Alde’s Stuff

Anthony Perez over at Amped News writes an opinion piece that the “Death of the PS3″ is a “Hasty Conclusion.” I have to agree with him on that one.

From his article:

“Last year, at NotifyWire.com, we tracked several one thousand dollar Xbox 360 bundles which would sell out in less than two minutes after becoming available,” said Ian Drake, president of NotifyWire.com, an online retail tracker that sends out “In Stock” notices via e-mail and newsletters. “Just today, a one thousand dollar PS3 bundle was in stock at eToys.com for over 12 hours. This sort of thing didn’t happen with the Xbox 360 until well after Christmas.”

An overpriced bundle not selling out for 12 hours, how SLOW… The only reason the companies come out with these bundles is to keep the Ebayers from flipping the boxes. Very few people want to pay over $1000 for a 60gig PS3 with 5 Games of which they may only want one. Yes, you can return all the games to the store, that is, if the store allows you to only return the games, or only return the bundle.

Also, keep in mind that the Xbox360 was selling to an empty market; there was no other “Next Generation” console game system out there. PS3 has to compete not only with the Xbox360, but with the Wii. And, uh, the Wii goes for $250 retail. No where does anyone EVER (until I read the Toronto Star article cited below) mention the price differential between the products. When the anti-ebay Wii bundle is $550, guess which one the non-traditional gamer is going to buy first.
Also limiting them is the, well, lack of hardware. The Times of London states:

“The big challenge for Sony is that PS3s are so hard to get,” said Baker. “To a large degree, many people are buying something else. Sony is going to have to battle to retain the market share that they’ve had historically. They’ve given a two Christmas advantage to Microsoft, and a one Christmas advantage to Nintendo.”

Sony also dug themselves a big hole with no real premier launch title out there. But then again, Sony didn’t do well with the PS2 either. PS2, I was amped for NHL2001 & SSX, but it wasn’t until March that I got it with Gran Turismo 3. And I remember even then having to sit and hit reload on Amazon.com over and over again. I still think it will be that way in March. People like me sitting on the sidelines waiting for Gran Turismo HD, or Final Fantasy XIII. And because of the price, we’re all more than willing to wait, too. Especially with Final Fantasy XII just recently out for the PS2.

The Toronto star says:

While holiday sales totals are the subject of much speculation on the web, the research firm NDP Group estimated that Wii’s sales figures for November were more than double those of its technologically beefier competition. According to the firm, Wii sold 476,000 units, while the PS3’s figures were put at 197,000.

…

“I think the Wii still has not proven that (the motion-sensitive controller) is anything beyond a gimmick,” Crecente said.

“The PS3 still has a lot of work ahead of it. It still has to make a game that’s not derivative of what’s already out there, or a sequel of something that’s already been produced.”

Indeed, it will be interesting. And also mind you this is going on with a background that Intel/AMD/NVidia/ATI are pushing the PC gaming envelope even further.

I was reading an article over at Ars Technica about AT&T and their issue getting Fiber and Television service to customers.
See, AT&T wasn’t going to spend a lot of money on their broadband. They never have. What they intended to do to compete with Verizon’s FIOS was to use the existing fiber they had planned to put out to their neighborhoods and then leverage the existing copper to the house and use VDSL instead to achieve fast speeds. This is why it’s taken them forever to finally roll this out. It wasn’t until Feb 2006 that VDSL as we know it was standardized. Previous to that it was an Infineon (Israeli company) based chipset that Broadcom also licensed. In essence it was 10Base-S, but because the IEEE 10Base Committee was disbanded in 1997, they would not re-seat to approve a new technology. ADSL hardware providers used this against Infineon to sell against them, claiming that why should someone use a non-certified technology. The true benefit of 10Base-S was that it provided 15mbps of symmetric throughput at short (less than 3000 feet) loop lengths. Perfect for the Hotel Market. The other benefit of 10BaseS was that it had built in CPE device feedback to the head end; it could tell if there was remote link, what speed that link was at, and a true remote report of RSSI and interference. In that sense, just like Betamax was to VHS, 10Base-S was a far superior technology than ADSL.

