unless the burden or expense of the proposed discovery outweighs its likely benefits, taking into account the needs of the case, the amount in controversy, the parties resources, the importance of the issues at stake in the litigation and the importance of the proposed discovery in resolving those issues.

This is the basis for asking for cost shifting. Usually the responding party pays for its discovery when the material is stored in a form that is useful in the ordinary course of business, though there are exceptions. Electronic backup data or other data that has to be synthesized in a burdensome way (not just a simple database query) is not in this class. Thus courts have held that cost shifting may be appropriate. It is important to note, however that the cost of a privilege review of the material is a separate issue and that a determination of whether material needs third party clearance is not an ordinary privilege review.

The courts have had a number of theories, but there are two key cases which each created tests

Rowe Entertainment v. William Morris Agency 205 F.R.D.(S.D.N.Y 2002) used these factors

1. the specificity of the discovery request;

2. the likelihood of discovering material data;

3. the availability of those data from other sources;

4. the purposes for which the responding party maintains those data;

5. the relative benefits to the parties of obtaining those data;

6. the total costs associated with production;

7. the relative ability and incentive for each party to control its own costs;
   

and

8. the resources available to each party.

If we apply this to SCO v. IBM, SCO clearly wins on 3 and 8. Only IBM have the code and SCO is broke. IBM clearly wins on 5, the discovery is of no use to them other than the use generated by SCO's review of it.  If that counted, the test would be meaningless. What about the 5 remaining tests.

IBM is way ahead on the likelihood of finding anything. This is tricky, because the relevance needed is to SCO's claims. If the test were the likelihood of affecting the outcome, IBM would win, hands down. Even with the relevance to claims standard, IBM is ahead, though, because the claims are vague and SCO have already gotten a great deal of code. They ought to have found something, maybe not much, but something to support their theories.

IBM is probably ahead on the cost test. This discovery should easily cost more than $10,000,000. I think that the need to document the cost is a reason IBM has yet to move for SCO to pay.

IBM probably wins the test of cost control incentive. SCO have no incentive to control costs and IBM no incentive to inflate them.

The specificity issue is either an IBM win or a split decision. The versions request is specific in a sense. All means all, but the definition of a version is a problem, unless it means every checked in version of every file regardless of the relationship between the files. Even there, you're plagued with thousands of cases of different files with the samename.The most important contributors is a win for IBM. This almost redefines unclear unless it is interpreted as those who checked in the most documents.

This makes it 4 1/2 IBM, 2 1/2 SCO, with one to go. Given the SCO's strength on lack of ability to pay test, if SCO had another clean win, SCO would probably be off the hook, with the proviso that each side could win in part. This would be a loss for SCO, because even 1/3 of the costs would be a big hit. There is still a ton of money to be spent on the patent counterclaims, and there's only $5,000,000 in the escrow less whatever SCO has spent to date.

The last test is use in the "ordinary course of business". To the extent data has to be recovered from backup IBM win. They also probably win the most important contributors issue. IBM also probably win the production of all versions issue. So I give this one to IBM.

The result is somewhere in the range 6-2 IBM to 5-3 IBM.

 Zubulake v. UBS Warburg LLC, 217 F.R.D.(S.D.N.Y. 2003) is the other commonly cited case. It divided electronic data into 5 classes depending on how much effort it took to retrieve, recognize and convert the data. Backup tape was the hardest because of the costs of things like identifying text files and formats and resolving duplicates.

The Zubulake reasoning would apply to IBM where they have to determine what constitutes a particular version or otherwise use the version control system in ways it is not usually used that are burdensome.

Having categorized the discovery data, Zubulake then applied to the portion that was judged extaordinary, 7 tests in order of importance.

1. the extent to which the request is tailored to discover relevant data;
similar to Rowe 2

2. the availability of those data from other sources;
Rowe 3

3. the total cost of production, relative to the amount in controversy;
Rowe 6 with regard to rule 26(b)(2)(iii)

4. the total cost of production, relative to the resources available to each
party;
Rowe 6 with regard to Rowe 8

5. the relative ability and incentive for each party to control its own costs;
Rowe 7

6. the importance of the issues at stake in the litigation
From rule 26(b)(2)(iii)

7. the relative benefits to the parties in obtaining those data.
Rowe 5

Under these tests, SCO wins 2 and 4 and probably 3 and 6 as well. IBM wins 1, 5 and 7. The result would be that SCO would probably be ordered to pay a percentage of certain of the costs. What percentage I don't know, but I think it would be in the 20% to 75% range.

However the actions of the participants will only partly be determined by the law. If IBM moved now, neither Judge Kimball nor Judge Wells would rule until after discovery. If they made a substantial award to SCO, SCO would appeal on the grounds that it denied them the ability to pursue their case. If the appeal was heard, there could easily be reversal at least in part and a remand, particularly if the review was de novo (IANAL, so I don't know whether it would be). This would slow the case down. Even the motions would be a drag on things.