ROBERTA

Roberta is an African American woman in her forties who was active in the 1980s crack scene. She now works as a counselor in a community-based program. She was given my name by another person I'd interviewed, and she called me. Among the large number of people struggling for survival in Baltimore, the money for an interview – $20, in my case – is enough incentive to keep the phone ringing off the hook. The street term for research subject payment is study money. But the person who referred Roberta, I think, approved of what I was trying to do with the interviews and said he might tell certain people to whom I should talk to get in touch with me. In this case, I was glad he did.

I asked Roberta where she'd like to do the interview after I picked her up in front of her office. She liked the idea of a lunchroom. We walked in, and I, of course, immediately started obsessing about the sound quality, but she found a table near an electric outlet and pointed out how convenient that would be. We ordered at the counter, waited and chatted about our work, and then took sandwiches and soft drinks back to the table. I set up the recorder and microphone. None of the other customers seemed to notice much, or to care.

As far as the main purpose of the interviews, we were finished in the first few minutes. Roberta noticed the rapid arrival of crack cocaine and the subsequent dramatic change. She also mentioned Jamaicans as key figures right away, talking about how she cooked crack for them. What I want to show is how she complicated things even as she confirmed the key issues.

Before I do that, though, I need to introduce some background knowledge relevant to all the excerpts to come. To foreshadow the conclusion, I'd like the reader to notice that I am at this point smuggling in global material that was not the focus of any interview.

Crack was not the first kind of cocaine to appear in Baltimore. “Powder” cocaine, cocaine hydrochloride, had been around for years. It was the glamour drug of the 1970s and early 1980s for mostly affluent, mostly White users. Because of its high cost, a mixed injection of powder cocaine and heroin, called a “speedball,” was a luxury item in the streets. But as Colombia ramped up powder cocaine production in the early 1980s, and as the glamour market became disillusioned with the negative effects of prolonged cocaine use, simple economics set in and caused prices to drop as demand waned and the glut of product grew.

Affordable powder cocaine was in the streets more than ever before. Users had already learned the easy way to convert powder cocaine into smokeable form by using simple products like baking soda. The smokeable form produced a more powerful effect from a smaller and therefore more affordable quantity of powder. The details need not detain us here; the point is that anyone who could afford a little bit of newly inexpensive and highly available powder cocaine could easily convert it into smokeable form in the kitchen. As Roberta will say, this pattern developed some time before crack appeared.

But isn't smokeable cocaine the same thing as crack? Yes, by some accounts, but by others, it isn't. Roberta differentiates between the two, calling the homemade smokeable stuff “freebase” and the imported, ready-to-smoke stuff “crack.” This is a bit confusing because freebase was a term used in the 1970s and later to mean powder cocaine turned smokeable through a completely different and more dangerous process that produced a purer product.

Just before the transcript segment reproduced here, Roberta had been talking about these earlier developments. She also described what a good cook she was, and how she first cooked for herself to make sure she knew what she was getting, and then for Jamaicans in return for money and a guaranteed supply of the drug. In excerpt (1), I try, in line 1, to bring her back to the takeoff of crack, the main issue I'm trying to get at in these interviews. She has already mentioned Jamaicans as important figures in the crack story.

What always strikes me with analysis of this kind is that you are confronted with how sloppy, or how good, a researcher you are. Local discourse, in an ethnographic study, involves an ethnographer. In this case, I want to know who started the crack epidemic, the different kinds of people in line 1, but I keep the reference vague. I want to hold the space as open as possible, since I'm interested in how Roberta might fill it in. That's the “good interviewer” part.

Now look at the details and notice the “bad interviewer” part. Instead of leaving it open, I implicate her earlier description of Jamaicans: how they were harder, how they weren't playin. The fact that Jamaicans ratcheted up the level of violence in the street markets is mentioned by everyone and documented by a number of major and minor events. Still, it's embarrassing to see how I offered an open question and then aimed implicatures right at the Jamaicans. However, Roberta had already introduced the topic, and I did want to hear more about it, so maybe this was conversational politeness, or indirection, rather than bad interviewing.

I continue fishing for a mention of Jamaicans with folks who weren't around the scene before (line 5), which gets only some, and a suggestion of kids, which in turn gets definitely in line 10. This ambiguity is developed in more detail later. What is clear, though, is that it wasn't only the Jamaicans; it was the entire scene that developed around crack where violence reached new levels. So hard and not playin didn't implicate only Jamaicans.