Today’s VDSL uses tried and true DMT encoding on the line, rather than QAM. Telco’s are much more friendly to DMT. It does lose the ability to do real remote troubleshooting, I don’t think that Ikanos had put forward in-line CPE reporting. VDSL (or VDSL2) has been tested to do over 100mbps Symmetrical and even 200mbps Symmetrical at short loop lengths. It’s not like having a real Multi-Mode fiber terminate at your house, but it allows for AT&T to save money and time in deployment.

But yet again, the Internet’s Village Idiot, Ed Whitaker, wants to hijack the process that unfortunately every other Telco has to go thru. People here in New Jersey bemoan the fact that they don’t have FIOS yet in their area. What they don’t know is the same reason their taxes are so high is the problem with FIOS. New Jersey requires the telcos (and before them, the cable companies) to negotiate with every township, rather than allow them to negotiate at a much higher (county or region) level. New Jersey has over 560 seperate townships. (A reason taxes are so high is that with 560+ townships, there is an incredible amount of redundancy of services; seperate Police/Fire/Administration for almost each and every township = lots of redundant government workers).

I can understand AT&T’s frustration, but welcome to the game Ed, the cable companies have had to do this before you, and I don’t think it’s fair to allow you to jump to the head of the line.

The second post in the series about WiFi Free Lunches, this one about Airports.

I’ve done site surveys and have installed airports in the past. It’s a bigger mess than the hotels.

The first problem is that all airports tend to be run by the government of some sort. It’s usually the city that runs them, but some are county affairs, and some even run by the state. I’m unsure how Airports outside the United States are run, they may even be controlled by the Country.

Why is this a problem? The people that run Airports do so as Feudal Lords. They are the lord of the keep, and the passengers are their vassals. No, I’m not kidding. When I’m done with my current job, I’d love to go run the network infrastructure at one of these large airports. The side benefits are astounding, and the amount of money I’d have direct control over boggles the mind.

These Airport managers believe that they’re doing you a favour by allowing you to sell something to their passengers. For example, an airport that was insanely close to a hotspot of technology didn’t have WiFi for the longest time (up until 2004 if I recall correctly). I was there for the site survey and contractor bidding (we had won the contract to provide WiFi only after we said no the first time, and it was awarded to someone who went out of business after being awarded it). We walked around the entirety of the airport with an AP and antenna on a pole (This was pre-911). We came up with a final number to bill the airport. It was something like 340k or so. That’s running a lot of cable, dealing with union contractors and very little profit built in.

The airport’s response: You pay for the installation and hardware (all 340k), you pay us $2k a month to provide WiFi to the people in the airport, you provide us with a revenue share of each use, and any contract that involves roaming or a third party has to be viewed and approved by the airport. Oh, there’s no free advertising either; you need to talk to our advertising branch of the airport to negotiate paying them to advertise. (Those fees would have been 3 times the amount of the installation).

That’s why you pay for access in an airport.

Airports like Las Vegas are few and far in between.

Every time I read something about some company “winning” an airport bid, I laugh. Unless the airport is a small regional one (Like Atlantic City, or Richmond Va), the company that “won” basically had to buy the business. Ask Logan International (Boston). I did a site survey and a AP plan for there. They were truly ahead of their time because what they wanted was to be THE internet provider for their site, and then resell management of that bandwidth to it’s tenants. But the FCC has stated that they can’t stop anyone from putting up their own WiFi APs if they want (Tmobile and American Airlines).

All the big Airports: O’hare, Atlanta, San Francisco, JFK, Newark, Denver, Houston, Dallas, Dulles. None of them are free. It won’t be free there, no matter what Jim Thompson wishes. It’s much the difference between American/Continental/Delta and Jet Blue/Southwest. As long as the business model requires WiFi vendors to pay outrageous fortunes to have presence in an Airport, expect to pay for WiFi. If the Airport can make a buck at doing it, expect them to make that buck and charge 2 more.

Next I’ll discuss Muni-WiFi and why it’s so overhyped and is so underperforming.

Check out the site Wii Damage which has set itself out to document the problems that the Wii controller has done in the name of fun and games.



People have broken windows, pint glasses, LCD TVs, even PDAs. Please make sure you have a firm grip on the controller when you’re playing Tennis…

Over at Ironic Sans, he had a great idea: That someone in Tokyo should build a building shaped like Godzilla.

