Ex Bibliotheca

The life and times of Zack Weinberg.

Wednesday, 23 October 2002

# 9:10 PM

Leonard reports the creation of a nifty utility. And I'd like to say that I don't get why people like putting spaces between the parentheses and the argument list, either.

# 6:55 AM


My mother points out that I misspelled bain Marie earlier — the first word has an I in it. This is no doubt why I couldn't find Maria the Jewess's history online. She wasn't French at all; she lived in Egypt in the third century BCE, or possibly the first century CE, sources differ.

In the "why didn't I think of that?" category, a chap name of Tkil suggests that the way to prevent your car's wheel from turning when you're trying to loosen the lug nuts is not to jack it up until they're already loose.

# 6:10 AM

more "trusted" computing

Seth responded to my earlier comments on his owner override concept.

Let's back up from the technical details and talk about goals. The status quo is that if you've got a computer and a chunk of data in a computer-readable format, you can do whatever you want with the data. In particular, you can make an unlimited number of perfect copies of that chunk of data, and transfer them to other people, without any effect on the original. There are people who would like this not to be the case, and they have designed technological measures such as Palladium which could prevent it in the future.

I am perfectly happy with the status quo. In fact, I prefer the status quo to the alternative. However, it is possible that a system similar to Palladium could be designed which I would prefer to the status quo. I'm the customer; the people pushing Palladium and/or other "trusted computing" initiatives have got to convince me to buy new hardware that implements it. (Let me remind you that most corporations can be prevented from doing things by not fucking paying for the product.)

So, on the hypothesis that we are going to design a new computer architecture incorporating something similar to Palladium, what functionality should it have, and equally what functionality should it not have, to make me consider it something worth buying? Here are some examples of both categories. Features that would be useful:

  • Store data such that only I can read it, where I am authenticated to the system by some mechanism more secure than a username+password pair.
  • Transmit data to a remote server for storage, and retrieve it later, without practical risk that the operators of that remote server will be able to read it.
  • Transmit data with an assurance that the intended recipient, and only the intended recipient, will initially be able to read it. (This mechanism must not prevent the intended recipient from relaying the data to a third party.)
  • Issue an unforgeable authorization to transfer a specific sum of money from my bank account to someone else's bank account, with assurances that the money will be received by its intended recipient, and that the intended recipient will get only as much money as I wanted them to.

Features that would be undesirable:

  • Store data such that it can be read by only one physical computer.
  • Transmit data such that only its intended recipient will ever be able to read it (i.e. such that intended recipient cannot pass it along to someone else).
  • Transmit data with an assurance that at some future date I can render any or all extant copies of that data unreadable.
  • Receive data and force me to honor any sort of sender-imposed restriction on its use or distribution.

Now, can we design devices and primitive operation sets that permit the implementation of the desirable features, while preventing the implementation of the undesirable features? I suspect we can. However, I suspect the result doesn't look much like Palladium; I suspect it's more like a normal computer with a smart-card interface.

In this vein, I'd like to point out Richard Stallman's opinion piece on "treacherous" computing (as he styles it), and also his much earlier essay The Right to Read.

other political items of interest

The organization Transportation for a Livable City has released a roadmap for improving transportation in San Francisco.

This open letter encourages the FCC not to bail out failing telecoms companies.

Thirty-five questions that haven't been answered, but should be.