Line 11 asks where they came from, a pronoun in search of an anaphor: Which “they?” Kids? People not from around here? Hard people who aren't playin? Roberta doesn't know. Probably she doesn't know whom I'm talking about. I suggest she may not know the story. I remember at the time wanting to take the pressure off; our chemistry was good, and by this point we were both enjoying the conversation. But as I look at it now, I realize the utterances in line 13 are a direct challenge to her competence as a research subject in general, and in particular as a witness to a history in which she participated.

Maybe I did realize it unconsciously, though, since in the next excerpt I start a rescue operation by referring to something she mentioned earlier (line 16), and here, earlier is just a few minutes ago. Something works here, because she overlaps and takes off, returning to the theme of the Jamaicans. Let's look at the next continuing segment of this transcript.

In line 17 Roberta returns to the Jamaican theme, though she also mentions people from New York, a dual reference that will come into play in transcripts from other people presented later. I backchannel with a foregrounded oh, signaling that she is on a trail in which I am interested. She then narrates her increasing personal involvement with Jamaicans, including foregrounding and echoing my wow in line 21. Her narration is interspersed with my own overlaps that encourage the flow of the story. She justifies the story – letting strangers move in – in three ways: She would have a steady supply; her children would not want for money; and crack itself drives one to do such things.

I now have my story about a sudden Jamaican presence linked to crack. She has her story about why it happened in her personal world, which explains it with situational attribution.

In the final section of the transcript considered here, I do what ethnographers often do – paraphrase to wrap the story around my understanding of it, to check what it is I think I know after hearing it.

My paraphrase is less about Jamaicans and more about the interrelationships of cocaine use patterns. Jamaicans weren't important because they introduced inexpensive smokeable cocaine; they were important because they were a new and powerful marketing force. Global influence wasn't about a new way of making a drug; global was about a more aggressive way of selling it.

Roberta's way of telling her story made it clear that crack didn't just happen out of the blue, which is how it has been and still is presented in many popular and professional sources. In Baltimore cocaine supplies had already increased, and people had already learned how easy it was to make smokeable cocaine, which Roberta called “freebase,” but which others called “crack.” These changes developed locally in Baltimore, according to her, without the kind of social earthquake that was to come with crack.

What registered in Roberta's life was the appearance of a new marketing force. This new force involved Jamaican distributors from New York. They provided what Roberta calls “crack,” ready-to-smoke cocaine, at a low price, and they moved in and took over local markets with a violence that exceeded normal street practices of the time. They were harder and didn't play. My paraphrase of all this is confirmed by her overlaps, some foregrounded, in lines 33, 35, and 37.

For Roberta, local knowledge is global knowledge because of her personal involvement with Jamaican distributors. But notice how the two of us have shaped the story around just this theme. The actual transcript isn't just about Jamaicans. Roberta's story leaks in several ways, because there are several other stories she could have told that I won't help her build.

Roberta was the best “global” interviewee in all the oral histories, because her local world linked directly with the global changes embodied by Jamaican dealers who appeared in her life. Local discourse can shore up global research by establishing that global events had major impact from the point of view of a local population.

But what about all those side issues? As Roberta said, it's not true that the invention of a simple way to convert powdered cocaine to smokeable form caused the crack epidemic. Users in Baltimore learned to do that well before the takeoff. And it's not true that Jamaicans simply appeared and started the epidemic. The supply of powder cocaine in the streets had already gone through the roof, and smoking it was already widespread. Moreover, the potential stories that didn't go anywhere in the early part of Roberta's and my conversation show that others were involved – other New Yorkers, other local people, children – and indeed, Roberta herself is an example of how people other than Jamaicans helped get things going. She was a cook. Clearly Jamaicans played an important role in the takeoff; but equally clearly, they didn't just appear out of nowhere and do it all by themselves.

Our conversation supported the general story I wanted to check out. Crack did explode onto the scene, and Jamaicans played a notable role in bringing that explosion about. Because of her life experience and my research interest, it was easy enough for Roberta and me to build a narrative together. But the process of building the story, as portrayed in the transcript, may also explain the limits of local discourse. What's interesting here is that the undeveloped stories also index global events, but that reading isn't clear from within the conversation itself. Instead, it requires a prior global analysis to warrant the interpretation.

Before developing this issue further, let me look at a couple of other examples to show that an ethnographer's involvement isn't the only problem.