Go and read, it’s a great idea. I think the Megatokyo crowd would heartily approve, nothing better to remind people of the zany monster movies than a permanent homage in architecture.

The folk over at Distributed.net, have been working on a preliminary client for the PS3, and since they’ve sorta semi-publicly acknowledged this (I have inside information I was sitting on), I’ll describe what they’ve gotten themselves into.

The client is currently being worked on, and have an assembly-optimized RC5-72 SPE cores that are running at approximately 24 Million Keys per Second.

Now, according to their Speed Database, an Athlon 64 4000+ gets appx 10Mkps, while a Core2Duo 2.6Ghz get appx 9.6Mkps.

Now, I’m sure you’ve read my post about PS3 datacenters, especially the point where I state they have 6 user-available SPEs. That 24 million kps is per SPE, so 6 * 24 = 144Mkps on the SPE’s alone. I’d be afraid to see what this thing will do with DES. Or for matter of fact, Triple-DES. I’ll remind you that normal Triple-DES encryption uses only 2 keys, not 3. Triple-DES encrypts the plaintext with key #1, decrypts the cryptotext with key #2, then re-encrypts that output with key #1. Therefore you can easily put the two DES keys into the 128 bit registers… Voooom….

There’s still some issues to solve, and an optimized PPE core must be written, since the current PowerPC cores on the PPE is very abysmal. Also, the client is written for PS3 Linux, and that a native PS3 port would require a PS3 dev kit that is currently unavailable to the dev team. And the lead on the project lives in a country where the PS3’s currently cost $3000 US.

So, if you’d like to donate one or know someone at Sony that could help as well, Distributed.net’s donation page is http://www.distributed.net/donation.php.

There’s no Xbox 360 client under development, well, mainly, because Microsoft doesn’t want Linux on the box (though Free60 is trying their best), so conversely, if you know someone who’d be willing to donate a Xbox360 development kit, that link above works just as well.

Stumbled across a blog post in which the author asked:

Why isn’t airport and hotel Internet access a standard free feature?

His conclusion:

The Internet is a series of tubes and is here to stay - therefore it should be looked at in the same light as television service since for many travelers Internet access is a necessity.

Seriously. He blames that people are stupid, like Sen. Ted Stevens, and thats why it’s not free.

I also hit Andy Abramson’s blog where he pines about hotspot use and that:

1. Speeds being horrible

2. Latency being a problem

3. NAT Traversal being an issue due to the way access points and routers are networked.

Now, I worked for a company that sold internet access in Hotels, Airports and Retail locations. Odd are good that if you’ve ever used a WiFi hotspot, you used one that ran across my network. So having been on the inside, I’ve got some news for both Andy and Frank: Bandwidth isn’t free. Or more vaguely: There Ain’t No Such Thing As A Free Lunch.

Now, this part is gonna get really technical, so if you don’t want an insiders view of the Hotel WiFi industry, move along…

Large hotels may have an IT staff. These people aren’t much IT as Remote Hands; they call corporate and do what corporate tells them to. Smaller hotels may have an IT person. Usually this person is the same person that maintains the Plumbing, Air Conditioning and Electric, i.e. the Hotel Engineer.

So therefore most hotels can’t afford to do the internet themselves, so they contract with someone like iBahn, GuestTek, Wayport, or whatever numerous other providers there are out there. Depending on how much the hotel wants to spend, will dictate what price is passed on to the consumer.

Average hotel rates for wired and wireless in an entire hotel are $50 to $250 per room installed. That involves the wired rooms using either existing wiring (Cat5 is EXTREMELY rare in a hotel. CAT3 is very rare. Usually it’s unrated twisted pair) or pulling new cabling. Even if you didn’t do wired, you still have to run wire or come up with some solution (ADSL/VDSL is $1100 for a head end and $20-50 per brick). So that 103 room Holiday Inn Express with Wired with Wireless overlay would be about 18k for the total install. Quite less for Wireless, perhaps 6-10k all installed.

The underlying problem here is that hoteliers want the contractors to pay that 6-10k. And then they’ll want the contractor to pay the hotel a monthly revenue share for the luxury of providing their users internet access. (Don’t laugh, this happened up until 2005). That’s the mindset you’re dealing with here.

