Recommended Resources
Cyberhaven.com Offshore havens, asset protection, global investing and other useful techniques.
The Year 2000 Bookshelf Books to help your evaluate the Y2K problems you face.

Gary North's Y2K Links and Forums - Mirror

Summary and Comments

(feel free to mail this page)


Category: 

Too_Late

Date: 

1998-05-23 16:54:49

Subject: 

Cost Per Line of Code: $6.46, Says Gartner Group

  Link:

http://www.techweb.com/se/directlink.cgi?IWK19980525S0037

Comment: 

It used to be said that it costs about a dollar per line of code to repair a system. Well, maybe sometimes, and in some locations.

Gartner Group now estimates the cost at six times the $1 figure.

How did this new estimate appear without warning?? Probably because of the fact that hardly any organization has advanced beyond assessment. Since so few had begun repairing the code, nobody had accurate estimates. Now the grim reaity is becoming apparent.

All of the budget estimates are now toast: Citicorp's $600 million (400 million lines), Chase Manhattan's $300 million (200 million lines), and -- most obviously -- General Motors' $600 million (two billion lines).

Does this mean that the problem can't be fixed? Yes. And what if the cost were 50 cents per line? It would mean that the problem can't be fixed. If the vast majority of organizations refuse to spend any money to fix it, and if there are not enough programmers to fix it, and if bad data spread to re-infect systems, and if old embedded chips can't be replaced -- all of which are true -- then it doesn't matter what the cost is per line of code. Managers are simply not going to spend the resources, including time, to fix the problem.

But at least this figure blows the whistle. It can't be fixed. It won't be fixed.

But nobody pays any attention to the whistle.

Nevertheless, if this estimate is verified, then the debate is over. From this time forward, it's just a matter of sitting back and watching the CEO's opt for early retirement.

My friends, it's over. For amusement's sake, and for the sake of the cause, I'll keep posting stuff on this site, but it's all superflous from now on. It's over unless someone can prove that Gartner missed it completely or this report got it all wrong. Don't count on either.

This was in INFORMATIONWEEK (May 25).

* * * * * * * *

Similarly, the earlier estimated cost for year 2000 fixes of $1.10 per line of Cobol code now looks way too low. In an in-depth study of three companies, Gartner Group found that all three had erred egregiously in their estimates of where their companies do their computing. Though the companies had said usage was 80% on the mainframe, 20% on client-server, Gartner Group found the reverse to be true. The IT advisory firm has accordingly upped its estimate. Taking all computer languages into account, plus PCs, servers, embedded systems, networks, vendor compliancy, supply-chain analysis, and everything else falling into the scope of the three companies' year 2000 projects, Gartner Group now estimates that the rule of thumb for average year 2000 costs is $6.46 per line of code-nearly six times higher than its original estimate.

Link: 

http://www.techweb.com/se/directlink.cgi?IWK19980525S0037

Return to Category: Too_Late

Return to Main Categories

Return to Home Page