(FORTUNE Magazine) – American stock exchanges are now in one of those "In the future, everyone will ride to work on monorails" phases. That's the stage institutions pass through whenever a major shift in the business environment leaves everyone uncertain how the future will work out and scrambling to inject themselves into it somehow.

Today's stock market events certainly qualify as a major shift. Electronic communications networks, or ECNs, are stealing market share from the established exchanges; the exchanges, in turn, propose to go public; Nasdaq and the ECNs threaten to become exchanges; everyone wants to trade all night on the Internet; and while we're at it, let's banish those funny fraction prices.

Behind all of this is a classic market share battle between a group of technologically advanced upstarts (led by the ECNs) and established brands like the New York Stock Exchange and Nasdaq. Think of Amazon.com vs. Barnes & Noble three years ago, and you'll be close--although in this battle, regulation plays as big a part as technology. The outcome is likely to be what it usually is whenever competition intensifies: Profit margins will shrink (at least at first), new alliances will be made, and customers will reap lower prices and wider choice.

But there is also a more fundamental power struggle being played out here. The same forces that gave rise to ECNs--regulation, new technology, and old-fashioned entrepreneurship--are slowly dragging the center of the securities industry from the floor of the New York Stock Exchange to an as-yet-unshaped electronic marketplace. That's not going to happen tomorrow, it won't wipe out the NYSE, and it probably won't make any perceptible difference in an ordinary order for 100 shares of AT&T. But for those concerned about the long-term fairness and stability of the market, this is the battleground.

To understand what turf is being contested here, let's review some market basics. The stock exchange business, for all its jargon-riddled, acronym-happy complexity, amounts to a simple matchmaking enterprise. Trading is like dancing: You need to find yourself a partner. To draw people in, the NYSE and Nasdaq have allowed matchmakers to set up shop on their dance floors.

If these matchmakers can't set you up with another client, no problem. They have the capital to hire a cadre of taxi dancers, and they'll rent you one, earning a profit on the spread between the cost of hiring the dancers and the rate at which they rent them out. Taking part in the trade this way is much more profitable than matching dancers directly. The prospectus for a recent IPO by LaBranche & Co., a large NYSE matchmaker, revealed that the firm partook in an aggressive 30% of the trades it handled, but those trades accounted for 75% of its total revenue.

On the New York Stock Exchange, there's one matchmaker per stock--the specialist. According to Salomon Smith Barney analyst Guy Moszkowski, specialists enjoy an operating margin of an astounding 55%. On Nasdaq, competing firms known as dealers or market makers cover each stock. (The average is 11 dealers per stock.) Lacking the specialist's exclusive franchise, they have more modest margins of 25%. Both the market maker and the specialist agree to put up their own money and maintain a buying (bid) and selling (ask) price for a particular stock. That means investors can always count on being able to trade--though they may not always like the price.

These matchmakers have two types of clients. Picture a guy walking down the street yelling into his phone, "Buy Intel! Buy now!" He just entered a "market order"--buy at the current price, whatever it may be. This guy is the middleman's bread and butter. The other customer, who places what's known as a "limit order," is a hassle. She'll only dance with guys with blue eyes and hairy chests, say, and she'll buy Intel only at 70 or lower. These special requests go into the matchmaker's "limit-order book," which used to be literally a paper notebook. The ECNs that have so shaken the established exchanges are nothing more than a cheap, fast, electronic version of that notebook.

The changes now sweeping the equities marketplace got their start in 1975, the year fixed commissions were abolished. Congress ordered the Securities and Exchange Commission to push Wall Street toward a "National Market System." In this ill-defined platonic ideal of a marketplace, information would flow freely, markets would compete, and investors could meet without a middleman. The easy stuff got done. For example, after some wrangling, the Intermarket Trading System was put in place to link the NYSE with competitors like the regional exchanges. But things stalled when the SEC started talking automatic execution on an electronic system, which would eventually cut the middleman out of orders. New York greeted that idea as enthusiastically as the NRA embraces One World Government. Like another idealistic scheme Congress proposed that same year, the metric system, the plan was mainly ignored. And like a mom who forgets she meant to send her punk son to Sunday school, regulators let the issue slide.

Until Mom caught her little boy stealing. In 1994 Nasdaq market makers were taped harassing colleagues who dared to make the bid-ask spread too narrow. (In other words, they were enforcing a price-fixing scheme.) The securities industry was busted. The punishment: a $1 billion antitrust civil settlement--the largest in history--and the SEC's 1997 Order Handling Rules, which were designed to set the industry back on the righteous path toward the National Market System.

Before these new guidelines went into effect, a market maker who got a limit order he didn't like--that is, one that would hurt profits--could ignore it, pretending no match was available. The new rules required him to post limit orders on Nasdaq or send them to an electronic communications network that would post them where everyone could see them.

