From Michael Robertson of Lindows.com via The Doc Searls Weblog is another rant about lack of support for Linux in WiFi hardware. Is Intel's Centrino Techno-Latin for "No Linux"? ... Most worrisome is Intel's lack of Linux support for their new Centrino chipset which they've called their "most important announcement since the Pentium." Intel says that 300 million dollars will go into advertising this new product for mobile computing, but Intel isn't making the small investment to provide Linux drivers. [from: JB Wifi]

Locustworld - Breaking down barriers?

Disclaimer: I have nothing to do with Locustworld. I'm merely an interested observer.

The more I discover and read about Locustworld, the more interested I become. What they are trying to do is breaking new ground in all sorts of fields. It's also exposing the holes and issues in several technical areas that we currently take for granted.

What are they doing
On the surface, Locustworld appears to be a WiFi hardware manufacturer selling WiFi access points that are capable of forming a self-building and healing mesh. Each box can provide local coverage to WiFi devices. Each box may have a connection to the internet to provide a gateway to the devices it serves. Each box also seeks out other boxes and builds links and routes to them. The end result is a growing mesh of WiFi coverage that has multiple routes to the Internet accessible from any WiFi device within the mesh.

How are they doing it
Locustworld is actually not a hardware manufacturer. At the core is a Linux distribution using a whole set of open source tools. This distribution has been tweaked and trimmed so that it will run with no hard disk and can be booted from a CD or flash memory "disk". This distribution is freely downloadable and the CDs can be used to boot any PC with the right hardware into a MeshAP.

The hardware Locustworld is selling is outsourced and they have two designs based on commodity pieces. The main part is a small ITX motherboard with commercial flash memory and a commercial WiFi card. it's got all the usual keyboard, mouse, monitor, USB ports you'd expect and with no hard disk or fans there are no moving parts.

Administration is possible by connecting a keyboard and monitor but the main route to administration is via a central website. This handles address allocation, software upgrades and setting definition.

So Locustworld isn't actually a hardware manufacturer. From a hardware point of view, they're actually a reseller. So are they a software company? Well all their software is open source and almost all of it comes from elsewhere.

So why is this interesting?
Lets take a few points and dig a little deeper.

Hardware
All the hardware is off the shelf, commercial, commodity hardware. It's just the packaging which is unique. This sets them apart from the other WiFi AP manufacturers who use proprietary hardware all the way through. Even when they buy chipsets from Broadcom or TI, the rest of the hardware is a custom build. The problem with this is that this forces the AP manufacturers to also be software companies. Something I'm sure they would rather not be. The result for the customer is that even though they use the same chipset in almost the same design, you can't run Avaya software on Buffalo hardware. And you have to cope with the inevitable bugs, firmware upgrades and general flakiness because the software development always lags the sales and marketing need to get the hardware to market.

Device Drivers
There are a number of efforts in the Linux community to develop device drivers for WiFi hardware. However "There are no 802.11g or 802.11a cards currently supported by Linux. Avoid buying any kit from Atheros, Broadcom or Texas Instruments. Atheros have even issued a cease and desist order to the open source project for their driver. ..." So we've got a major limitation here where the manufacturers want to retain control and they write for Windows first, for the Mac if you're lucky and Linux last. This doesn't bode well for Linux desktops if you can't get device drivers for the latest WiFi kit. This also points up the limitations in a business model where software is used to keep hardware proprietary. It's understandable, but it's the customer and market who ultimately lose out.

Speed to market
Intel is issuing press releases about mesh networking in the labs but isn't shipping anything. Meshnetworks is spending VC money on marketing but haven't shipped anything (I think. At least that's how it appears). Nokia were working on a mesh networking system but pulled out. Meanwhile by using existing hardware and software a tiny UK company is shipping product.

Automatic Upgrades
I had a discussion about all this and asked why the WiFi manufacturers didn't follow Locustworld's lead and ship commercial hardware with Linux based open software. The response was that end users didn't want to cope with manual upgrades. But how is that different from the current situation where you're locked into firmware and software upgrades from people who don't want to be software manufacturers? The Meshbox system is designed to download upgrades automatically which surely is a better solution?

Security
Let's say you buy a WiFi AP that has a built in firewall, gateway, NAT, DHCP like the current round of combo boxes. Now let's say that Microsoft come up with a new Messenger that needs specific ports open or a specific MS product like UPnP. Will your AP supplier keep up with this? Let's further say that someone discovers a hole in a common system (like a Linksys router that has an easily broken password for instance), how well will your AP manufacturer deal with this? There's a very strong argument for either using open source software or software from a major manufacturer (like Cisco or MS) for anything that is used as a security gateway. Otherwise, who knows how long a fix will take to appear.