JAMES

James is an African American in his forties who works as a counselor at a methadone clinic. He and I had talked before about heroin, and our chemistry was good then. I had learned a great deal from him, as I did from Roberta, but James was also interested in academic advice since he was taking courses at a local college. In fact, after the interview I helped him with a sociology paper he was working on. When I first asked him if we could get together and talk about crack, he was happy to do so. I drove to his office for the interview. He accepted the $20 interview fee, saying he'd buy something for his child.

James was fading from the street scene as crack came in. He had reached that tipping point in the long struggle of the chemically dependent where he started to move away from the street. He made it clear that he did not play a central personal role in the development of the crack scene. His knowledge is more distant than Roberta's. He agreed that crack appeared suddenly and boomed rapidly, but he had a different take on whether Jamaicans brought this about. Right before the transcript excerpt presented here, I was talking about how the market changed with crack when compared to the old days of heroin, since a lot of small-scale entrepreneurs could get into the business. I asked if he thought this was true.

James talks about New York and Jamaican together, just as Roberta did. When I follow up with a request for specific ethnic identifiers (line 6), James says most of them were African American, which he foregrounds, but he adds that there were Jamaicans as well, which I echo with hopes for the future of the story I'm trying to build – a hope he obliges with another echo that repeats and foregrounds his earlier statement that there were a lot of them. So far, so good, kind of. But Jamaicans and crack were supposed to have arrived in Baltimore at the same time.

James and Roberta are similar in that both identify Jamaicans as a significant new presence in the Baltimore drug scene, but both note – Roberta somewhat indirectly, James explicitly – that a lot of other people were part of the New York invasion as well.

James derails me in our conversation, however, because he speaks of the New York – including Jamaican – presence in Baltimore as something that preceded crack, not as something that brought it about. He further complicates things by saying that what they brought in was heroin, not cocaine, with heroin foregrounded in line 3 and the modal adverb basically foregrounded in line 9. So the Jamaicans (and the other New York boys) hit town before crack, and they brought in heroin, not cocaine.

This is bad enough, but it gets worse:

James lays out a scenario in which the New York boys not only appeared earlier than crack. They were mostly distributors of heroin; they also had nothing to do with the arrival of crack on the streets of Baltimore. I follow his narrative along with a stunned litany of huh and ok, the lack of overlap likely having something to do with my lack of enthusiasm for the mess he's making of the story I want to tell. I follow with an effort, in line 22, to link the New York boys with powder cocaine, but that just confuses things for a few turns and doesn't help rescue my story anyway. The foregrounded statement of interest on my part in line 24, jeez that's interesting, is an understatement. James is telling a story that no one I've heard from or read about has ever told before – or since, for that matter.

Well, if you don't love counterexamples to your most cherished notions of how the world works, you have no business doing this kind of research, so I start looking for James' alternative story:

James does have an alternative theory: This wasn't about Jamaicans, or anybody else in particular. Anybody and everybody was selling, with anybody foregrounded in line 27, because it was so cheap, again foregrounded. In that same section, he says that people went to New York or to different places to get it. Then in line 31 he makes clear what a hot commodity it was, foregrounding how fast it would go and how easy it was to get rid of. No one in particular brought crack into Baltimore. It was just there, all of a sudden, a booming market where anyone could make money.

At the end of line 31, and in 33, James slides into a topic that brings him to something that Roberta also mentioned – something that wasn't developed in our conversation, either. The crack scene brought young kids into the picture in a way that earlier scenes had not. Noticing crack meant, for both of them, noticing the appearance of kids, probably because of their use as lookouts and delivery boys and in other service roles in the crack economy. Again, this is not news; the increasing use of children in crack distribution was an item reported early and often in accounts of the crack scene. In part, this use of children blossomed with new laws inspired by the legislative and media frenzy after basketball star Len Bias's overdose death in 1986, laws that targeted crack cocaine with severe mandatory sentences. Using juveniles as the human conduit for movement of the drug reduced the risk for adults.

James has to be wrong about crack and the New York boys, Jamaican and otherwise. There is just too much evidence that says that Jamaicans did, in fact, play a major role in bringing the crack epidemic about in Baltimore and several other cities and towns in several regions of the United States. But he is not, of course, wrong intentionally. And in fact, he is wrong for all the right reasons.

