Oral argument — Susan Long v. IRS

May it please the Court, I'd like to reserve a couple of minutes, two minutes I think, of my time for rebuttal. This case, which is just the most recent chapter in a long-running series of cases brought by Susan Long against the Internal Revenue Service to obtain information under the Freedom of Information Act, essentially presents two issues. The first is one of statutory language, specifically two phrases in the statutory definition of return information. The statute provides that return information does not cover data which is in a form that cannot be associated with, and then it has the next phrase, or otherwise identify directly or indirectly a particular taxpayer. Now, here's where the controversy focuses. The language in a form that we submit is critical. Identification is important, but the first and most important aspect of this litigation is that the tax return information must be put in a form that represents a new statistic, and then as a second matter, the IRS must determine that it cannot directly or indirectly identify a particular taxpayer by some form of perhaps reverse engineering. Now, this is not something new. This is the holding of the Supreme Court in Church of Scientology in a form that language and made it very clear that the IRS will not comply with Section 6103 if all it does is remove identifiers. Rather, the data must be reformulated, and in this particular case, the data is very narrow. The controversy between the parties as to this first aspect of the case, and I'll get to the second in a minute, as to the first aspect of the case, is simply the presence of what are called cells, which are just data entry points of one or two, that is, the data reflecting one or two taxpayers on two IRS reports, forms 37 and 38. So, I looked at what the IRS submitted. I'm terribly sorry, I can't hear you. I said I looked at the documents, the statistical data that the IRS submitted in its supplemental excerpts. Yes. I don't see how even the one cell violates the Haskell Amendment. How would you go about identifying from that data? Well, if it's done now with a computer program, I am told, that can scan these and pick up the cells of one or two taxpayers in particular audit data changes. How is it done? And how are we to rely on what you are told about some computer program that exists out there? I mean, how actually is it done? I do not know how these are scanned as a matter of computer science, but I know that there are cells of one and two on forms 37, tables 37 and table 38. That's also true of some of the other tables that are not in controversy here, the inventory tables 35 and 36, but the computer and the IRS experts who scan these are able to do this and they will release to Ms. Long in the tables 37 and 38. And keep in mind, table 37 is almost 4,000 pages long. They will release all of the data except those entries that reflect cells of one and two, because it's quite clear that with a cell of one, there is no reformulation at all. And that data... Is your argument that there must be identifying information of the name of a particular taxpayer? Oh, they would go far beyond that, Your Honor. It would include his name, his address, his Social Security number. But those are not in the cell? I don't think so, no. I think that this would simply be a data change rather than a name change of one particular taxpayer, so that it would be his return item. It wouldn't be his address. That would be stripped out in any event, but it would be an audit change item that his capital gain item of $20,000, for instance, had been increased or decreased by a certain amount. But it would be one particular taxpayer. The same thing for a cell of two. No name, no address, no social... no clear identifiers, but it would reflect only two taxpayers' change data, audit data. Now, with that, you can see that there has been no reformulation at all with respect to the cell of one, and it's minimal with respect to the cell of two. It's minimum enough? The IRS says no because the statisticians work... When they do things like the statistics of income, they work on what they call the rule of three, and you can see how it works from the reverse engineering point of view. If I know that my audit change was that I had a tax deficiency of $100, and there is, in fact, a cell of two, which shows that there is a net entry of... and say I have a competitor, and I can figure out who that competitor is because some of these sets, by industry, for example, are rather narrow or rather specific. And if there is a change in his entry of a capital gain item, I know who it is because these would be grouped by industry. In some cases, the groups are very broad. In other cases, they are not. But there is one way to make absolutely certain that there could be no identification through a reformulation, and that's to include a third taxpayer's data changes because no matter how much I know about myself and my competitor, I have no idea which are his and which are the third person's. But the AIMS Table 37, what's in it? It's just the people who are audited for that year? I can't give you a categoric answer on that, Your Honor. I think it primarily is audit changes, yes. And so are you – is what you're saying is you're auditing the same people every year, so if they appear in a certain place in the statistics one year, then – I mean, I don't understand how you would compare from year to year what the audit changes were. Well, they're done – actually, they're prepared monthly. Okay. They're prepared monthly, and we give them to Ms. Long in CD form monthly. Okay. And they're basically, as I understand it, a kind of task management tool. How efficiently does the IRS expend its resources? That's a really good public service, don't you think? Oh, yeah, I love being on this side of the – I love being on this side of the – Right. But anyway, what we're talking about on this aspect of the case is very, very narrow. The reformulation of data on Tables 37 and 38. 