Software

Declassified

Which is a better real estate investment, luxury apartment buildings or self-storage locations filled with unused exercise bikes and old copies of National Geographic? I can make a strong argument that self-storage is better -- far better. Those storage places charge about the same rent per square foot as very nice apartments in the same city, yet they typically aren't located in very nice parts of the city. Storage units are smaller so there can be more of them, reducing the financial impact of tenant flux, if there were any, but there generally isn't because we all hang onto our junk and need someplace to store it. There are no residents to complain, no toilets to fix late at night, no heat or air conditioning, almost no utilities at all. If the renter fails to pay his rent, YOU HAVE ALL HIS STUFF and he has no tenant rights. And after you've owned the place for 20 years, making a good living for every one of those years, someone comes along to buy it for many times more than you paid because by that time maybe it IS in a very nice part of the city and they want to build an office building or -- just maybe -- luxury apartments.

Self-storage is a fabulous real estate investment, but people don't get that because it isn't sexy. And in exactly the same way, most people don't understand that the most profitable section of their local newspaper -- the section that probably makes the difference between success and failure for the paper as a whole -- is the classified ad section. Lost pets, apartments for rent, help wanted, and cars, lots of cars, are the financial backbone of American newspaper journalism. Or were.

Daily newspapers are dying, we're told, killed by a generational shift to the Internet. In a big-picture kind of way this is true, but if we drill down, the details are startling. Look at Google News, a compilation of popular news websites, most of which think of themselves as anything but parts of the Internet. Most Google News entries are from newspapers, magazines, and television stations, while far fewer are Internet-only products like this one. And since Google News links directly back to each of those sites, feeding them readers, if we want to blame any organization for the decline of newspapers, this time it can't be Google, nor any other Internet news site I can think of.

If not Google, then who? Craigslist.

I have worked at newspapers, and every department except for display and classified advertising is a cost center, not a profit center. Even the circulation department, which takes in money from subscriptions, NEVER takes in enough for the paper to break even, thus remaining a cost center. Of the two ad-related profit centers, classified ads aren't as sexy, but they are generally more profitable because there are no sales commissions involved and service and customer hand-holding is minimal, meaning little overhead. It is precisely the newspaper equivalent of those self-storage places -- modest but extremely lucrative.

Craigslist, as we all know, is classified advertising for the Internet. Started and still led by Craig Newmark, Craigslist is as close to non-profit as you can get for an Internet business that can also claim to be commercially successful, with busy operations in 450 cities and many countries. Only Craigslist doesn't actually exist in those 450 cities, just in a few computers in San Francisco. The business is low overhead and can survive quite well, thanks, by charging only for broker apartment listings in New York City and employment ads in only seven cities. All other ads are free.

The link between this column and Craigslist is tenuous but real: Jim Buckmaster, the current president and CEO of Craigslist, coded some of this site in the late 1990s, most specifically the late and unlamented (by me) reader forum.

Jim and Craig are proud of their operation and see little reason to change the way it works, which is what vexes not just newspaper classified ad departments, but ANY company that accepts money for advertising. Ebay, for example, doesn't like Craigslist any more than does the San Jose Mercury News, and eBay actually owns 25 percent of Craigslist through a private stock purchase that eBay hoped would give them important insight into how Craigslist succeeds.

That success couldn't be simpler: FREE has a mystique that is hard to fight. That makes Craigslist Linux to eBay's Windows.

The only real lesson to learn from Craigslist is, itself, anathema to eBay. What makes Craigslist special is that it is, for the most part, free. And the only way to compete with Craigslist is by making any competitive service free, too, which eBay is rightly unwilling to do.

The sad truth for eBay is that no matter how much it studies Craigslist, short of going free, itself, eBay will never compete with Craigslist for local users. There is no way to compete with free in a comparison of automated services. No way.

So Craigslist wins and eBay loses. But that's not all. Yahoo loses, too, as does every other site with paid advertising. This week's deal between Yahoo's HotJobs division and more than 150 daily newspapers is intended to look like competition for Craigslist, but it isn't. There is no fair way to compete with free.

Craigslist will always win when it comes to local classified advertising, simply because Craig and Jim's operation is big enough and popular enough and cheap enough that nobody else will ever be able to catch up. Local classified advertising on the Internet is Craigslist's opportunity to lose, not any other company's opportunity to win, because it can't be done.

This fatalistic view of mine leads to arguments, but I maintain it is valid and applies to more than just Craigslist. It also applies to YouTube, for example, and is the major reason why Google is buying YouTube -- because Google can't afford not to.

