Let's talk about RFID. But first, let's imagine the Internet as it might be. Suppose every ISP required its users to buy only its own brand of modem. And use only its own proprietary Web browser. And connect only to Web sites certified by the ISP to work with that modem and Web browser.

How big do you suppose the Internet would be today?

That's pretty much the situation we're in now with RFID.

A few weeks ago, I wrote about how Wal-Mart is trying to restart its stalled RFID effort. In 2003, Wal-Mart mandated that its suppliers attach RFID tags to pallets of goods they deliver to the retail giant, starting in 2005. Some have, but many have refused.

Wal-Mart's solution: Charge suppliers a fee for attaching an RFID tag. No RFID tags on your shipment? You get dinged US$2 per pallet. That seemed like a smart move to me, because it sets out a clear, easy-to-calculate cost for suppliers still deciding whether RFID is worth the price.

I still think it's a smart move. But after hearing from readers, I don't think it will move any suppliers to RFID.

Why not? Because tags are a trivial part of RFID's price. As one reader put it, "A pallet already costs US$15. Another $2 is no big deal."

In fact, for small suppliers, US$2 is a great deal, because the RFID world today looks a lot like that imaginary, highly proprietary Internet. RFID tags may be standardized, but that's not true of the software that supports using them for warehousing and inventory management. That software may nominally meet formal standards, but it won't interoperate with any competitor's software.

So when a big customer like Wal-Mart goes with Vendor X's system, all of Wal-Mart's suppliers have to buy software from Vendor X too. They can't buy from another vendor or roll their own, because it must be certified by Vendor X before it can talk to Wal-Mart's software. The suppliers are a captive market.

And when Target goes with Vendor Y, and Sears goes with Vendor Z, the number of RFID systems required for captive suppliers just keeps going up -- at monopoly prices.

No wonder US$2 a pallet sounds like a bargain.

What makes suppliers so sure that's how it will go? Because they went through this two decades ago, when big customers mandated the use of electronic data interchange documents. As with RFID, there were EDI standards, but there was no actual EDI interoperability. And suppliers were captive -- and got soaked.

They don't want that to happen all over again.

What they want is something like the Internet, where modems and networks and Web servers and browsers all interoperate, no matter where they come from.

Wal-Mart can make that happen. The world's biggest retailer could start designing its next-generation RFID systems today, using a set of interfaces that can be used by any RFID software -- commercial, home-brew or open source. Wal-Mart can force interoperability.

That's in Wal-Mart's interest as it tries to get more from its investment in RFID. It's in the interest of small suppliers that need the competitive pricing that interoperable RFID systems will allow.

And it's also in the interest of the RFID vendors who are still fighting to remain proprietary. See, proprietary RFID can never grow beyond the resistance of captive suppliers. But interoperable RFID could become huge -- with software-vendor profits to match.

When they finally understand that, they could make RFID as big as the Internet.

Frank Hayes is Computerworld's senior news columnist. Contact him at frank_hayes@computerworld.com.