Back then, the only ECN of any consequence was Instinet, a network in which institutional investors could trade in secret with each other and with market makers. Suspicious of this private party, the SEC effectively forced ECNs to post their quotes on Nasdaq's comprehensive trading bulletin board, known as a Level II screen. It displays all market makers' and ECNs' best prices and quantities. Buyers line up on the left and sellers on the right, with the best prices at the top.

Allowing ECNs onto this system was a crucial change. Previously, an ECN could stay in business only if it could attract both partners to its dance floor. Once it was allowed to advertise trades on Nasdaq, however, it just needed to supply one party. The other side could come from anywhere else in the Nasdaq system. By forcing Instinet into the public marketplace, the SEC had inadvertently flung open the public marketplace to ECNs.

Day-trading firms, which for years had sought greater market access to Nasdaq, rushed to set up ECNs. Island ECN (largely owned by the online brokerage Datek), Attain, Archipelago, and NexTrade, for example, all started off handling day-trading orders. Not to be left out, brokerage firms and other traditional players backed their own stable. REDIBook was launched by the NYSE specialist Spear Leeds & Kellogg. Strike was started by Bear Stearns.

Altogether, the nine ECNs that are hooked up to Nasdaq's trading system have captured an estimated 20% of Nasdaq shares traded, with most of the volume going to Instinet and Island (see chart). The growth rate has been phenomenal. ECNs have been capturing an additional 1% to 2% of share per quarter, and experts predict that they could take 50% of Nasdaq's business within a few years.

The ECNs all have different clientele and pricing structures, but the basic formula is the same: They internally post the size and price of limit orders they get from their subscribers and automatically execute when a match is found. When a Datek customer places a bid for 1,000 Microsoft at 90, the order is sent instantaneously to Island, which scans its outstanding offers for a matching seller. If it finds one, Island automatically completes the trade. If not, the order will post on Nasdaq as soon as it becomes Island's best Microsoft bid. Once it is there, anyone with access to a Nasdaq screen can take it. "What we are is your gateway to the National Market System," says Kevin Foley, chief executive of Bloomberg's ECN, Tradebook.

One of the great advantages of this model is privacy. Internally, ECNs list only the sizes and prices of orders, not the identity of the trader. On Nasdaq, orders are identified only by the ECN's name. This is vital to institutional investors, who are frequently burned by trading desks that "front run" the institution. To return to the dancing metaphor, if an institution has a busload of dancers who want red-headed partners, a matchmaker might run onto the floor and hire all the redheads first, then offer them to the fund at a higher charge. "Fear of front running and information leakage is the entire electronic trading business," says Foley. In addition to anonymity, ECNs automate some of the services that middlemen provide institutions. For example, ECN software can let funds post large, potentially market-moving orders in reserve. Charged with a 100,000-share bid, it might coyly post a bid for 1,000 shares that will automatically refresh itself or accept larger offers until all 100,000 have been accumulated.

What is confusing about ECNs is that they behave like a broker--by taking orders and disguising trades--but they make money like an exchange. When they take orders, ECNs charge a razor-thin commission ranging from 0.0015 of a cent to 4 cents per share. Market makers, on the other hand, rely on the spread, which is typically between 1/16 and 1/4 of a point (or 6.25 cents and 25 cents). When prices switch to decimals next spring, the spread (and with it, market makers' profit margins) will shrink.

ECNs are at the forefront of the highly publicized push toward after-hours trading. Up to 40% of online trades are placed after the market closes, when investors get home from work. But for now, these nocturnal orders only pile up at trading desks for the following morning. The result is an increasingly chaotic open that strains the capacity of the current system and may not give investors good prices. (So you should confine your own market orders to daylight.) For the thinly staffed ECNs, however, extending hours is relatively easy, and the prospect of building volume by absorbing some of the nighttime overhang is hard to resist. The ECNs have formed a consortium to share after-hours trade data so they can keep making money after the established exchanges go to bed. Some ECNs already operate after dark. Island, for example, trades from 8 A.M. to 8 P.M. weekdays, and NexTrade never closes.

Building volume remains the critical job. According to Moszkowski, most ECNs only generate enough traffic to match 5% to 10% of their orders internally.

It takes volume to build volume: The best way to draw dancers to your dance floor is to have plenty of other dancers. So, like new Websites trying to attract eyeballs, ECNs will do anything to get "liquidity," that is, limit orders in their book. Many have sold equity to partners who can send them more orders. Archipelago now counts Goldman Sachs, E*Trade, and American Century among its owners. BRUT and Bloomberg's Tradebook get clients from brokers they have already wired for their trading and information systems. BRUT also announced plans to merge with Strike. Other ECNs pay for liquidity. Island, for example, rebates 0.10 cent of its 0.25-cent-per-share fee to anyone who places an order.