38 is a much smaller table. It reflects much smaller tax payers. I do want to mention the second aspect, which is a broader and more policy-oriented aspect of this litigation, and that's we're dealing with a consent order that was executed in 1976, and times have changed. Okay, that is a good point, but you didn't move to – Was this just a response to you're not providing the data again to Ms. Long? The most candid answer, Your Honor, is that I don't think either side really has covered itself with glory in terms of being timely and moving forward with this. We did ask Judge Peckman to please let us go back under the supervision of the court and redo the consent order to take into account the fact that – and she said, I'm not going to do it because you're not in compliance. You're not in compliance because you won't release the unredacted cells of 1 and 2. Did you ask her that in a motion, the denial of which is appealed to us? Well, we didn't – we have raised the point. We did not specifically raise it as an appeal issue because it is committed to her discretion, and we think that it really would be in the interest of all the parties. The government has thrown a huge amount of litigation resources at this, and so has Ms. Long, and so has this court, and so have a series of district judges. Really, I think that although it might take on a certain air of locking a pair of scorpions in a bottle, it really wouldn't be a bad idea to start over. So this kind of controversy – It might be a great idea. The only question I've got is, is it really properly before us right now? Well, perhaps, Your Honor, you could frame an order in terms of remanding to the district court for her to reconsider that aspect of her order. Yeah, but that – you need a motion to ask the court to do that, and you haven't done that. We did move, and she denied it. But you didn't appeal that to us. No, we did not. You could probably move again at any time to modify the consent order and just have a new case. And if this court were to find that we have satisfied our statutory obligation by turning over everything except the cells of 1 and 2, that would wipe the premise of, I think, of Judge Peckman's order off the map because we would be in compliance with her order to turn over those papers. As I understand it, she had the same problem I'm having with this, which is I don't have enough data in front of me to even understand how you could identify a particular taxpayer. I just don't see that in the record. I understand. I find it mystifying to look at them also. And what I'm satisfied with is this, that it is not merely identification. That is where Ms. Long and that is where Judge Peckman began and ended. Even though they quote the language of the statute, even though they acknowledge the presence of the precedent of Scientology, they keep talking about, I look at this table, which is this massive numbers, this tiny type, and I don't see how you could identify anything. And the answer is to the people who run this kind of program, you can do it. The minute you expand the table, the cell. Well, maybe if you put in front of the district court that program and say, see here, we can identify, and that violates the Haskell Amendment. And I don't want to give you legal advice on how to present your case. It's a fascinating idea to actually run it as a hypothetical. All right, we'll run a program for you. See if you can identify someone and show the district court judge, see with modern technologies and all the developments since 1976, times have changed. But I have a serious question as to whether or not identifying a particular taxpayer by name is what is separating this. It's not a question of facts. It's a question of whether the amendment covers identifying a particular taxpayer, even if you don't know the name. Because the cell of one is of a particular taxpayer. You just don't know the name of that taxpayer. But it is, exactly. And it is precisely that point. It is a cell of one is indeed associated with that and only that particular taxpayer. That's why it runs afoul of the Haskell Amendment. Because the way the statute is written is that cannot be associated with or that does not otherwise identify, i.e., name and address, social security number, a particular taxpayer. Precisely. It is a two-step operation to detox those records, reformulate them, and then be sure you've scrubbed the identifiers. But Judge Wadlow's suggestion is a fact, that perhaps the parties could agree that we can sit down and run a mock program. It might end this kind of problem once and for all. It might. We're past the time, but