The typical hotel contracts go something like this: Hotel pays for all capex and owns the hardware outright and the contractor manages it for them: Hotel makes the most money of the revenue share. Or, Contractor pays for all the capex: Hotel makes the least amount of money from the revenue share. Or Hotel wants to give it away: Hotel pays a fixed rate per room per night.

So why do hotels make you pay for access? They like the unlimited upside of the revenue share model. Plus, they don’t have to raise their rates in comparison to their competitors to cover the capex/monthly fees.

I will also remind you at this point that DSL has no service level agreement. For example: AT&T has a 48 hour ticket pick up window. That means their DSL technicians have 48 HOURS after you enter in your issue to just PICK UP the ticket. And there is no guarantee of time to repair. Therefore real hotels use T1s and Leased Line circuits with SLAs. And depending on where the hotel is, you could have a $300 a month T1 to a $15,000 a month T1 (Just ask any hotel on the beautiful islands of the Bahamas how much a T1 costs). Average is somewhere near $700. A Month. Before anyone’s bought anything.

So, if a hotel is paying, oh $700 for a T1, and paid $18,000 for all the hardware, and have a revenue share model in which they recoup $6 of that $10 in fees, it will take 3000 uses to recoup the hardware costs total, and 117 uses to pay for the T1 per month. So, if that’s a 3 year contract (industry standard), that’s only 190 uses a month over 36 months to break even. Which isn’t too tough for most hotels. And as the contractor, you make $760 per month on the break even hotel use.

This, of course, is fine and dandy, until the moment someone picks up a phone to call for help. This is MAIN reason hotels contract the service out. Take a hotel chain like Marriott. With hotels of 500+ rooms, if you had a meager (and I mean REALLY SMALL) take rate of 10%, with an average heads on beds rate of 50%, you’re talking 50 users per night. Out of those, 10% will call for help, so you can expect 5 calls from that hotel per night. These cost the contractor between $5 (if they farm the callcenter out to India), or $20 (America) per call (based upon some concocted callcenter rate of $/hr for the agent, plus opportunity cost, etc). So, as the contractor, you need to clear that amount to generate profit (in this example, $100). So, with that 50 uses, if you get $4, that nets you $100 in profit per day if you keep all your calls in the states.

But should the hotel give internet access away, the contractor would instead charge a flat rate fee, plus per call charge. Now the hotel is losing money, not making it. It’s gone from being a Revenue Center to a Cost Center. You can derive no new profit from having WiFi. Oh, and with free WiFi, take rates tend to skyrocket to 70%. So, 350 users instead of 50. And if 10% have problems, that’s 35 calls instead of 5. Is this starting to make sense? And if I’m giving it away, you can bet the contractor is charging per call and a base MRC. Your downside is infinite, your upside is nil.

I’d short serve the industry here too if I didn’t mention the Legal aspect of being an Internet Service Provider (Because if you’re a hotel and you provide Internet Access, You’re now an ISP). When Gill Sperlin of Titan Media sends you a DMCA takedown notice for Woodsman Gay Male Porn movie sharing. Who is sharing, and how do you stop them? Or you receive an FBI Subpoena for one of your users accused of child pornography? Or Credit Card Fraud?

I’ve stayed at hotels where internet access is free. And they all suffered from the same issues that Andy stated. At night, when the hotel filled up, I couldn’t even listen to 96kbps Shoutcast streams. I’ve been on free WiFi that my signal strength was -85dbm, just barely good enough to maintain a 1meg signal as long as someone didn’t walk down the hallway. I’ve gone to a Panera bread store, but couldn’t get my email because the Horde server I was connected to was blacklisted because it had the word “chips” in the domain name, and therefore was banned for being a gambling site.

In summary, Hotel internet access is not free for 3 main reasons:

1. Hoteliers are the third cheapest bastards in the world (after Airports and Internet Leechers)
2. Bandwidth is not free (even if the government is paying for your city to have a mesh wireless network hooked up to the internet, some taxpayers somewhere are paying for your access)
3. Network hardware isn’t free. Silicon and gold don’t just fall onto PCBs and get magically soldered together. Copper’s natural state is not to string itself into 4 pairs of small gauge wire.
4. Legal compliance costs for law enforcement inquiries and copyright infringement aren’t cheap either.

