Last night, care of Helen, we blagged our way to a viewing of The Barber of Seville at the Royal Opera House, organised as a promotional evening by uniqueliveevents.com, a newish consortium formed by BT Broadcast Services and Shooting Partners. The opera was broadcast live from France, and displayed projected via JVC’s QX1 projector (“the highest quality image currently available in the world using DILA technology”). Where we were seated, there was a lot of fan noise from the projector, but the image and sound quality was very good, and the whole uniqueliveevents.com proposition is an interesting one. Their pitch is that they have a network of venues wired up and ready to go, and that they can provide ‘full service’ in getting events streamed live to any of them — or to a location of your choice. The core of their offering is distributed broadcast of large events to people who can’t otherwise attend — linkups with outdoor events, or local viewings of, say, sold-out concerts. I’m going to give them a call next week to get more details of their cost structures and business model — their sales guy seemed to have enjoyed a little too much of the hospitality to give us much detail by the end of the second act.

I recently replaced the motherboard in my mp3 box with an EPIA 5000 to get rid of the fan I needed with the EPIA M-series. I’ve also connected it up to my plasma screen to interface to Otto. With a modified Trident video driver (the version I’m using is also archived here) and a tweaked XF86Config care of Alastair Robinson (who wrote the driver), I now have 720×400 widescreen (16:9) output, which means non-stretchy video output on the plasma under X. Very nice.

The next Project JXTA Engineering Meeting is happening on March 18 in San Fransisco. Teasingly, the following topics are on the agenda:

• Alternative Peer Group implementations to demonstrate the flexibility of the JXTA architecture. (One size does not fit all needs.)
• Simplifying the Application level JXTA API to ease adoption. (perhaps by wrapping JXTA with the BSD Sockets API or JavaSockets API.)

I’m hoping that this suggests that the JXTA team are giving some thought to the meta-goal of using JXTA to build a ‘platonic network’ on top of TCP/IP, so that current applications can simply slot-in JXTA underneath their current socket calls and get all the advantages of JXTA without the pain. I’m also hoping they give some serious thought to performance issues — my Great Hope for JXTA is that it becomes a very simple API for exploiting distributed resources on the Internet without having to worry about firewall issues and topology, and the above two issues are important markers that this might yet happen. I’m eagerly awaiting the minutes from that meeting…

There was a post on some feed somewhere a couple of days ago saying that classical music sales haven’t been hurt at all by p2p, because no-one can be arsed downloading an opera — strikes me that there are any number of things that seriously efficient distributed chunking ala Gnougat will make possible (including pirating Tosca).

For example, the simple rsync script I have between my ripping machine and my mp3 jukebox works perfectly (chunk size == MP3 file, and plenty of wifi bandwidth), whereas the same system would fall apart for sending a movie over the public internet unless you add in decent error detection and recovery, and even better, smart caching and routing.

But rsync or similar works fine between my two machines at home for mp3 sharing, and in fact for that purpose it would work perfectly across the Net, to provide a service which in itself might form the basis for something far more interesting, if niche, than Napster — some neural network thing that categorises newly acquired music and shares it out via rsync to those of your social network the system thinks would like it. Easyish to do, and completely automagic — with something like Otto as a frontend, new stuff would just get randomly slotted into the feed that your local jukebox plays out, which would be pretty cool.

Or I can set up a local folder called ‘new stuff’ where a script automatically symlinks new music, and your machine rsyncs to that, or I have a folder called and your machine picks up new things entered there.

For me those thoughts are much more exciting than Napster and its ilk — generally I want to hear good new things I don’t know about, rather than hunting for things I already know — generally if I know about it, I’ve already got it or a simple route to it — its the unforeseen but lovely things that I want to have streaming into my jukebox…

This is an instance of something very specific I’ve been thinking about recently — I think the potential for automagic video-on-demand and streaming music (cf Tivo/Sky+ or Otto+rsync+some intelligent agents) is that these tools turn very lean-forward media experiences (multi-channel TV, MP3+time consuming p2p search), into the kind of lean-back experiences that ‘old’ TV and analogue radio have presented for most of the last century — experiences that wash over the viewer/listener, which is I think what most people actually desire from these media and their modern descendants.

Of course there’s also the meta-issue of programming — at the moment my Otto system has no idea what time of day it is, or what mood I’m in, or even what music segues well into what — metadata is important here as well. But even without, my Otto setup feels like ‘radio playing the kind of music I like’, not ‘tech I have to fight with to find things and schedule them’. The experience changes from active to environmental. Tune in and drift off. Experientially that’s a Big Important Difference.

My earlier post was a little unfair on mudlondon and other geolinked MOOS, I think. By my own definition — location is what becomes of places when things happen there — sites such as mudlondon may well become places, assuming stuff actually does happen there — people gathering, talking, building. My hidden bias and agenda, which I should have articulated at the beginning, is that for the geolinkage to mean much to me personally, the experience of the virtual place must in some way play off, rather than simply representing, its real referent — the fascination for me is in the dissonance between the experience of real and virtual places, the experiential moire generated from the interplay of knowledge of the real and experience of the virtual. I’ve always felt that the resulting fracture is what makes possible a poetry of augmented space — something similar to the effect of lexical ambiguity in textual poetry, shimmering between interpretations. Tokens, not metaphors.

Following from a posting from Dave Weinberger, I’ve been thinking about hotels. Hotel guests constitute potential temporary communities of interest: groups of people dislocated from their familiar context, but with similar needs. Everyone wants to know where to eat, what’s going on, and everyone has an interest in their temporary residence: the quality of accomodation & service, the overall experience.

Hotels have a real opportunity to add value through the provision of infrastructure to help their guests to network & work as a community, sharing their experiences & building local knowledge.

It seems that the hotel sector is slow even to provide more than slow, expensive dialup networking, and missing completely the opportunity to provide local network services. The quickest, most effective way for innovators to gain a market edge would be the provision of inhouse wireless nodes. This would provide valued guests with very flexible internet access, while exploiting the characteristics of reception to provide a platform for valuable community services, including pseudo peering for guests to share knowledge amongst themselves. In essence, such a platform would build on the ideas of presence being exploited by the emerging generation of location-based people-finding systems being developed for conference events. Guests could locate & network with other like-minded guests, and share knowledge related to both the hotel itself & its locality, without the central resource costs of a centralised ‘online concierge’. Thoughts to explore in depth later.

[later: Dave also points to a news story relating Intel's tie-ups with hotel chains to provide at wifi hotspots. It's unclear from the story if these deals provide any value-added locality services on top of the wifi access layer, but in any case, a step in the right direction.]

I’ve looked at the via82xx ALSA source code, and 48000 is hardwired all the way through it. Haven’t had the time yet to dig out the chipset documentation to see if I can hack it. In the meantime, I’ve managed a massive improvement in top-end quality with S/PDIF output by moving the resampling responsibilities from the chipset (which seems to offer only a very poor linear interpolation, hence the ringing) back into mplayer. With the command line

mplayer -ao alsa9 -af resample=48000:0:1

mplayer upsamples using an integer-based polyphase processor, which gives much cleaner output for complex audio. There’s also an option to do polyphase transcoding using floating point math (replace the ‘1′ above with a ‘2′, see the documentation), but on the Epia at least, this generates a playback error in mplayer. But in any case, mplayer is doing a much better job than the hardware, and that will have to do for now. Oh and for completeness, I should mention that the basic configuration is from the ALSA section of VIA’s Application Notes, which seem accurate. After install, you just need to use alsamixer to unmute the EIC output, and set the EIC level to zero.