Naming and numbering schemes
Picture the scene. You fire up a MeshAP. So does your neighbour and you link them. You agree an IP numbering system between you. Later, another neighbour starts a MeshAP and uses their own numbering system. You find out about each other and link the nets and suddenly all your numbering has to be done again. So what's being built here is a walled garden part of the internet run by the end users. Unfortunately, the existing IP registries aren't set up to cope with end user addresses. As the Wiana
Faq says,

Current Internet address assignments are based on a hierarchical structure.

The existing Internet registries are not set up to deal with networks where every part of the network is owned and operated by end users.

A large scale ad-hoc wireless network is a different physical medium to the classic Internet

Existing Internet "trust" is achieved out of band, there is no per address authentication.

Significant fees are required to obtain IP address ranges.

"Applications for wireless address space were refused and the structure of these registries does not facilitate addresses for end users. Even with IPv6, allocations are described as "limited"."

Now there's plenty of other current problems with IP addressing and NATed gateways. On one level we really want to have every internet connected device to be uniquely addressable. Particularly for technologies like Voice over IP. This also brings with it huge security and safety issues because we're currently relying on firewalls to protect these end devices. IPv6 was supposed to solve all the naming problem, but the upgrade path is awkward and there's still hardly any support out there from the existing infrastructure. Like DNS, this needs to be decentralised outwards, but is currently mired in political and commercial interests.

Locustworld's solution to all this is to set themselves up as a registry for the 1.*.*.* range mostly because they needed something and there wasn't anything else out there. But this is already causing conflict with other groups who need to see more formal arrangements and are concerned about Locustworld's longevity and status. It's also distinctly sub-optimal as a long term global plan.

Business Plan
How do you bootstrap a hardware company without using large quantities of VC? Simple, don't produce hardware, produce software. Locustworld produced a hardware spec, but they have no hand in the hardware production except as a reseller. Ok. So how do you build a software company that sells free, open source, software? Simple again, you don't sell software, you sell support. So is there anything here at all? Well a tiny company is shipping product and changing the world. If they can make a living from this then do they actually need 1999 style capital?

Broadband provision
Again, picture the situation where numerous private individuals start installing Meshboxes and linking them together. Some, but not all of them will add in a cable or ADSL link to the internet. Right now this is almost certainly breaking the T&Cs of their broadband provider. In the labs, Locustworld have got aggregation working so that a mesh with 3 ADSL lines could provide a burstable rate of 3* ADSL speed to any device in the mesh. From the broadband ISPs point of view they are potentially losing customers and being hit with more bandwidth demand. But look at it another way. They could install MeshBoxes themselves as a last mile or long mile solution and achieve cheaper rollout and provision while also getting coverage outside their current areas.

This particular battle is just starting as the Telcos get increasingly concerned by WiFi usage at the edges of their network. Locustworld makes all the arguments harder. If we now factor in the hotspot providers as well and note that they are frequently the same telcos, we have another problem. To take an example, Meshboxes could be used to extend the Bryant Park free WiFi up through Manhattan and in the process, cover the same area as all the Starbucks and McDonalds hotspots that have recently been announced.

Is this real?
Reading the mailing lists, and observing, there's no doubt that Locustworld have a fascinating product. There's also no doubt that it's not quite ready for primetime yet. I'm concerned that the very small team may get overstretched

I've just seen this on the MeshAP mailing list.

Don't bother looking at any cards unless there is Linux support.

If there isn't, it is because the manufacturer refuses to release the specifications and should be pressured in to doing so.

There are no 802.11g or 802.11a cards currently supported by Linux.
...
Avoid buying any kit from Atheros, Broadcom or Texas Instruments. Atheros have even issued a cease and desist order to the open source project for their driver. ...

Feel free to flame the manufacturers, vendors and chip producers. The more sales they lose as a result of this policy, the better.

Atmel is Linux friendly and has good driver support. Lucent and Prism2 also have good support.

So what we have here is the most innovative uses of WiFi being blocked by the hardware manufacturers. We shouldn't expect the hardware manufacturers to be good at software and it's quite hard to see why they should even want to be in the software business. But deliberately blocking software development and refusing to release specs that allow software development seems fairly stupid.

I had a discussion with Glenn about some of this and suggested that it would make sense for the AP manufacturers to base their AP boxes on commodity PC hardware and then to outsource all the software development to the Open Source movement. He felt that end consumers wouldn't want this as they wouldn't want to have to cope with software upgrades. But frankly the current situation of repeated firmware upgrades to deal with bugs is just as bad. [from: JB Wifi]

Tim Woolford writes. TotalTelecom - Industry Viewpoint : 2003 - Is this 'The Year' for the UK WiFi industry?

In December 2002 I wrote about London WiFi hotspots and highlighted two issues: lack of demand and the need for some form of roaming between hotspot networks.

...