James was fading from the drug scene when the crack epidemic started. In fact, he says elsewhere that the first time he heard of crack was when women (almost entirely women, he says) appeared in a treatment center where he was working in the Baltimore-Washington corridor. Recall Roberta's story as well, together with her stories elsewhere about Jamaicans using local women to deal crack. More women were involved in the crack epidemic than in the earlier heroin epidemic, an increase that is also mentioned in other sources and documented by epidemiological indicators. Roberta and James also illustrate another frequent observation about crack: Major outside distributors operated by infiltrating – James's word – African American communities, using local people to get started, and then bringing them under control to dominate the market.

James is basing his story on what he saw, but what he saw was the crack epidemic after it was off and running, with dozens of entrepreneurs, widespread use of children, and a lot of people motivated by the opportunity for economic survival in an increasingly dismal 1980s inner-city world. Roberta, in contrast, tells a story about how the epidemic started, a story based on her personal involvement in it as it developed.

James also may have based part of his story on a condition that inspired the Jamaican distributors. According to the story that I describe in Agar 2003, Dominicans dominated the early crack market in New York. Individual entrepreneurs from other cities along the I-95 freeway corridor, such as Philadelphia, Baltimore, and Washington, came to New York to purchase crack for resale. Supposedly, Jamaican marijuana dealers spotted this pattern and began to set up distribution systems to gain control of these new markets. They developed their crack business by avoiding competition with the Dominicans and developing markets outside the city.

And what about the heroin? James was an active heroin addict before crack arrived, so his stories about New York boys, including Jamaicans, bringing heroin to Baltimore set up some new questions. The early 1980s was a period when heroin from a new source, Southwest Asia, was flooding U.S. and European markets. This source blossomed in Afghanistan and Pakistan during the civil war between Soviet occupation troops and mujahadeen and developed in part, like other illicit drug production systems of the latter part of the 20th century, because the trafficking helped fund anti-communist forces. It's not far-fetched to imagine that this new influx of heroin to the United States from Southwest Asia created new opportunities for entrepreneurs who wanted to distribute it from New York down the Atlantic seaboard. So New York boys, including a growing number of Jamaican immigrants beginning in the early 1980s, might well have taken on this role and brought the new heroin into Baltimore. James is probably right when he complicates the crack story by noting that Jamaicans, and other New York boys, were already serving Baltimore markets with heroin before the crack epidemic.

The conversation with James leaks in many of the same ways as the conversation with Roberta –- prior drug trends, not just Jamaican distributors, involvement of youth. But he also directly contradicts the global story that I want him to support, as Roberta in fact did, that Jamaican distributors played a major role in bringing about the crack epidemic in Baltimore.

James was, by his own account, a more distant observer of the crack scene. A plausible way to understand the limits on his story is this: From a distance, he saw several different aspects of the changing illicit drug use scene, like a photographer creating a time series with a snap here, a snap there. To use a linguistic metaphor, his story is a coherence constructed across a series of disconnected observations encoded by perfect-aspect verbs, whereas Roberta's is a story subsumed under a single imperfect verb. James's last imperfect verb was the arrival of Southwest Asian heroin when he was still an addict, and that pushed to the forefront as he talked.

Is this just a way for the ethnographer to weasel out of a problematic interview? I don't think so. The reason I don't is that global research – strongly supported, not only by ethnographic sources – shows why James was wrong about some of the details of crack, and probably right about the connection between Jamaicans and heroin. As with Roberta, it turns out that there is more of the global in the conversation than just the issues I was after. And as with Roberta, those global issues were seen and interpreted in this article after global knowledge was gathered from other sources, not with understandings gleaned from the conversations themselves. But unlike Roberta's case, the limits on James's local discourse derive from his social and psychological distance from the developing crack scene.

ANDY

Let's take one final look at a crack conversation, this time with an Anglo-American administrator of drug services. Andy, as we'll call him, is in his fifties and is a former user of many substances, including heroin, beginning when he was a high-school student. We had talked on a previous occasion about illicit drug trends in the metropolitan area, and I had worked with youths in one of the programs he supervised. I drove to his office to conduct the interview. As with Roberta and James, the chemistry was good, and Andy has more experience than almost any of us at being interviewed, by both media and researchers. I did not offer to pay him for the interview, and I doubt that he expected it.

In our conversation, Andy was most interested in telling me about the city/county task force he participated in. He is proud of it because, when he heard about crack in New York, he decided that there might be time to do something about it before it arrived in Baltimore. He showed me documents, press clippings, videos his group had made, and pamphlets they had prepared. He felt that crack had had a lesser impact than it might have had without this intervention.