One view of Google buying YouTube says what Google bought is 40 million eyeballs, which assembled somewhere is harder to replicate than a bunch of streaming servers spitting out content. So YouTube reached a critical mass of eyeballs and was bought as an ad-serving platform, the story goes. It was cheaper for Google to buy YouTube than it would have been to throw a bunch of engineers in a room and simply copy YouTube, which would look to have been the other possible maneuver.

Only this isn't entirely the case. Google bought eyeballs, yes, but in YouTube they also bought a superior platform with social networking aspects that Google was probably not in a position to just copy. This is odd to explain, but YouTube's growth vector was such that Eric Schmidt had to have figured out on an empirical basis that Google could NEVER catch up. If you copy a feature but don't spend money on marketing (that's the Google habit) and the outfit you are copying is far enough ahead, you NEVER catch up because they will have eaten the audience before you get to it.

The video-sharing game was already over and Google knew it. So it wasn't just eyeballs they were buying, it was a seat at the table.

By the same token, no present competitor will ever beat Craigslist. It can't be done without cheating.

What this means for Yahoo and those daily newspapers struggling to survive and compete with Craigslist is that competing won't work. This is a war that can't be won, no matter how much money Google or Yahoo or Microsoft or eBay has. And those daily newspapers with their reporting staffs and international bureaus? In the long run and absent a successful new business model, they are doomed, because it comes down to location, location, location, and for that Craigslist can't be beat.

Keeping the Peace

The video player died in our minivan a few weeks ago, sending my kids into a death spiral of Wiggles withdrawal. In retrospect, it would probably have been better never to have the video player, but it seduced us with the promise of hypnotically transfixed children who would never again ask, "Are we there yet?" I had to have it back, only better. So I built a video server for the minivan.

What I wanted for the van was the equivalent of what we had already working at home, where MythTV running on a $300 Linux box stores every episode of Arthur, Dora the Explorer, Dragon Tales, Rolie Polie Olie, Cyberchase (my son Channing's favorite), Jimmy Neutron (my favorite), and many other programs -- more than 500 shows in all. It's squeaky-clean content in every sense, including the best of all -- no peanut butter fingerprints. I just had to find a way to move this capability to the car for almost no money.

My output device of choice was the Sony PlayStation Portable or PSP -- a handheld game system with a widescreen LCD that plays movies beautifully. We had one, thanks to a grandmother who had no idea that two-year-olds don't get such things for their birthdays, and when they do get them they try to flush them down the toilet.

Though not advertised as such, the PSP is EXTREMELY water-resistant.

Cars are hostile environments for computers. I wrote an entire column once on why we don't have hard drives in cars (it's in this week's links), so I knew that for a server I'd need a hardened, yet cheap, box, which I found in the Norhtec MicroClient Jr. from Thailand. Because of enlightened government computing policy, Thailand has the cheapest non-Microsoft PCs in the world and the MicroClient Jr. is among the least expensive. In volume it sells for $90, but I paid $120 plus an extra $70 for WiFi capability. I might have saved the $70 and used a USB WiFi adapter I had lying around, but the box has only one usable USB port (the front two were diasbled on my unit) and I wanted that for storage.

For $190 I had a diskless, fanless, completely silent PC with a Via processor and 128 megs of RAM. To this I added a copy of Puppy Linux, which is a very good lightweight distribution you can boot from a CD, though in the MicroClient Jr. I used a CompactFlash card from an old digital camera as the boot drive. For the data drive I used a huge four-gig USB flash drive that came from who knows where: I don't recall buying it, but it was sitting on the shelf.

This is not a very ambitious project, really. The MicroClient Jr. is a little smaller than a Mac Mini and can run on 12 volt DC, so I mounted it under the driver's seat, stealing power from the seat motors. The USB flash drive is about the size of a pack of cigarettes, if you remember what that looks like, and I used Velcro to stick it to the side of the MicroClient Jr. The little PC runs fine as a server, and there are many open source programs for transcoding almost any video into the H.264 or MPEG-4 formats preferred by the PSP. The PSP already has WiFi capability and the components are never more than four feet apart. Best of all, I was able to put 53 shows on the data drive.

Clearly this is a product someone should be making and selling, probably Sony. If we can get Grandma to pop for a second PSP or even a third, the server should be able to keep them all running just fine. Remember, no decoding takes place in the server, so only bits are being schlepped. And since I've had the system running, not one kid has asked "Are we there yet?"