However, of greater importance to the technical and service aspects of hotspot deployment will be the continued development of innovative, realistic and ultimately sustainable revenue models. 2003 could be the year WiFi comes of age in the UK. [from: JB Wifi]

This blog is coming from Harrowgate via a mesh of Locustworld wifi access points linked via a wireless link to a van in the car park with a satellite internet connection. One of the meshAPs is a punters laptop with a USB WiFi nic booted off a locustworld CD.

This is part of a workshop at the Hardcore Wireless conference. [from: JB Wifi]

Today's hotspot mystery shopper saga comes to you from Benugo, 116 St John St, EC2. The deal is that you buy something and they then give you a access code which gives you 30 minutes of free access. So I buy today's special and sit down to try and make it work. DHCP comes straight up. I can ping the gateway but nothing else. I fire up a browser and that's dead as well. After 10 minutes of footling around I try IE6 and of course I'm taken straight to captive login page. So if you use a Mac, I guess you're stuffed. Anyway, the first access code has apparently expired so I get another one, put it in and I'm on. There's a little

Once I'm signed on, there's a little window showing the minutes left.

The adsl speed test reports
Downstream 473 Kbps (59.1 KB/sec) 510 Kbps (inc. overheads)
Upstream 241 Kbps (30.1 KB/sec) 260 Kbps (inc. overheads)

Even though I'm apparently enabled, Phoenix still doesn't work.

They've got some mains sockets so you can get power. And there's small posters on the wall to tell you they have access available. I get the impression that I'm not that unusual, but that nobody would be able to help me get going. Certainly there's nobody else with a laptop here.

And with 8 minutes left I'm now rushing to finish what I'm doing.
[from: JB Wifi]

I think this article at Techdirt Wireless nicely sums up the state of GPRS vs WiFi and the traveller. Here we have a tech savvy, American in London on business with a box full of gadgets trying to get Internet access. In theory, this represents one of the prime markets for paid WiFi hotspots along with 3G. So how does it go.

- First the Hiptop doesn't work with EU GSM as it's a US GSM only device.
- Then the T-Mobile GPRS works with a tri band Sony-Ericsson phone and it's bluetooth connected to the laptop. But at $15 per megabyte!
- Then the hotel room broadband connection didn't work. There's no mention of the price but we know these things tend to be on the expensive side.
- The few Starbucks with WiFi are on the opposite side of town (He doesn't say but presumably he's staying in the west end and the wifi enabled Starbucks are all in the City.)
- So it's down to the EasyInternetCafe. This has a SurfAndSip and appears to have WiFi, but the web page is not obvious. There should have been a pre-paid £5 for 24 hours option but there's no obvious way of paying and nobody in the EasyInternetCafe to ask.
- So then it's the usual 1.5hours for £2 on a terminal which works first time and is as painless as usual.

So here we've got the captive audience, several failures, and the real competition ends up being the nearest Internet cafe. So what this tells me is that there is a market for paid WiFi hotspots. But the price has to be as low as or less than a typical internet cafe because that's where the real competition lies. He also points out that this is a one off situation (that will get repeated) so until there's roaming across hotspot providers any sort of flat-rate monthly subscription doesn't work either.

And frankly all that applies to GPRS and 3G as well. IMHO per-megabyte pricing is just never going to fly. [from: JB Wifi]

What is the Xpertweb Talent Co-op ? [from: JB Ecademy]

The Internet Topic Exchange has just appeared on my radar. A very interesting experiment in using trackback to build channels of blog entries.

Pick a channel on the Exchange, drop the trackback URL into the trackback field on our Ecademy blog form and you can join in too. [from: JB Ecademy]

I want an Instant Message client that automatically goes into "busy" when I'm using certain applications. I'm not quite sure what the test should be, but I suspect that it shouldn't be that hard to be able to tell that I'm probably busy and don't want an interruption based on my current use of the machine. The problem is that even though there is a function in MSN to be able to say "offline" or "busy" I forget to use it. Most clients put up an "Away" message when you don't use the PC for a period of time, but while that's useful for the caller, it's actually useless to me. I guess what's really happening here is my own irritation with being interrupted which is why I don't much like telephones, mobiles, IM and other technologies that demand your attention just when you don't want to give it. It's also why I like Email so much because I can work at it when I want to and not when the caller wants me to. This also raises the question of why so very few IM systems have a store and forward mode. Why can't I send an IM message to someone who's off line, away or busy and them receive it when they next become live?

There's a lot of talk in this area about Presence and the ability of callers to see whether you're available on a whole range of systems not just IM. This could turn into a real tyranny of attention if we don't make sure that the systems make it easy to say "I'm unavailable". But unfortunately this is not high on the developer's list of priorities because it works against the need to encourage people to use the system for the system to be a success. But then I'm beginning to hate IM so much for it's interruptions that I'm seriously considering just not running it.