I was sympathetic to Andy's pride and impressed by this rare event in the history of illicit drug epidemics in the United States – a genuine effort to prevent something rather than reacting with harsh laws after the fact. I felt bad as I kept trying to shift the conversation from the task force to how crack eventually did, in fact, devastate the city. I didn't want to rain on his parade. My mixed emotions are apparent in the mitigating phrases in the question that opens this excerpt:

Andy, like Roberta, points out that smokeable cocaine was available before a distribution system brought crack to the city. Smokeable cocaine was easy to make out of powder cocaine, and powder cocaine was all over the streets. Notice that Andy does not mention Jamaicans, just New York, which I had topicalized in my question in the first place. The next excerpt occurred later in the transcript.

Now, in an interesting shift, one that again parallels Roberta, he talks about already to sell crack in line 1 and attributes that to the now familiar New Yorkers. I ask, in a politically correct way, if those New Yorkers were Colombian or Jamaican or who knew who they were (line 8). (It is amusing that this “politically correct” switch got thrown only in the conversation between two old white guys.) At this point, Andy's distant position vis-à-vis the dynamics of the street scenes comes to the fore with his straightforward statement that I don't think anybody knew (line 9). His closing comment in line 15 personalizes it with I and includes the additional assertion that he didn't know who was controlling heroin, either, at that point in the 1980s.

Recall James's conversation, where he also spoke of new heroin coming into Baltimore right before crack, something he also attributed to the New York boys, including Jamaicans. Recall also the description offered earlier about how new conduits for Southwest Asian heroin were opening up about this time. What was happening around Andy, in the 1980s, was that distribution systems were shifting and smokeable cocaine was spreading, as was the new Southwest Asian heroin. As a program administrator, he was distant from the details of the dynamics. He often attributes what he knows to law enforcement. Important for him was that a major new pattern was developing for which he needed to program. Andy, like James, was distant from the actual social processes within which crack took off, but he was working with “local relevance” constraints different from James's because of what he needed to do as a program administrator.

LOCAL DISCOURSE AND GLOBAL RESEARCH

In the end, then, is local discourse limited in a project that aspires to global analysis? The three cases examined here all had different limits. In fact, most local discourse probably shows the same kinds of limits treated in this article. Local discourse will not provide much material on the distant locations and processes that are essential to global research. This was certainly true in the studies Reisinger and I have done. In spite of the obvious intelligence, creativity, knowledge, and articulateness of interviewees, and in spite of good chemistry between interviewee and interviewer, the interviews themselves were limited when measured against global research goals.

Why? Several reasons have come up in the discussion of specific cases in this article.

Reason 1: It's our own damn fault. The excerpts show the relevance of Ochs & Capp's (2001) work on the “living narrative.”. The oral-history transcripts examined here are records of stories under construction with a co-narrator, not the neat narrative of a practiced teller of a tale worn smooth by frequent use. I, the interviewer, had a coherence that I wanted to build (Linde 1993), but the “multiple tellers” and “embeddedness” and “open linearity” and “fluid moral stance” that Ochs & Capp describe limit narrative development on the part of a single speaker.

The limits on the interviewee are exacerbated by the hegemony of the interviewer. He or she is probably paying the interviewee, and he or she is usually in the pole position in the Q/A adjacency pair. Imposed limits are topic-based as well, depending on what we interviewers want the living narrative to become. There we are, in the transcript, in all our shame and glory. For anyone who knows how to “see” the details, the co-construction of the local discourse and the ethnographer's role in its successes and failures are pretty obvious.

In the samples presented here, I'm determined to talk about the crack explosion and the role of Jamaicans in bringing that explosion about. The transcripts, as described earlier, “leak” all over the place and go outside that topic framework. The leaks show that, for the interviewees, there was more to the crack epidemic than dramatic increases in use and Jamaican distributors.

Weren't the additional topics that leaked out important for global research as well? For instance, the oral histories make clear how, at the local level, crack distribution was not nearly as controlled as the old heroin system from the 1960s and 1970s. Jamaicans were only one of a number of “kinds of people” who were identifiable as crack took off, and within that (and other) clusters, there was no single organization. And “New Yorkers” looked like a more salient category than “Jamaicans,” before, during, and after crack. Did I encourage any of this? No. I wanted a yes-or-no answer about Jamaicans. Interviewees were offering me some anti-Aristotelian advice. I didn't take advantage of it.

