Posted on by lhl

Quantum Mechanics: Not Just a Matter of Interpretation

It has been widely accepted that the rival interpretations of quantum mechanics, e.g., the Copenhagen Interpretation, the Many-Worlds Interpretation, and my father John Cramer‘s Transactional Interpretation, cannot be distinguished or falsified by experiment, because the experimental predictions come from the formalism that all such interpretations describe. However, the Afshar Experiment demonstrates in an interaction-free way that there is a loophole in this logic: if the interpretation is inconsistent with the formalism, then it can be falsified. In particular, the Afshar Experiment falsifies the Copenhagen Interpretation, which requires the absence of interference in a particle-type measurement. It also falsifies the Many-Worlds Interpretation which tells us to expect no interference between “worlds” that are physically distinguishable, e.g., that correspond to the photon’s passage through one pinhole or the other.

Fascinating discussion in the comments.

Seach under “Vaidman bomb” for an interesting take on interaction-free measurements. This is related to the “seeing with no light” article in Scientific American a few years back. Surprisingly, one can build experiments which have radically different results contingent on adding a component with arbitrarily small probability epsilon of interacting. The bomb part is phrase this way: Suppose I have a photo-detector connected to a bomb trigger. With what probability can I determine that the photodetector is there or not without setting off the bomb? The answer is I can do this with arbitrary small probability of setting off the bomb. This sounds very similar to Cramer’s description of Afshar’s result, and is perfectly understandable in all three interpretations of quantum mechanics.

more…

Sorry, folks. The waveform was collapsing along nicely there for a while, but in pure non-deterministic fashion, we’re back to no one actually knowing anything…

Posted on by lhl

Peter Van Dijck‘s blog, Guide to Ease is great and chock full of all kinds of ideas and links that I have no time to follow right now.

Lately I’ve been closing tabs instead of blogging. Forget email, Google needs to organize my bookmarks/browser history. Soon to dump: geolocation, geowanking links, more Gmail thoughts, deconstruction; oh, also should write about Wil Wright, who spoke on campus Thursday. That was frickin great.

We’ll see how I survive the next couple days. Have a couple large projects/papers to (try to) finish.

Posted on by lhl
Posted on by lhl

Tracking some gmail (related) conversations:

So, Technorati has introduced the idea of conversation tracking in their iconography, but it’s not actually displayed or ordered as one. (I was going to include technorati links to all these, but it doesn’t didn’t because it’s not actually that useful for tracking conversation flow from these links).

Posted on by lhl

Last summer I volunteered to do a site redesign for the Southern California Linux Expo. In the end, for what I thought were timing reasons, they decided to keep the old site.

So, a new redesign just went up today for next year’s site (I’m on the planning list still). Ummm, ok. Granted there’s a a bit of ego-stake here, but trying to look at it objectively, do people really think this looks better than this? That’s sort of scary.

Posted on by lhl

My mail configuration is still messed up on my new server, but I got around to configuring CRM114 at work (moved my trained .CSS files over and it worked flawlessly). I’d almost forgotten just how much of an amazing thing it is to have a spam-free inbox. It’s like being able to breathe after being completely congested.

Although huge productivity impact numbers are thrown around, I think that the general quality of life and mood impacts of spam haven’t been adequately accounted for.

Anyway, while I was at it, I updated my Mail.app smart training script. One functional fix (new versions of CRM114 add an X-CRM114-Version header that I now also drop), and then writing install and usage instructions at the top (zowie, documentation!)

Posted on by lhl
  • Lawrence, Kansas or Paradise – follow along as our intrepid hero tours w/ one of his bands and debugs GMail on the road
  • Real Dialogue: The Tech interviews Jack Valenti [PDF] – an old friend passed along an interview w/ Jack Valenti by Senior Editor Keith J. Winstein in the April 16, 2004 edition of The Tech. Worth a look.
  • DECREASE YOUR ERDOS NUMBER! scientific coauthorship – score an Erdos number of 5 (a research project by Bill Tozier)

    During that period, the seller will provide expert technical advice on research projects in the fields of evolutionary algorithms, machine learning, agent-based modeling of complex biological and social systems, complex systems research in general, social network theory (including business and marketing applications), engineering design automation using machine learning algorithms, artificial life, and any of a number of other specialties (a more comprehensive list available on request; a complete curriculum vitae will be provided to the winner).

  • Transform! – awesome
  • Pashua – Dialog GUI for Perl, PHP, Shell, Python, …
Posted on by lhl

I got am email from the author of CLIX requesting elaboration on my comments on the interface and I felt it only fair to write a substantive response. That actually took a while to do, and since I went through the trouble of it, I might as well share.

CLIX is a simple (and very appealing) idea, a cmd-line trainer for newbs (somewhat similar to The Regex Coach), but the current implementaiton, a single-pane multi-table (w/ descriptions cut off and ordered by title) actually seems pretty hostile to the audience this tool is serving. A double-click drops a sheet which lets you edit and run the command.

So, my biggest suggestions were:

  • the ability to search/filter
  • make sure that the presentaiton was oriented to tasks
  • make sure that enough of a description got listed to describe what each item did; right now you need to either expand the description a lot or double-click into a modal sheet just to see the description

To do this, I made one possible suggestion, a 3-pane interface

          /--------------------------------
/---------| [Search__________] [Filter[v]] |
|category |--------------------------------|
| + xxxx  | Name                  category |
| + xxxx  | Name                  category |
|         | Name                  category |
|         | Name                  category |
|         | Name                  category |
|         |=============AV=================|
|         | (viewer)                       |
|         | Name                  category |
|         | Description                    |
---------|                                |
          | cmd                            |
          |                                |
          | shell results                  |
          |                                |
          --------------------------------/

This is by no means the ultimate interface, there are a lot of different ways to go about presenting this, but what you get:

  • instant familiarity, looks like Mail, right? conceptually works the same as well
  • categories on the sidebar gives easy nav/filtering organized around tasks
  • search at the top allows for quick access (filter restricts search to certain fields; make the default/top category an ‘All’ view so search is attached to the category views, would avoid conceptual messiness of searching from all or selected categories
  • w/ the preview pane, you longer need a modal sheet to run the program. You get the bonus of easy scanability – full description + easy navigation (just like how you interact w/ email)
  • the fields in the list view should probably be customizable (and adapt, ie: category would disappear when you’re not in a global view/search)

RFEs: track date/frequency of use of commands to automatically build lists of commonly and recently run commands; possibly also allow ‘starring’ (ala gmail – a quick way to mark favorites)

If anyone has any thoughts (critiques, more effective alternative layouts), feel free to discuss.