So there you go. I want to be able to say "Go away and leave me alone!" ;-) [from: JB Ecademy]

Kevin Sites Blog is coming from a CNN reporter on the front line in Iraq. [from: JB Ecademy]

Bruce Sterling and Derek Woodgate: Tomorrow Now So I'm surfing around, catching up on a day or two's news, and I come across some mega coverage of the SXSW conference via BoingBoing. I've read everything Bruce Sterling has ever done so this caught my eye. Then I noticed that the author and owner of the blog is one Heath Row, who just happens to be an Ecademy Power Networker!

If you got some serious time to spend on this, there's way too much content about way too many interesting topics from way too many interesting people. Great job, Heath. [from: JB Ecademy]

Yes, it's another episode of Get your war on

- Listen: They're called French Fries, they're greasy and they taste great with mayonnaise. F***ing dealt with it.

- Mayonnaise? Holy s***?, are you British? [from: JB Ecademy]

Is anyone going to the London WiFi Meetup tonight? If so, I may see you there.

Jerusalem, 33-34 Rathbone Place, London, 7pm [from: JB Wifi]

The inevitable counterpoint to World of Ends. EmptyBottle.org: World of Assholes

Essay question: Compare and contrast, World of Ends,

1. The Internet isn't complicated
2. The Internet isn't a thing. It's an agreement.
3. The Internet is stupid.
4. Adding value to the Internet lowers its value.
5. All the Internet's value grows on its edges.
6. Money moves to the suburbs.
7. The end of the world? Nah, the world of ends.
8. The Internet’s three virtues:
a. No one owns it
b. Everyone can use it
c. Anyone can improve it
9. If the Internet is so simple, why have so many been so boneheaded about it?
10. Some mistakes we can stop making already

with World of Assholes

1. The Internet is complicated
2. The Internet isn't a thing or an agreement : it's a place.
3. The Internet isn't stupid, but it's filled with stupidity.
4. Adding value to the Internet adds to its value.
5. Value on the internet goes unnoticed unless some high-traffic node connects it to the mainstream.
6. Money moves to the greedy.
7. The asshole of the world? Nah, the world of assholes.
8. The Internet’s three vices:
a. Americans dominate it
b. The wealthy populate it
c. More inhabitants does not automatically mean more value, except to those who want to sell you something
9. If the Internet is so complicated, why do so many seem driven to try and simplify it?
10. Some mistakes we can stop making already.

And apologies for the bad words. Heh, I'm only quoting here, ok! [from: JB Ecademy]

WiFi Ecademy DailEnews
Which led me to this.
Reiter's Wireless Data Web Log
Which led me to this.
Werblog
Which led me to this.
Japan Media Review -- A New Set of Social Rules for a Newly Wireless Society
Which led me to this.
Keitai Log

It's a smart mob world. [from: JB Wifi]

The Doc and David Weinberger have produced another great manifesto. World of Ends: What the Internet Is and How to Stop Mistaking It for Something Else. This should be required reading for any corporate type person with responsibility for "Internet Strategy".
[from: JB Ecademy]

Following my success in getting free wifi in a Starbucks in Faringdon road, I got this today.

Dear HotSpot user,

Yesterday marked the announcement of our plans to launch the UK HotSpot service. By the end of May, we will have an impressive 56 hotspots in Starbucks coffee shops across the UK.

This announcement follows the successful pilot programme in six Starbucks stores in London and Birmingham. Since the launch in August 2002, nearly 1000 of you have trialed the service.

These hotspot areas mean that you can relax with a coffee and, using a 'Wi-Fi' enabled laptop or PDA, can easily check your work e-mail, surf the internet or download multi-media presentations, without the hassle of a docking station or bulky cables.

This scheme has already worked incredibly well in the US, with over 2,000 HotSpots already in place. Discussion with airports and hotel chains means that the concept should be equally as successful in the UK, giving a real opportunity to showcase our service.

To ensure maximum flexibility and choice for all, from 31 March, four pre pay tariffs will be introduced, ranging from £5.50 for an hourly pass to £47.00 for a month pass.

More detail about the new T-Mobile HotSpot venues will be updated on www.t-mobile.co.uk/hotspot shortly.
Regards

Webmaster (JR)
T-Mobile UK

£5.50 for an hour is on the edge of what I'd consider acceptable. [from: JB Wifi]

The Register : Swisscom buys Megabeam and WLAN AG

Swiss telecoms operator Swisscom today announced the acquisition of WLAN hotspot firms Megabeam and WLAN AG. Financial terms of the deals were not disclosed.

I hope Megabeam got a reasonable price. That's a pretty fast turn round for the VCs involved! [from: JB Wifi]

Here are some words as I try and type without making any typing errors. [from: JB Ecademy]