Another feature of the “living narratives” is that the crack epidemic developed in the context of product flow and marketing of other illicit drug commodities. Crack was not just something that Jamaicans invented and then brought in and sold as brand new. The cocaine and heroin markets were in motion at the time, and the use of smokeable cocaine was already known and practiced. Illicit drug epidemics, and illicit drugs, cannot be isolated, even though the drug field, not to mention this author, have done that repeatedly throughout the history of drug research.

In addition, the transcript fragments presented in this article make clear how economically driven the crack epidemic was. All three narrators weave into their oral histories, in their different ways, how one important thing about crack was that a lot of people could suddenly make some money. By the 1980s, many neighborhoods had become what one Baltimore researcher called “warehouses for the poor.” All illicit drugs offer opportunities in the underground economy, but with crack, more people than ever before could get into the business and earn a little cash, all the way down to the street level. A little cheap powder, available to anyone, could be easily converted by anyone.

Roberta and James also foregrounded the increasing involvement of youth in the local distribution system. As mentioned earlier, repressive laws passed shortly after crack appeared increased risk for adults. At the same time, the dismal inner city of the 1980s offered even fewer work opportunities for adolescents than for adults. In such a world, an increase of youth in the most publicly visible – and hence most vulnerable – roles in the crack market made unfortunate sense. Both Roberta and James also described a major increase in the involvement of women, a contrast to the male-dominated drug scenes that preceded crack. There were many more single mothers like Roberta trying to make ends meet in the 1980s.

All these issues – multiple distributors, multiple products, economics, increase in youth and female participation – are keys for the global side of the analysis as well. For the present, I just want to note that whatever Roberta, James, and Andy had to say about these issues got nudged to the side because they weren't topically linked, in the interviewer's mind, with the focus of the oral history. However, I also want to note that global research, using non-ethnographic sources, has allowed me to interpret what they said.

Thus, one reason for the limits on local discourse lies in the fluidity and path-dependency of “living narratives.” Does this mean that local discourse simply needs to open up into an infinitely long process so that all possible topic frameworks can be explored? Under these conditions, would the original idea work? That is, can we derive global explanations directly from local discourse?

No, because of Reason 2 for limits on local discourse: relevance. For example, as a former user and a policy person in contact with police sources, in our conversation Andy put together his version of what was happening. His version was incomplete, like Roberta's and James's, though like them both, he was aware that a new drug had exploded onto the scene and that new marketing forces were at work. In the end, though, what difference would it have made for Andy to know more about the global dynamics of the crack trend? Or what difference would it have made for Roberta or James? I'm not sure it would have made any difference at all, as far as what they needed to do in their local moments. Roberta needed to get crack. James needed to counsel a new breed of client. Andy wanted to prevent the drug from rolling south into Baltimore from New York City. How and why the situation came about was less important to them than the fact that it had. They all, in their different ways, had to deal with it.

The global dynamics that I was after probably wouldn't have mattered much to the three people involved in crack scenes. Global explanations just aren't very useful to those dependent on the drug or to those charged with intervention. To metaphorically extend Sperber & Wilson's (1995) work, transnational events just weren't pragmatically “relevant” to their local moments. Who cares about history and transnational influences when it's a hassle just to make it through the day? Drug-dependent persons, overworked counselors, and underfunded administrators have more pressing problems to deal with.

This is an interesting problem. A limit on local discourse is really about the lack of relevance of global analysis to more pressing needs of everyday practice. Global research just isn't that useful to “the community,” if I may use that massively ambiguous term. Or is it? I think it's useful as a kind of clinical political intervention. Providing locals with a global perspective, though it probably won't help them make it through the day, does give them a view of things that they didn't have before (if you've done your job right), and perhaps some material for those in that community who struggle against the powers that be. Maybe what we can do is to offer a mixture of psychiatric reframing, critical theory, and outside agitation.

Let me offer an example from some past research. In the 1980s, I did a study of independent truckers. After researching archival material to find out the story behind the rules and regulations affecting their “leases” with trucking companies, I gave a talk at an independent truckers' meeting. Everyone knew about “the lease,” but no one knew where it had come from – including me, until I researched it. The response was one of the most positive I've ever received from an audience. Knowing where the lease came from didn't help with their daily routine, but it was enlightening and interesting to that audience anyway.