I had a question, if I may. So my question is on your second issue. Am I correct that the IRS and long have not used our courts' mediation facilities? Well, I think the answer is yes. You have not? Okay. Then the other question I've got is sort of related to that. Like I'm understanding what you're suggesting, not to be that you'd like to mediate, but that what you'd like to do is have this ordered back to the district court where she could, you know, modify her consent order other than by agreement. Well, actually, I had given some thought to the possibility of having a master or a mediator sit down with the parties and try to say where are your current, keeping in mind this is a terribly dynamic area of database creation and data retrieval, where are your current problems, how can we write a consent order that will, to the greatest degree possible, anticipate future problems. You know, early on in this case, we were litigating with Ms. Long whether a bunch of tables were similar to, as a factual matter, tables that were referred to in the 1976 order. Those tables, some of them haven't been created in 20 years. And we didn't appeal our losses on those issues. And it would be nice if we really didn't have to go through this kind of thing. I don't have a problem with the idea of using the mediation auspices of the Ninth Circuit or of the district court. I'm sure Mr. Nelson can talk to that, too. We did sit down, Mr. Nelson and a couple of colleagues of mine, in Washington in 2006 and tried to work out how can we get along doing this, and it just plain fell apart. Well, in the cases I've been in, we normally haven't asked parties will you mediate to argue that inquiry. But sometimes we've issued orders that have said the parties can use our facilities if they want to mediate. But in those orders, the ones I've done, they've said that if any party doesn't want to, they should tell the clerk. And then we don't know which party doesn't want to mediate. Well, I think also part of this has to be for the court to frame the order in terms of how are we going to anticipate future problems based on a very old instrument that was drafted by some U.S. attorneys in the field and which for a time simply disappeared. And this litigation has been going on for a very long time over such things as what are similar records, how do we redact particular types of information, how much access, and how frequently Ms. Long has frequently asked the IRS to provide data on a very, very short timetable, which consumes a huge expenditure of manpower. This seems to me to be appropriate for sitting down and working it out under the auspices of the court. All right. Thank you, counsel. Thank you. I've gone way over. Probably all plus. That's all right. May it please the court, I'm Scott Nelson representing Susan Long. I want to start with Judge Gould's question about whether the parties had made use of the court's mediation facilities. When the IRS took an appeal from Judge Peckman's 2006 order, we did negotiate under the supervision of the court's mediation program until the IRS withdrew its appeal. Then we spent another year negotiating outside of the mediation program with the IRS through lawyers in the tax division of the Department of Justice. It was only after that effort to negotiate, which took over a year and a half, broke down that we brought the second motion to enforce the consent order, which resulted in this appeal. And when we were asked, when both sides were asked by the circuit mediator whether we wanted to mediate this appeal, both sides said, in light of the experience that had led us to this point, that it didn't seem that that would be a workable solution. I'd also like to address, before getting on to the heart of the appeal, which is the return information question, the issue of whether the IRS moved below for the relief that it is now seeking, which is an order requiring the parties to renegotiate the consent decree. What the IRS moved for in the district court was not an order requiring the parties to renegotiate the consent decree, because the meet-and-confer process that Judge Heckman had earlier ordered had already broken down. It asked for the consent decree to be modified against Ms. Long's wishes to limit the types of information that she could obtain. And that was what Judge Heckman denied, saying that she had ordered that the government, if it wanted to move to modify the consent decree, do that before it started redacting things of its own accord and withholding them. That was the respect in which she said it had not complied with her order and had thus not properly moved to modify the consent decree. The IRS has not identified any legal error in her decisions with respect to not modifying the consent decree, with respect to ordering the production of anything other than the cells of one or two that are the subject of its appeal. So I think there are two problems with the IRS's request that this court vacate Judge Heckman's order and require her to order the parties to negotiate a new consent decree. The first is that they didn't seek that relief below, and the second is that they haven't identified any legal error in Judge Heckman's orders Well, what they're saying is that the requirement that they produce information about cells of one and cells of two