sophisticated email filtering clients will, almost inevitably, implement a p2p network with good trust properties. Vipul's Razor is already a significant step in the right direction.

Such a client could then easily implement a real attack resistant trust metric. If both the sender and receiver clients participate in compatible p2p networks, then approval for the message could be expedited. Otherwise, the message would have to take its chances.

Assuming that this trust-based message approval works well, then in time it will probably get a lot of penetration. When that happens, you could probably just turn off untrusted smtp delivery.

But how do you know in advance whether a trust system will work well for email? Building it and fielding it in the existing email infrastructure is really hard. Plus, there are lots of different ideas on how to do this (my stamp trading network is but one). How do you know which one is best?

To me, the answer is clear: build a prototype. Such a prototype would not have to work with the existing email infrastructure, or attract a huge user base (just large enough to test whether it is spam resistant). Once the prototype proves the ideas, then the difficult work of integrating into the existing email network can begin.

Subpixel positioning

I have a hacked up version of Ghostscript that does unhinted antialiasing with subpixel positioning. See before and after screenshots.

Now that I've posted the screenshots, I'll have to make the patch really work. I have code for making the cache subpixel-aware, but it's not 100% yet (the "after" screenshot was made with caching disabled). Also, I'll need to add a few configurable parameters, especially to turn off hinting by default in aa mode (right now, it's compile time).


Denver was fun. We went up to Winter Park on Monday, and the kids got to play in the snow a bit.

Alan tested, as we expected, highly gifted. Less expected was the large discrepancy between verbal skills (off the scale) and visual/spatial skills (mazes and the like), at which he is basically average. They recommended an optometry consult. That will be interesting.

When we got back, I think Max made his first pun. He said, "miss spider", which either meant that he missed his favorite spider book, or that he was referring to more of the title words. Of course, it's possible I'm just reading the pun into what he said, but even so I wouldn't be surprised if it were for real. All the people around him love language and delight in puns, and he's a quick study.


Desmond Tutu: Apartheid in the Holy Land. This is one of the sanest things I've read, about an insane conflict.

Latest blog entries     Older blog entries

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!