Maybe relevance helps us understand limits on local discourse, but relevance doesn't mean that groups who are part of a global story aren't interested in hearing it once you, the researcher, have figured it out. Besides, who's to say that a global researcher is the only one who can provide this service? Who's to say a “local” hasn't already done so? There will always be locals who think globally. In the old days, such people turned into what we used to call “key informants,” a kind of person about whom Morris Freilich wrote long ago in a book called Marginal Natives. Nowadays such people are more likely to be “colleagues” or “consultants,” as exemplified by several ex-addict global theorists of the drug field from whom I've been lucky enough to learn over the years. And anthropologists aren't the only ones who have noticed such people. Numerous studies in the diffusion of innovation, for example, describe cutting-edge innovators as social types with one foot in the local world and another in the larger world (Rogers 1995). They are few in number, but they bridge the two worlds, and through them flows the “news” about innovations.

In spite of such counterexamples, though, the material in this article – not to mention many other projects, not to mention everyday life – shows that it's unlikely that local discourse will serve up the information we need for global analysis. People are just too busy with their everyday affairs. Local discourse has to link to global analysis, about which more in a moment. And local discourse reflects and may well offer views on transnational events, but they probably won't do the job, either. A perfectly understandable reason for these limits is the lack of relevance of the global story.

In addition to relevance, new limits appear when we consider deixis, Reason 3 for limits on local discourse. The oral histories confirmed that the crack epidemic was indeed an explosion. but when it came to the role of Jamaicans, the results weren't so neat. Roberta supported the idea strongly and in several places in her interview. James contradicted it, in the sense that he said Jamaicans were active, but earlier, and they had nothing to do with increasing crack use. Andy didn't contradict the idea; instead, he didn't know the details, though he did briefly mention Jamaicans in a part of the conversation not used in this article. These differences among the stories, as noted earlier, lent themselves to an analysis in terms of social-psychological distance of the narrator from the dynamics of the crack epidemic.

Because the crack scene moved and changed so dramatically in the beginning, an oral history could change depending on the time/space coordinates of the story, the deixis. Deixis is also the right term for how the oral histories changed depending on the proximity – physical, social, and psychological – of the narrator to the events being narrated. Thus, we have different spatio-temporal and psycho-social locations, different spatio-temporal coordinates of events observed, and different oral histories.

These limits on discourse are the down side of the concept “emic.” The value of traditional ethnography lies in exploring a particular perspective on the world; however, a global analysis includes multiple perspectives of different types distributed across widely different social locations. As historiographers sometimes note – and as recent “embedded” news coverage of the Iraqi invasion showed – in medias res isn't always the best place to figure out what is going on and why, though(as roughly a century of ethnographic research shows) it's the right place to learn what it means when you have to deal with it.

FRACTAL ENDING

And so, in the end, I land in an argument that local discourse is limited for several obvious and good reasons, always with the exception of the occasional resident local/global analyst. The resident local/global analyst, like any others, including the researcher, offers global explanations that serve particular ideological interests. In fact, the explanations might be flatly wrong, though probably not with intent to deceive. They might also be paranoid, like the occasional early-1970s theories that the CIA introduced methadone to kill minorities – understandable with the Tuskegee syphilis experiment in the background, but off the mark nonetheless.

The “living narrative” that is the self-organizing conduit for connecting ethnographer and interviewee follows in part, or maybe mostly, along the lines of ethnographer's interest. With global research goals pushed to the forefront, an ethnographer asks people to talk about things they probably don't know much about. The reasons they probably don't know include a lack of pragmatic relevance for daily life and the limits imposed by a particular material, psychological, and social location in space and time.

But didn't the local discourse analyzed here contain several global themes in addition to the ones I focused on? Yes, it did.

Maybe it is difficult to get much of the global directly out of the local. But once you do get some of the global, the local suddenly makes a lot more sense. Oddly enough, this is a lesson I'd learned without recognizing it in my earlier study of independent truckers (Agar 1986). I didn't get around to the kind of analysis used in this article until after the study was done, because it was only after I understood from archival work the regulatory and economic environments of trucking that I was able to show why a particular transcript fragment was so interesting.

Why is this? One way of thinking about it lies in the concept of fractals. One famous example of fractals goes back to the concept's inventor, Benoît Mandelbrot, who noticed that fluctuations in prices on the commodities market looked similar at whatever time scale one drew the graph. Whether the chart is for a day, a month, or a year, there are some similarities in patterns. Another famous example is the problem of measuring the coastline of England. From a satellite photo, it looks ragged. If one flies over it at low altitude, it looks ragged in similar ways. If one puts his nose on a stretch of shore, it looks ragged in similar ways too.

