Tuesday, May 30, 2006

ACORD is a Guideline NOT a Standard

Apologies to my readers who are not in the insurance industry and do not care about the ACORD XML "Standard". I have gabbed a lot about it lately. Perhaps this applies to other "industry standard" xml efforts. I know that it also applies to my experience with OAGIS. I also recently posted on canonical message format as well (related). I hope that this post and the comments that may ensue will put this topic to bed for a while.

While at the ACORD conference last week in Las Vegas, I challenged myself to interrogate every vendor, carrier, and ACORD employee on whether ACORD is a standard or a guideline.

To a person the consensus was that it is a guideline.

Perhaps I am just a stickler for words, but this really bothers me. I listened to executive after executive and industry analysts yapping and banging drums about "standards" and how everything is just great in the insurance world on the wire. I listened to the Gartner insurance dude spaz out about ACORD "standards" and then recommend that carriers fork it (excuse me, use ACORD as a guideline) in one sentence. The insurance analyst panel (Gartner, TowerGrouup, Celent - liked their guy Matthew Josefowicz the most)) was pretty comical - one of the more arrogant groups I have seen. Sniping at each other, challenging each other to dare to question the standards -- saying things like, "the standards have never been as close as they are today ...". They certainly all know more then I do about the insurance business, but someone needs to remind them that their PowerPoint presentations don't compile. They should take a look beneath the covers of a couple SOA projects and check to see what is really on the wire.

I talked to several aggregators and AMS reps who said that in their eyes ACORD is at best a guideline. They did say that since carrier xml is often ACORD-like and at least has some similarities, it makes things better then if everyone started from scratch. I can buy that - but that isn't standard.

I listened to a presentation from a reinsurance intermediary (Guy Carpenter) on an event driven impl. They forked the ACORD standard and have no regrets. By the way that was the best presentation of the show (even though it was called "Using ACORD XML Standards to Build Enterprise SOA"). I have a feeling that the presenter, Paul Fox didn't name it that.

I challenged some CIO level folks who were talking the standards bit. They quickly folded and agreed with me.

What bothered me the most about the ACORD conference was that there was no serious discussion about the realities of ACORD XML in practice today - no sessions, no critical thinking - zip. Everyone just seemed to wink and nod and say nice things about standards. Also, zero talk about 2.0.. Perhaps I just wasn’t in the right sessions.

Every sane person I spoke with is following the prune strategy accepted in 2.0.

I still believe that it is appropriate to use the ACORD guideline as an starting point for a insurance entity's canonical form - provided it is aggressively pruned and there is no requirement to keep up with future versions of the ACORD guideline. Most importantly, don't spread the illusion that your canonical form is "standard" -- that there will be some magic interop some day. There will not. This is ok. Owning your canonical form outright is far more important then conforming to a standard that doesn't really exist.


c. john palmer said...

guideline is the only way to look at ACORD. I've been using that mantra for years. I was impressed with how fast they(ACORD) went from the AL3 format to XML in the early days - but AL3 is also a guidleline not a standard.

IMHO your are right on track to say it is better than nothing. Coming from the carrier space - getting similar XML from agency management system vendors (who hate each other) and leveraging XSLT to start building real Straight-Through-Processing into our systems is getting us on the right track.

But don't get me started on SPX...

Anyway - keep on pluggin away

Rick Jelliffe said...

I think most standards are 'guidelines' in your sense.

The trouble I see in a wide range of financial/insurance systems is that they all substantial and inertia-bound investments in software systems with particular schemas, and these do not necessarily match the 'standard' schemas.

Given this, I think it is often prudent to see these industry standards as long-term influencers rather than instant interop pixie dust. And the long term may be 5, 10, 15 years, depending on the capital cycle, regulation and market requirements.

Which means that the success metric for the devloping organization is not so much "can everyone interop today" but "is this getting us towards interop?"

Everything that rises must converge!

Scott Welker said...

Very useful 10+ years later. Thank you!