So far, ECNs have taken only about 4% of the Big Board's volume. The chief reason: The exchange dominates trading in listed securities, controlling some 84% of the share volume in those stocks. The pool of liquidity is so deep that in 75% of the orders, buyers meet sellers without the specialist taking a position as a principal. On the Big Board, as in an ECN, volume begets volume. The NYSE's volume also allows it to charge transaction fees that will be hard for ECNs to undercut. In many trades, the exchange's and specialist's fees come to less than 0.4 cent per share. So why go anywhere else to trade NYSE-listed stocks?

Institutions have a good reason to look elsewhere--front running. But a trusted specialist can provide unique insight. If you're a big institution, a shrewd floor broker can judiciously parcel out your trade to avoid moving the price against you. Their judgment and savvy are something that no ECN computer can ever fully emulate.

Moreover, New York has shown in the past that it can adapt to competition. In the 1970s broker Bernie Madoff popularized the third market--in which nonmembers trade listed stocks off the floor--by undercutting the exchange's then-thick margins on retail orders. The NYSE responded by lowering fees. To compete, Madoff now pays half a cent for trades sent his way. The usual estimate is that the third market's share is about 10% of the volume in listed stocks.

To compete more nimbly, New York and Nasdaq both plan to shed their cumbersome ownership structures by going public. Both proposed IPOs were immediately greeted skeptically, and the NYSE's plan was almost universally pegged as a lousy investment. "It's like Henry Ford is making his assembly line, and the biggest buggy maker is planning an IPO," says Harvey Houtkin, Attain's CEO.

Another potential breach in Fortress NYSE concerns the exchange's Rule 390. Keeping with the spirit of the Buttonwood agreement, the NYSE's collusive founding document, the rule forbids member firms from making a market in a stock listed before 1979 except on a registered exchange. "Rule 390 is the last vestige of the Buttonwood agreement," scoffs Steve Wunsch, president of the Arizona Stock Exchange.

SEC Chairman Arthur Levitt has hinted strongly that he wants Rule 390 junked, but NYSE President Richard Grasso says he has no intention of doing so. The argument may be moot, however, because of a 1998 regulation that allowed ECNs and, more important, Nasdaq to become registered exchanges. So far, Island, Archipelago, and NexTrade have begun the process, and Nasdaq will soon. "I think there's the potential for this to explode," says National Association of Securities Dealers President Rick Ketchum. Once that happens, there is nothing to stop NYSE members from dealing Coke, say, on Nasdaq.

Of course, the possibility that member firms may be allowed to trade away from the floor doesn't mean they will. After all, they can already go on Nasdaq to make a market in stocks listed after 1979, but they seldom do. Sources disagree as to whether that reluctance arises from exchange politics or just from the difficulty of competing with the NYSE's awesome liquidity.

However well the NYSE staves off ECNs now, the electronic marketplace in general may be too powerful an idea to resist forever. Regulators clearly have not forgotten the National Market System: Levitt has mentioned it in two recent speeches. "Basic economics tells us that the greater the supply and demand that congregate in one place, the more efficient the price setting," he said in one speech, calling for a "virtual limit order book" that would display all orders from all the markets to all participants.

That kind of marketplace could have less room for the middlemen of Nasdaq and the NYSE. Indeed, the entire electronic communications revolution is about nothing if not reducing the middleman's role. Matchmakers will still matter wherever market participants have a hard time meeting--for thinly traded stocks, for example. And as institutions have grown larger, they need a middleman's capital to help them complete trades. But otherwise, an NYSE specialist is not indispensable. Middlemen often argue that their capital will add stability, especially during market turmoil. But a specialist needs only to have $1 million in liquid assets and be able to buy 22,500 shares of the stock they specialize in--30,000 if it is in the S&P 500. "It's ridiculously low," says one Street executive. "When this freight train turns around, no specialist firm on the floor will be able to stop it."

The traditional market structure was created of, by, and for middlemen. For the past 25 years, technology, regulation, and entrepreneurs have been chipping the edifice away. In the future, even retail investors may learn that the instant gratification of a market order sent through a middleman is expensive. "What is the value?" asks former SEC chief economist Susan Woodward. "If the market maker is not there, you might have to wait a little longer [to see your order clear]. That's it. We're not talking about three weeks, we're talking about 90 seconds." Woodward, now chief economist for the Internet-based private-equity marketplace OffRoad Capital, predicts that middlemen will stop committing their capital to make markets in highly liquid stocks because they simply won't be able to compete profitably. In that world, it seems safe to say, 55% margins will be hard to come by.