The basic idea here is that some process works over and over again, and it takes the results of its previous step as input to the next step, then the results of that step as input to the following step, and so on ad infinitum. The logic is both iterative, occurring over and over again, and recursive, in that it uses its own results to run again. The result is that a structure is built up, a structure of patterns at different levels of scale, in which the patterns are produced by repeated applications of a simple algorithm.

What does this have to do with the oral histories? I can't help but think there's some relevance, because of anthropology's longstanding fascination with “themes.” We're always after patterns that replicate across domains and at different levels of scale. If X is true for “our people,” then we should see X in life stories, daily situations, religious systems, subsistence practices, and so on. We seek the “patterns that connect,” to use Bateson's phrase.

The fractal concept sets up an interesting way to think about global research with reference to local discourse. How do we show a link between distant global events and local realities? The answer: We look for fractals –patterns in those different locations generated by the same algorithm.

This all sounds pretty rich. It sounds as if the transcripts are thick with global patterns, and they are, as described earlier. In fact, this implies that if a global pattern matters, it will in fact appear locally, because the same algorithm is operating. The problem was that I couldn't explain the local fractal without the global research. The algorithm didn't start locally. It was designed and initiated at some distant center of power. The origins of those global fractals were learned from historical, archival, media, academic, and statistical sources that carried information about those centers of power.

It is, of course, hard to pin down what came from where. Prior experience can't be discounted, either, since I'd done a lot of ethnography with drug-dependent persons over the years. But with a goal of figuring out the interactions of global and local processes, together with a goal of explaining major changes that affected many different sites, local discourse lost its central position in this research. The effort remained ethnographic in its epistemology, but not ethnographic in its genres of primary data.

If one wants to locate fractals in global research, it makes sense to look for them first in data with global coverage, not data from a single site. The fact that a crack epidemic had some unique characteristics in Baltimore was less interesting than the fact that the crack epidemic in Baltimore had characteristics like crack epidemics all over the United States. And the fact that crack epidemics all over the United States occurred at about the same time meant that the explanation wouldn't be found in Baltimore. As our project developed, and as we looked at more and more cases of illicit drug epidemics, we realized we weren't studying Baltimore much at all any more. And when we did listen to Baltimore voices, we recognized the same algorithms we'd already identified as relevant to many other social locations.

So what are these algorithms that work over and over again at different levels of scale? In this case, candidates don't look difficult to find, though here I can only suggest them. For example, “reduce employee costs at the lower end of the job scale” is an algorithm that began affecting the United States in the 1970s, culminating recently in the “jobless recovery.” The origins of such an algorithm are found in locations distant from impoverished neighborhoods – in government and business, for example. In a poor neighborhood, the same algorithm helps us understand an adaptive response – an increasing turn to the underground economy.

As another example, consider the algorithm “implement Draconian laws to punish adult drug users.” The legislative hysteria that the crack epidemic produced is well documented. One adaptation to this algorithm at the local level would be greater avoidance of arrest risk, achieved in part by using youth in drug-market roles where vulnerability is greatest – in public places with evidence in their possession.

On the one hand, then, fractals suggest a way to go after local/global interactions by showing that the same process, or algorithm, is producing similar patterns in different social locations and at different levels of scale. The algorithms begin in social locations where power allows their design and forces their iteration. They end in individual lives where survival depends on adaptation. The patterns are not identical, but they are generated by the same process. The local/global patterns connect through the iterated algorithm.

On the other hand, fractals suggest a way to show the limits and value of local discourse. The initiation of an algorithm is most visible at centers of power, best learned in material that documents activities in those centers and spans multiple sites. As Reisinger and I learned in our research, such information is available in archives, legislative hearings, media, and aggregate data sets. But local discourse, as the examples in this article show, expresses the algorithm as well, by way of expressing its effects on local practice.

In a nutshell, fractals provide a metaphor to show global/local links. Global data sources explain the origin of the algorithms. Local discourse shows their iteration and their effects on agency at the individual level. It seems to be an idea worth pursuing, especially since it reintegrates the time-honored anthropological notion of “theme” into contemporary research. In the cases analyzed here, it shows why the oral histories – both the ethnographer's topic framework and the leaks that occurred – reflect transnational events rather than explain them.