MILLER ON PROCUREMENT.

Software has never fit really well into the competitive process. It is much better suited for schedule purchase on a sole source basis or via regular sole source.

The reason is that no software firm starts out to create a piece of software that looks jut like anyone else's software.

It is very difficult to compare and if you do an RFP you either shortchange yourself or a portion of the vendors depending upon what you put in the specification or what you leave out.

If you write an RFP for spreadsheets, Internet access, DBMS, sorts, etc. unless a standard exists, and it seldom will, you, of necessity, must draw a common boundary in order to compare product specs.

This means you ignore features you don't understand or did not include in the spec and evaluate on some common grounds. Even if you use a value award or you make liberal usage of desirable features with bonus dollars, it is much less than an exact science.

But most agencies don't even make a skillful and honest attempt when they buy software. Let us suppose that an agency is trying to procure client/server software, a new e-mail system or a DEMS. They sort of throw up their hands at the complexity and break the rules.

A common tactic is to call in three or four vendors, run them through a dog and pony show with no specs or only the most vague of specs, run a test or two, a pilot or two and, viola, the IRA or Justice has a new software standard which was achieved in a manner similar to palm reading or looking at birth signs.

We have seen agencies with weak procurement controls, like the USMC, even do a checklist for hardware or software much as you see in trade magazines and declare the standard to be the product with the most checkmarks.

This probably has no relationship to agency needs and reflects the lack of understanding, so common as to the product specs and the agency needs.

None of this would stand up in court as a selection method. But it is far, far too common. Vendors are sufficiently naive that sales people jump at the chance to get involved in these, sloppy at best, wired at worst, selection processes and don't think the problem through.

At best they have one chance in three or six to be selected. At worst, they have zero chance. And it will not be a rational, disciplined process as federal procurement is supposed to be per the FAR in 1.4, Part 10, etc.

So, eventually the favorite is picked and the losers look so sad when they realize they took part in this face with that eternal salesperson's hope beating in his/her heart.

By then protests are probably impossible because of the ten day rule. Recently Oracle went through this with one of the worst agencies of all, the Navy, eventually lost the farcical selection process and filed a protest at GAO only to lose for lack of timeliness.

In effect, GAO said, "You chose to take part in this pathetic process, you lose and now you want to accurse the dealer of finding aces on the bottom after you played all night. Get out of here!"

This is a very difficult problem with which to deal. We all hope during the process to talk the techies into a sole source for us or a schedule buy for us, or into MIPRing finds to Contract X and buying us.

It is asking sales people to be too honest to avoid the process. They are driven by orders and honesty is not in the top ten list. Yet, if you chose to play that game you will probably lose and by playing the game you are doomed to accept the result.

Honest software RFPs are like men seven feet tall, very, very rare.

Much depends upon the skill and your prior experience with the contracts people. If they look dumb or cross eyed at me I assume that it will not end well and start to force them to play the FAR game. They will not like it and your VP will probably not grasp it but you will probably benefit in the long run.

Essentially, the biggest problem you have will be the greater quantity of sales people and the large clout someone else in the industry will have, except in those rare cases where you are the dominant player in your niche.

If you are Lotus or Tivoli the current process will probably fit you to a T. But if you aren't, don't be naive and let your naive hope overcome the cold light of intellectual assessment.

Software should either be done with an honest sole source per FAR Part 6 or an honest schedule buy per FAR 8.4 or an honest RFP per FAR 15. We seldom see it properly done.

And the scratch it together plan in some cases, such as a DEMS with a hurried pilot and selection of a product may well result in a product that doesn't work, is outgrown, is much, much too slow and the cost of the product may only be 10% of the total expenditure which the agency will be slowly sucked into in order to design, code and implement. Won't you look foolish in the IRS after all of this happens ad the product you wired the system to get won't even do the job?

Most agencies have had this problem, haven't they?