The Tower of Babble May 12, 2000 URL: http://www.financetech.com/showArticle.jhtml?articleID=14704037
And they're off! FIXML jumps out of the gate with FINXML trailing and FPML in a distant third. Entering into turn one, FPML jumps in front with FINXML falling slowly behind and FIXML pounding at a steady pace. The crowd of IT executives and developers jump to their feet, sweaty with anticipation, clenching their stubs, arms outstretched in prayer. If they've chosen correctly, the time and energy they've spent on developing networks and applications will prove to be a windfall for electronic trading and counterparty communication. If they lose ... well, no one really wants to think about that, and the horses are still pounding around the first turn, their anointed riders whipping them with all possible force.
If you find yourself confused about these emerging XML "dialects," you're not the only one. XML (Extensible Markup Language), among many things, promises a neat and tidy common data format, but, as with all widespread industry efforts, there are tensions, egos to consider and vested interests at stake.
First, a primer. Wall Street first recognized the potential power of XML as the base on which a number of grammars or dialects could be created for transmitting electronically formatted messages. They saw in XML the possibility of interoperability, providing the ability for one firm's systems to speak with another, but as more and more IT divisions jumped on XML as the future, they've created a number of dialects with no common translator.
"People started noting-and FIXML was among the first to do this-that XML has the potential of doing for computer to computer messaging what HTML has done to terminal-based input, which is to establish a lowest common denominator...and unleash a whole new, more efficient way of communicating," recalls Dwight Arthur, managing director at the DTCC. "Many organizations jumped at this opportunity, which has brought about a new golden age of interoperability. ... As with all golden ages, this one is wearing thin rather rapidly. ... (A)nyone who hasn't created their own version of XML-something is noticing that they're being left out."
FIXML, currently being developed under the auspices of a FIX Committee working group, is intending to handle equity and fixed-income transactions and communications just as FIX has done till now. FPML-created last year out of joint partnership between J.P. Morgan and PricewaterhouseCoopers, and now controlled by the FPML Organization, an industry consortium-is focused on over-the-counter (OTC) derivatives.
Unlike these two, FINXML was created by a vendor, Integral, as a key to its CFOWeb.com B2B trading site. Those investment managers and banks that sign onto the site need to use FINXML to connect their back-end processing systems to the CFOWeb browser-based front end, but Integral has availed the protocol to the industry for free.
Contrary to widely held misperceptions, both the FPML and FIX organizations believe their standards are rather complementary. First off, the two are focused on entirely different segments of the securities markets.
Brian Lynn, the program manager of FPML at J.P. Morgan and co-chair of the FPML standards committee, emphasizes that the two standards are potentially wonderful fits for one another. "The thing to understand is that the FIX standard is primarily about price negotiation and execution of predefined securities," Lynn says. "The OTC derivatives market is quite different. Each product that trades is potentially unique, and the terms are negotiated between the counterparties. So, the main issue for trading OTC derivatives is to have a full and rich description of the product itself, so the focus of FPML is on product description rather than the process."
Whereas FIXML incorporates both the message and the transmission, known as the session layer, FPML is primarily concerned with the message. How it gets from one party to the next is a secondary concern.
"The reason a session layer wasn't included is that this is viewed as an abstract description of the deal that could be transmitted over any medium," Lynn explains. "It can be sent over e-mail, it could be sent on a message queue, over the Internet. Any medium should be viable."
And FIX may just prove to be that "medium." Lynn admits that both organizations are working together to consider how each protocol may interoperate, and that one could "potentially" use FIX for the transmission of the data incorporated in an FPML-formatted message.
Similarly, the FIXML working group did a very interesting thing when it released the first spec in January of last year. FIX, including the most recent version 4.2, incorporates the message and session into one layer, whereas FIXML separates the two. Goeller points out that as technology continues to evolve, it may become more advantageous to those firms using FIXML to transmit the messages via another means.
"Currently, FIX works pretty damn well," he says in the current protocols defense. "The challenge is that there will be alternatives to this. As I've said, there are a lot of people working on this and there's a lot of Internet-based technology evolving which FIX may take advantage of."
Whether the FPML or FIXML group are posturing or genuinely interested in working together remains to be seen, but there exists a significantly larger rift between Integral's FINXML and the other two groups. Representatives of Integral joined the FPML Organization back in August, but they have participated little in its development, and tensions between the two groups are at all-time highs.
Harpal Sandhu, ceo of Integral, spared no feelings about the subject in an interview last month. "The proposition that we're giving people is it's open-source, it's free, and most importantly, it works," he said. "It's a commercial system as opposed to a specification. You can probably expect a much greater level of functionality and product support out of FINXML because there's a massive company around it rather than a consortium of people debating object models about how XML should be defined."
Defending the benefits of a consortium, Lynn responds that J.P. Morgan and PWC were looking to give FPML to the industry so that as it evolves, it comes to encompass the needs of the industry as a whole. "FPML is something created by the industry for the benefit of the industry," he stresses.
An industry observer was a little less restrained in response to Sandhu's assertions. "I definitely would say commercial considerations are in (Integral's) mind, and I think they're pretty clear about that," the insider says. "At a panel discussion ... a couple of months ago on which Harpal was speaking, he said repeatedly that they were a commercial enterprise, and needed to make sure all that happens ultimately contributes to the value of their firm. That's what J.P. Morgan and Pricewaterhouse were trying to avoid."
In a very commercially oriented pose, Sandhu essentially accused the FPML group of largely copying the work Integral had done in creating FINXML. "Look at the chronology. We announced FINXML in April of 1999 in a launch that we did with Sun Microsystems and Oracle. In the audience there were individuals from banking institutions including the primary sponsors of FPML that wrote very detailed spec papers on what they had seen, and that was used as the basis of an announcement four months later for FPML."
Ouch.
Lynn flatly denies any replication of the FINXML specification. He points out that there are certain business processes inherent in derivatives trading that would call for similar functions in the two protocols, but those are elements the industry has asked for. "I would expect any XML-based derived standard to include some basic elements, but express them differently," he points out. "For its part, FPML is based entirely on ideas and designs arrived at originally by J.P. Morgan and PricewaterhouseCoopers, and is now being refined by the FPML technical and standards committees."
He adds, "We've had reports from a number of the 40 or so people working on FPML right now that they weren't considering other standards, because some of the license terms were too restrictive. They would force them not to use any ideas that they would come up with in any other place."
If anything, Lynn says the forthcoming version of FPML-expected in May-has borrowed some elements from SWIFT, an active participant in the FPML standards committee. "SWIFT has some message formats that were created for confirming derivatives transactions and some information was taken from those with the consent of SWIFT," he details.
Accusations aside, the firm matter at hand that Goeller, Lynn and Arthur are primarily focused on is the future of the development of these protocols and the promise that they someday be able to communicate with one another. The current state of interoperability has lead Arthur to call it a "Tower of Babble." He and others are looking at other initiatives-like Oasis, BizTalk and EBXML-that will potentially serve as translators amongst all these dialects, a true lowest common denominator to ensure fluid and seamless electronic communication.
Despite all the haranguing, Goeller still sees XML and all the work that has so far been accomplished as a wonderful thing. "Look at all the conversations we're having with SWIFT and FPML and all these other folks that wouldn't normally have these conversations," he says. "We are having conversations about a thing called XML, and from that standpoint, its good."