When we shortly replace the old van, I'll take the server with me, of course, and put it in the next vehicle. I can see some real advantages to this solution, which is inconspicuous, doesn't block any mirrors, and generally keeps two hands per kid occupied, which limits my boys to kicking each other.

In other news, Microsoft's Zune is out and, as predicted, hardly anyone is taking it as a serious threat to Apple. But if it isn't a threat to Apple, then why do we keep writing about it? To a certain extent we do so because anything from Microsoft gets a lot of press, but some pundits are convinced there is another shoe to drop -- some deal or feature with which Microsoft will turn this sow's ear into a silk purse. I don't see this as likely FOR THE FIRST ZUNE VERSION. Apple has established a 6-9 month iPod product cycle. If Microsoft hopes to continue the trend of doing the job right the third time around, then the clock is ticking toward that time, which will be 12-18 months from now. THIS Zune is a placeholder meant to start the clock ticking at best. Two Zunes from now is when the game really begins.

Finally, faced with the prospect of an open source Java clone coming from the Apache Foundation, Sun Microsystems made the bold move this week of making most of its own Java system code open source. What's really surprising is that they put Java under the Free Software Foundation's General Public License, which is great news for developers but not immediately clear why it should be good news for Sun. Many companies that dabble in open source, like Sun, Apple and others, frequently tweak their licenses and rarely just throw the code under a blanket GPL. So why did Sun?

Because this isn't your father's GPL, that's why. Sun put Java under GPL v.2, which gives the original licensor some unique rights. Say you extend Java, under GPLv2 the way to give your improvements to the world is by giving them back to Sun. Or, if you'd like to keep those changes to yourself, it requires negotiating a non-GPL license with Sun, which means you'll have to PAY Sun to USE YOUR OWN CODE.

Under GPLv2 Sun benefits significantly more than it would have under the original GPL. Sun controls the code, it controls forking, and anyone who wants a special deal has to pay. For a product that was generally given away, anyway, going with GPLv2 will probably make Sun more money -- probably a LOT more money -- than the company would have made by keeping the source closed.

Now that Sun is losing less than $1 million per day, every penny counts you know.

Divide and Conquer

In the U.S. elections this week, a change of power took place in the Congress that had an almost instantaneous effect at the White House. Just a few days before, President Bush said of a potential Democratic takeover of Congress: "The terrorists win and America loses." This week his line has changed to "reconciliation" and "bipartisan effort." That's the way it is with dogma, which is heartfelt right up to the moment when it is no longer felt at all. That might be the best way, too, to understand Microsoft's recent deal with Novell and apparent embracing of Linux.

There was a time -- right up to a week ago -- when Microsoft appeared to feel, from the statements of Bill Gates and Steve Ballmer, that they could ridicule Linux out of corporate America or use a campaign of fear uncertainty and doubt (FUD) to undermine the open source operating system. There was nothing good about Linux, they said, and a lot that was bad. They even argued that paying Microsoft for Windows was less expensive than getting Linux for free. Yeah, right.

But Linux happens. And Linux isn't going away. As a server operating system, Linux is far more important than Windows and that trend isn't changing, something that Microsoft has finally acknowledged, not just through this agreement with Novell, but also through the PHP license Redmond also announced this week. But Microsoft is still Microsoft and has its own peculiar way of changing with the times as we see in this very interesting agreement.

What is actually going on here? For one thing, Microsoft seems to be paying Novell a lot of money, some of it for software licenses and some for patent licenses. Microsoft agrees not to sue Suse Linux customers for violating its patents (a sort of non-license license that is purported to gracefully cover the legal gaps between open- and close-source licensing) and the two companies will build an interoperability lab.

Nothing wrong so far. Novell gets money, exclusive access and co-promotion by Microsoft and Suse Linux gets a boost against Red Hat and the other distributions.

Then Ballmer, speaking a few days later with an Indian IT publication, offered to do the same sort of deal with ANY Linux distributor, even Red Hat, much to the surprise of Novell, which thought the term "exclusive" meant, well, exclusive. The ink was barely dry and Microsoft was apparently re-interpreting the terms of the agreement.

This was, of course, by design. The agreement with Novell IS exclusive and Microsoft has no intention of signing similar deals, but to suggest that it might has a dual effect of undermining ALL commercial Linux vendors. Novell customers and investors have to wonder if there is some fine print they don't know about, so Novell is undermined. Red Hat is certainly undermined, either by being suckered into negotiations that will go nowhere or, more likely, by not even approaching Microsoft since they know (but the market doesn't) that Ballmer's offer isn't for real. More Microsoft FUD.