violate the Haskell Amendment. Right. And she's continuing to require that. And I'm going to turn to that now, because that is the one respect in which they have actually raised an issue below and appealed it, and that, I think, is the only issue that is really before this court. Whether Judge Heckman erred in finding that the data in these tables, even if they were cells of only one or two satisfied the Haskell Amendment, that is to say that they are data in a form that cannot be associated with or otherwise identify directly or indirectly a particular taxpayer. Now, it's common ground that what the Supreme Court held in Scientology is that the term in a form means that the IRS must change the form in which it presents information that would otherwise be return information before it can be ordered to be released under the Haskell Amendment. In other words, if, as in Scientology, someone requests the IRS's files on a particular taxpayer, the court doesn't have any authority to order redaction of identifying information and produce that material. And that's because, as then-Judge Scalia noted in the opinion that was affirmed in Scientology, if you're just having them hand over the same piece of paper that in its original form in the IRS files has identifying information and all you've crossed out is the name, you haven't changed the form. Would you support the position that compiling and tabulating from the form to a table is sufficient to reformulate? If it's compiled and tabulated into a table like this one, where it's amalgamated with every other return that is subject to examination. Well, how about just the cell of one? Pardon me? The cell of one situation. If it's re-tabulated from a return into a table, which is a single return, is that reformulated in your view? Yes. It has changed its form from information in the files about a particular taxpayer to information in a mass of statistical data that can no longer, it's now in a form that can no longer be associated with a particular taxpayer. Well, so you would argue that a particular taxpayer means an identifiable taxpayer? I think that that is indeed what it means, and you can draw that from the fact that the statute says associated with or otherwise identified. Well, it says directly or indirectly. So what would directly or indirectly do to the question of whether it's an identifiable taxpayer? If it's indirectly identifiable. Indirectly identifiable means you can put together the pieces of the mosaic and figure out who the taxpayer is. And that's what the IRS in this case has utterly defaulted in putting in any evidence showing that that's possible here. In fact, we presented evidence below that it was not. We explained that the categories used in these reports are extremely broad categories of income, types of industries such that there's really no possibility that someone could look at this and say this cell of one is a particular taxpayer. But isn't the best protection against that to not have a cell that is directly associated with a particular taxpayer? Because your argument relies upon it's impossible for someone to do it, and there's no evidence one way or the other on that. What Congress did was to write a language that says that it cannot be identified with a particular taxpayer. It didn't say identified with a taxpayer, any particular taxpayer. Well, particular taxpayer I think is a different matter from just some taxpayer in general. It's obvious that a cell of one contains information that pertains to a taxpayer, but not a particular taxpayer, and I think they're reading- It does. It depends on what you mean by a particular taxpayer. It is of a particular taxpayer, correct? It is one taxpayer, but it is one taxpayer that no one else can know which taxpayer that is. So you're saying it's not an identifiable taxpayer. Well, if you don't know whether it's this taxpayer, this taxpayer, this taxpayer, or this taxpayer, I think that it is not within the meaning of the statute a particular taxpayer. But the IRS argues that there's ways to associate it with a particular taxpayer that are out there, and my frustration with that argument is, well, show us. Well, I think when Mr. Cohen was responding to you and he was talking about the programs that are developed, my understanding of what he was referring to is just the IRS's program that has now been developed so that they can identify where the cells of one are automatically and redact them, not that there's some computer program out there that will also then be able to tell some member of the public who looks at this table, you know, who that taxpayer is. And they've never presented any information in the form of evidence. You know, you can comb the excerpts of record here and not find any explanation of how you could possibly look at these tables and identify the particular taxpayer at issue. I put to you a hypothetical. Companies, publicly traded companies, have to report if they're audited or being audited, and they might even have to report the nature of some special expenditure. If they are reporting they're being audited, wouldn't it be a simple matter of some smart person with a computer to look through those tables and find the one that is being reported on and extract the information? You know, the categories are so broad, and, you know, I don't