Hrm, this is rather long. I’ll post about Airports at a later date.

Added content in response to Andy’s comment:

Ah, yea, I seemed to kinda dodge where I was going with your point.

Starwood… Ah that one was great. So they go and put out a bid to all the typical providers, and then the IT director decides he’s going to do it himself with Cisco’s help. 12 months later, with the wonderfulness of Cisco BBSM software (which they bought back from CAIS/Ardent when CAIS couldn’t pay their bills anymore and Cisco was caught short having given them millions of dollars in gear), there were 7 installed Starwoods. 12 months to deply 7 hotels. Needless to say that solution got punted and they went back to the standard group of providers.

What you also have to understand is how branding in a hotel environment works. Most hotels out there with a brand on them are not owned or managed by the company whose brand they fly. A hotel gets built but an independant management company, and all they do is license the name for a fee. This means they buy all their linen, towels, soaps, literature, etc, from that brand. The entire running and control of that hotel is independant (to some degree) from the brand. Just because it’s got a Marriott logo on the outside doesn’t mean that it’s owned by Marriott.

So, in the hotel space, you’ve got GuestTek as the #1 provider if you rate by the number of rooms installed (Oh, did you know that the #1 provider in the space just returned to positive EBITDA on $34 million in revenue? Small Small Small companies.) Somewhere up there too you have Wayport, iBahn (aka STSN), Lodgenet, and other smaller or even vendor provided (companies like SMC and the like with a stand alone firewall).

I know that Wayport and iBahn typically install T1’s to a property. Lodgenet and others typically install DSL (Yes, Lodgenet prefers to deliver the hotel DVD’s with the movie data rather than transfer the movies over the internet). With the price of point to point T1’s coming down because of the glut of fiber after the dot-com demise, it means that these T1’s are typically a lot less oversubscribed than DSL.

Having been in the ISP business now for 12 years, I’ve seen all sorts of customer access; from dialup to X.25, OC48s to the first GigE product sold at UUNET, oversubscription is the key to the quality of service. Consider your standard Tier-1 Provider (Level3, Verizon Business, AT&T, Qwest, etc). These companies tend to have multiple OC192 backbone networks with OC48 peering handoffs. Even if you did have a 10 to 1 T1 oversubscription rate in a city, you’d hardly notice.

Now, go further down the food chain to your DSL providers, like a Verizon, Megapath, Qwest, Bellsouth, Covad, etc. Here you probably see 100 to 1 oversubscription.

For example, if you buy wholesale DSL from Verizon, you typically bring to them an ATM circuit (DS3, OC3, OC12), and then have to pay per megabit of traffic into each LATA that Verizon serves. So that’s say 100 or so LATAs onto a single or multiple circuits. So if you’re on someone’s DSL line (or even Cable internet, you’d be surprised at the number of Hotels that do a bundle deal with Time Warner & Comcast), you’re not only sharing bandwidth with the people in your hotel, you’re also sharing the LATA’s bandwidth with all the other locations in that LATA.

Oh I forgot about loss of revenue for the hotels. Hotels are really upset with the loss of telephone revenue due to the proliferation of Cell Phones in the 90’s. They construct their buildings to inhibit your cellular coverage. No, it’s not that your cellular service sucks so bad you can only get it right up against the window, it’s meant to drive you to use the in-house PBX. Airports got over this by contracting to someone to provide all the GSM/CDMA access in the Airport, and then get a cut of each call being made. Hotels haven’t yet figured this out.

Since you can easily tell what is SIP and then what RTP is keyed off the SIP interaction, companies like Packeteer, Cisco and Juniper all have easy identification of VOIP streams, and then you either partition that bandwidth (like you can in a Packetshaper), or put the QOS all the way down to “Best Effort”.

And lastly, I forgot what the typical user does when they get into the hotel: Download porn. They fire up Kazaa, Morpheus, Gnutella, Bittorrent, you name it.

That’s the recipe for bad VOIP calls. Clogged pipes because of your fellow traveler, because it competes with the hotel’s phone revenue, because cheapest wins and cheap is oversubscribed DSL, and undermaintained or poorly planned installs get internet access at hotels to where you’d rather just stay home.