Microsoft chose to do a deal with Novell, rather than with Red Hat or any other Linux distributor, for several important reasons. Suse Linux is most popular in Europe, where Microsoft has had tough going lately with European Union regulators. Appeasement of the EU by cutting a cross-licensing deal with its largest seller of open source operating system software couldn't hurt, and might even save a fine. Choosing Novell hurts Red Hat in the U.S., where Suse hasn't really been a major factor. Novell is perceived as weak where Red Hat is strong. And finally, Novell has some legal leverage over Microsoft in the form of an antitrust lawsuit inherited from WordPerfect. Maybe this deal will get Novell to back off on the legal front.

Microsoft is a marketing machine, though, and they really want to start selling software into the Linux market and believe that -- with Novell's help -- they can. On top of the Suse and PHP agreements, then, expect Microsoft to try for a similar relationship with MySQL and to augment the limited cooperation already going on with Apache.

The plan is to ultimately offer a Chinese menu of server components, where you can run cheap open source bits on Windows, perhaps using Suse as middleware, but with the knowledge that Microsoft is going to incessantly try to get customers to "upgrade" to SQL Server, for example.

The grandest manifestation of this strategy would be to decouple Windows from its kernel and somehow pop the API and other bits right over onto Linux, assuming that would somehow make the product more competitive. At the end of the day all Microsoft cares about is money, really, so if they could give up a lot of support headaches and support a far smaller code base yet get the same money, of course they would do it. I'm not saying this is where the strategy WILL go, just that it is where the strategy COULD go.

Yeah, but who is going to make this thing actually work? That brings us to the Novell/Microsoft interoperability lab, which was also part of last week's agreement. The moment I read about that, the question that popped into my mind was, "Which company will be doing all the work?" If historical patterns kick in, it will be Novell. As the smaller and leaner company Novell has always been able to beat Microsoft on technology. This agreement will tie Novell to interoperability promises Microsoft doesn't care about nearly as much as Novell does, with the result that Novell will do most of the work.

We saw this happen before when 3Com tied its fortunes to Microsoft in the late 1980s with the lamented 3Com-Microsoft LAN Manager network operating system, which was ironically Microsoft's answer to Novell at that time. Then 3Com CEO Bill Krause felt the only way to compete with Novell was through an alliance with Microsoft. So 3Com bought its way into the relationship, ended up doing all the work (MORE THAN all the work if you count recoding Microsoft blunders), then had to BUY ITS WAY BACK OUT when the product failed.

After that deal was over and the blood had dried, 3Com founder Bob Metcalfe claims that a Microsoft exec told him, "You made a fatal error, you trusted us."

This new agreement shows exactly how weak Novell has become and how eager Microsoft is to grow its server business at almost any cost. It isn't bad for open source, but neither is it good for open source. It is probably good for Windows developers who can now hope to deploy .NET apps freely on Linux using Novell's Mono code, which just shipped its final release version (no coincidence there). In the end, Microsoft WILL turn on Novell simply because they can't keep themselves from doing so. Those who think Microsoft has changed its ways weren't listening this week to what Ballmer was dishing out in India. This is not a guy who talks off the top of his head.

If there is a saving grace here it is that WordPerfect lawsuit against Microsoft that Novell has not yet dropped and can continue to use as a weapon against Redmond. This is Novell's third antitrust case against Microsoft and it won the first two. Novell knows how to win such a lawsuit and has a huge evidence database available to it from those other cases. If Microsoft tries to screw Novell, then Novell can turn up the heat on that WordPerfect case, possibly releasing new evidence that can further hurt Gates and Ballmer. Remember, all Microsoft stories are true.

The real irony, of course, would be for the Microsoft/Novell relationship to go swimmingly yet have Novell STILL pursue that WordPerfect case at full speed, winning it of course. I can see it now: "You made a fatal error," the Novell exec will say, "you trusted us."

The $200 Billion Lunch

Remember Y2K? If you worked in Information Technology in the waning days of the last millennium, you probably remember Y2K as a combination of Christmas and the hardest workday of your life. We'd programmed ourselves into a potential disaster with the way computers handled dates, and fixing the problem took several years and a reported $100 billion. Well if you liked Y2K, you'll LOVE IPv6.

IPv6 (Internet Protocol version 6) is, of course, the next-generation Internet address scheme that has been around for a decade but generally not implemented. Instead, we came up with Network Address Translation (NAT), a kludge that allowed us to stretch the available pool of Internet addresses that would otherwise have run out years ago. NAT works well enough that we could rely on it for years to come, but then those feisty Chinese went ahead and decided to switch their whole country to IPv6, so now we have to do it, too.