know that there's any basis for saying that there's anything publicly reported that could allow you to link them up. You know, I suppose if some company were to say, you know, that one there in that table we think refers to us, then, you know, that would allow you to link them up, but then it would be the choice of the taxpayer to have revealed the information. I don't think that there's been any showing or even suggestion by the IRS that there's public domain information of the nature that you just described that would allow identification of a particular taxpayer. Unless the Court has any further questions. Well, I have another question along the lines of my question to your colleague in opposition to you, and that is, and I think you addressed this, but I want to be clear, is it your position that the question whether we could order the district court to consider, you know, to sit down with the parties and consider whether the agreement should be modified, that that is not properly before us? That is not properly before you, Your Honor. The IRS did not in the lower court request an order that the parties renegotiate the consent decree, and it has not appealed any such order. And we always can do what I said earlier. We can defer submission in a case, you know, for so many months or whatever, and let the parties decide if they want to mediate together. But I don't want you to say if you want to or not. I take from your comment, as I said earlier, normally when we do that, we do it with an order that lets anyone do it or not without us knowing. They just tell the clerk. But from your earlier comments, it sounded like both parties had addressed that earlier. Well, I think they have, Your Honor. And I'd just like to say one further thing about that. You know, as to the return information issue, it's a sharp legal issue. It's presented to you. It's a yes or no. There's really nothing for us to negotiate on that. We don't want redaction. They do. We disagree about it. Everything else, the other tables that are currently being produced, the other objections that they had to the production of those other tables that they're no longer pressing, there isn't, at this point, I think anything really there to negotiate. There's no issue where the IRS has demonstrated that there's anything wrong with the judge's order that those other items be produced, or indeed the tables 37 and 38 be produced, other than the cells of one or two issue, which is in front of you, that really requires, at this point, any negotiation at all. So, you know, I would frankly urge the court, you know, that it decide the return information issue up or down. Both of us are taking our chances on that. It seems like the issue of the data in Tables 1 and in Tables of 1 and Tables of 2 is framed for us to resolve. Exactly. Okay, I have a question on the legal issue. Okay, in 1989 we decided another long case, a case concerned check sheets. Do you know what those were or are? The check sheets, as I understand it, were documents that the IRS used as they were in the process of moving, you know, information from returns into databases. And the check sheets were pieces of paper that actually contained information that was specific to individual taxpayers and included identifying information. So this court's decision in long, in the 1989 long, I should say, I think it was 89, that the check sheets could not be redacted and produced was basically I think pretty much a direct application of Scientology. When there's individual taxpayer information that in the IRS's files has identifying information on it, the court can't require redaction of it to thus, you know, require the IRS to reformulate it. All right. I think I understand. Are there any further questions? No. I'm fascinated by the fact that this is a 1976 consent decree. It just seems to me it's a perpetual consent decree, and some motion should be properly made to the court. in the law and still holds force, but it is a consent decree. And so I agree that it's not properly before us, but could properly be brought as to whether or not it should be canceled and a new request made under new circumstances and that evaluated. May I respond? Of course. The reason I think that in this case it's less problematic than, for example, in an institutional reform type case to have a 35-year-old consent decree is that what we're talking about here is information, records of the IRS that are clearly subject to the Freedom of Information Act. The only issue that really exists as to whether these reports are subject to production is whether particular numbers within them are return information. But in general, it's clear that statistical reports of this nature are subject to the Freedom of Information Act. I'm not disagreeing with that. It just seems to me that rather than evaluate it on the merits, it's being evaluated under a 35-year-old consent decree. The court never reopens the question of burden, never reopens the questions that would otherwise be litigated if a new request was made under FOIA and it had to be evaluated. That's my only point. And if the IRS were to come in and make a convincing showing with respect to any of those matters, it might be entitled to have the decree modified that it failed to do so in this case. All right. Long v. IRS is submitted.