To a certain extent, it is Sputnik all over again. Some people see this as a place where there will be a commercial disadvantage unless the U.S. keeps up. It is comparable to NTSC vs. PAL television standards (hint: PAL is better but we don't have it).

To be fair, IPv6 is a far better solution to the problem of diminishing Internet address space than NAT could ever have been. IPv6 just expands the total size of the address pool by making the addresses substantially longer, with the benefit that the pool will be big enough for every device to have its own unique static IP address.

As things stand right now, something over 30 percent of Internet packet traffic is illicit, either spam email or attacks of various sorts. As such, a passive unprotected Windows system on the net can be infected with some kind of pathological code in a median time of minutes. Converting to IPv6 addressing would be a chance to at least get a finger into that leak.

There is also a very large market for being able to encrypt net traffic. IPv6 puts that where it belongs, down in the lower layers of the protocol stack. Right now we really have to put encryption in the top of the stack at the application layer.

The downside of all this upgrading is cost. Implementing IPv6 will incur an infrastructure cost of around $200 billion, and that's just for the U.S. Figure another $200+ billion for the rest of the world. In short, this means an IP feeding trough of unprecedented size.

The good parts about IPv6 include no more NAT, greater resistance to hacking (though that's only until the new IPv6 crime codes are introduced, believe me), and much easier tracking of data on the net. Figuring out your new IP address is easy, too, just add a string of zeroes at the front of your current IP.

The bad parts of IPv6 include having to replace most routers, as well as any performance hit that may come a jump in packet size -– today’s packets average 63 bytes, while IPv6 packets will weigh in at 87 bytes. But the real hit will come from inadvertently broken parts of the network, like anything based on Asynchronous Transfer Mode (ATM) technology. ATM uses fixed 53-byte packets with eight bytes of address. Switching from eight-byte to 32-byte addressing will decrease the packet data payload from 40 bytes to 16 bytes, which is not good. IF ATM survives it will require either a NAT-like kludge, new ATM equipment that runs 2.5 times as fast, or a simple acceptance that the new Internet is slower than the old one it replaced.

My Mama wouldn't like that.

But the U.S. military sure would. Future Combat Systems (FCS), the $125 billion (or $300 billion, depending whether oil changes are included) U.S. Army of the near future will absolutely rely on IPv6. FCS wants to make addressable over the Internet anything with an electrical system -- every flashlight, walkie-talkie, and Humvee. The FCS mantra is that everything that has electricity is a sensor, a node, an effector, or all of the above. That's a LOT of IP addresses. The same force is moving the civilian market, too, with RFID tags on everything.

All this government money is about to be spent on IPv6 upgrades because otherwise it won't happen. Nobody is going to do it voluntarily, so there is a federal mandate, and such mandates often come with federal money. This one sure does. People won't voluntarily upgrade because their systems are working just fine now, and will continue to work after the IPs run out -- they just won't be compatible with the later IPs. Until that SERIOUSLY affects their day, they won't spend the money to change.

Here's where I get on my soapbox.

Most readers of this column have known about IPv6 for years, but I doubt that many readers know a mandated upgrade is coming. It isn't my job to announce this stuff, yet it seems like that's what I often do. And I'll do so with a prediction that it won't be a smooth upgrade because we're too distracted with other issues and we'll turn this transition into an excuse to spend far more money than we really need to.

Instead, we should look for inspiration to the source of our most recent motivation to move to IPv6 -- China. In the current addressing scheme, China received a very small number of IP addresses, and this was causing them a lot of difficulty. If they stayed with the existing system it would have resulted in a nasty network kludge. So they made a national decision to implement IPv6 and put in a good network design. With IPv6 China has the address space they need and it is working well for them. Of course, the rest of the world is still on the old system and to communicate with China an address translation is needed. This is becoming a pain. Countries who want to do lots of business with China or who want to do lots of business through the Internet (India) are now seriously looking at their own IPv6 plans.

Look at it as leadership through good example. China has done something very impressive and now others are taking notice. We (the U.S.) think we control the Internet, but China is proving otherwise.

And what is happening in the USA? Well we have Net Neutrality. We have a telco rebuilding a national monopoly. We have Cisco and Microsoft working together on Network Admission Control (NAC). I can see a time in the near future when they'll try to charge me for every PC in my house. While China is building a national resource, our government is letting companies turn the public Internet into an expensive private toll road.

But we'll move to IPv6, that's for sure, if only to make sure Halliburton has plenty of business.