Megan Skiendel: Zara, hey — how's the week treating you, are you recovered from that content sprint?
Zara Reyes: Barely. I'm running on cold brew and spite. Hand me something interesting.
Megan Skiendel: Okay, I'm going to give you a scenario and I want your gut reaction. A Civil War general sits down — thirty years after the war — and writes his memoirs. He was there. He made decisions that cost thousands of lives. Technically, historians call that document a primary source. And also — simultaneously — corrupted evidence. Both things, fully, at the same time.
Zara Reyes: Wait — both? Like, not 'it depends on the question,' but structurally both at once?
Megan Skiendel: Structurally both. That's what we're sitting with today — the source hierarchy, how it actually works, and whether its defining feature is reliability or something else entirely.
Zara Reyes: No but — so the question is whether the hierarchy is a truth tool or a power tool. That's the thing we're actually trying to figure out.
Megan Skiendel: Exactly that. And the general is where it cracks open — because thirty years of memory degradation and thirty years of crystallizing your own legacy, those aren't the same problem, but the hierarchy kind of launders them together as if they are.
Zara Reyes: Right — but the part that doesn't fit is that Leopold von Ranke built external and internal criticism specifically to handle this. Like, that's the whole nineteenth-century professionalization move — you assess authenticity, then you assess intent and bias. So the framework knew about the problem.
Megan Skiendel: It knew about the problem and still made proximity the default weight. Ranke's contribution was rigor — the insistence that you go to the primary source, you don't trust secondhand synthesis. But the hierarchy he anchored still says: closer to the event, more reliable. And the general breaks that. He's maximally close. He's also maximally motivated to revise.
Zara Reyes: So the framework is a heuristic, not a law — like, it gives you default weights and then you're supposed to adjust. But the adjustment for intentionality is the hard part.
Megan Skiendel: And here's where I think it gets genuinely unsettling — primary sources are the raw materials, diaries, government records, trial transcripts, letters. The definition is about creation conditions, not honesty. So the same category holds Anne Frank's diary and a general's reputation-laundering memoir. They both qualify.
Zara Reyes: That's the contradiction. The hierarchy is supposed to be a reliability tool, and its own categories contain this — yeah, that's the thing we have to turn over.
Megan Skiendel: And that's the crack — so what that means is the category 'primary source' is doing two jobs at once, and they don't always point the same direction. One job is: how close was this to the event. The other job is: why did this get made.
Zara Reyes: Okay, those are the two variables. And they're not the same thing.
Megan Skiendel: They're not. Think about a grocery receipt from the day of a crime versus a note someone wrote explaining what they bought that week. The receipt is more convincing — not because the person who wrote the explanation is lying, but because the receipt wasn't made to convince anyone of anything. It was made to record a transaction.
Zara Reyes: That's — yeah. That's the whole thing, actually. A tax ledger from 1847 carries more weight than a memoir from 1847 not because the memoir writer forgot — it's because the ledger had no reason to persuade. It was made for administration.
Megan Skiendel: Proximity plus intentionality. Those two variables explain the entire hierarchy. That's the engine.
Zara Reyes: Wait, so external and internal criticism — Ranke's actual tools — are just the formal method for measuring each variable? Like, external criticism asks 'is this document what it claims to be,' and internal criticism asks 'okay but why was it made and how honest is it.'
Megan Skiendel: Exactly that. External first — authenticity, authorship, date, provenance. Then internal — intent, bias, reliability of the claims inside. You can't do internal until you've done external.
Zara Reyes: And Anne Frank's diary is — I mean, it's almost the perfect test case for why these variables matter. Real time, no posterity to perform for, no chance to go back and revise. The intentionality is just: survival, witness. Not publication.
Megan Skiendel: That's precisely why it carries the evidentiary weight it does. She couldn't have known anyone would read it. There's no retrospective self-justification baked in — which is the thing that corrupts the general's memoir structurally. Same category, completely different intentionality score.
Zara Reyes: So the complication — and this is where I keep getting stuck — is that contemporaneous official records score high on both variables, close to the event and low motive to persuade. But, wait, actually, no — low motive to persuade isn't the same as no motive. A government memo drafted in 1863 has institutional purpose baked in from the moment it's written.
Megan Skiendel: Honestly, that's the part the hierarchy undersells. Contemporaneous doesn't equal neutral. It just means the motive was present-tense rather than retrospective. Different shape of bias, not absence of it.
Zara Reyes: So the two variables aren't a guarantee — they're a starting position. Proximity and intentionality tell you where to place your default trust, and then external and internal criticism are how you adjust from there. That's the whole engine.
Megan Skiendel: And that's what makes the memoir problem messier than the hierarchy admits — because it's not just 'bias present at time of writing.' It's that the bias compounds. Daniel Stowell, working on the Papers of Abraham Lincoln, looked specifically at this — how retrospective accounts distort not because people are lying outright, but because memory is reconstructive. You don't retrieve a memory like a file. You rebuild it. Every time.
Zara Reyes: Wait — reconstructive, like you're actively assembling it each time you recall it?
Megan Skiendel: Each time. Which means every recall is also a potential revision. And Stowell's point was that documentary editors working Lincoln-era reminiscences — letters, diaries, accounts written decades later — had to build explicit protocols just to triangulate them. You couldn't treat them as direct evidence. You had to run them against contemporaneous records to even know what you were working with.
Zara Reyes: So the distortion was systematic enough that it required a formal protocol. That's not a caveat — that's a structural problem.
Megan Skiendel: Systematic enough that Stowell published on it in 2005 specifically because editors kept running into it. The general writing his memoir thirty years out isn't just forgetting details — he's actually, I mean — honestly, he's reconstructing the battle through every intervening year of political consequence, reputation management, maybe a court martial he survived. That's not memory decay. That's memory shaped by outcome.
Zara Reyes: Shaped by outcome — okay that's the mechanism the hierarchy doesn't capture. The time gap isn't the problem by itself. It's that the gap fills up with motive.
Megan Skiendel: Right — but the part that should be making us uncomfortable is that we just spent ten minutes establishing why contemporaneous official records are more trustworthy. And now I'm not sure that holds.
Zara Reyes: No, wait — say that more. Because the hierarchy would say the government memo wins over the general's memoir every time. Contemporaneous, institutional, no retrospective drift.
Megan Skiendel: And the counterargument is — look, the government memo written in 1863 had its motive locked in at the moment of drafting. It wasn't added later. The intentionality was baked in before the ink dried. So we're not comparing 'biased retrospective' versus 'neutral contemporaneous.' We're comparing two different shapes of bias — accumulated personal reinterpretation versus original institutional purpose.
Zara Reyes: So — actually, which is worse? Like, genuinely. I don't know.
Megan Skiendel: Honestly, neither is cleanly worse — and that's the thing the hierarchy papers over by just ranking them. It assumes institutional contemporaneous beats retrospective personal, full stop. But a memoir at least might defect from what the institution needed it to say. The distance creates room. The 1863 memo had no room — it was the institution speaking to itself, for itself.
Zara Reyes: Which is a wild thing to sit with — and it only gets weirder when you realize the hierarchy doesn't just rank by reliability. It ranks by who had access to record-keeping in the first place. That's the part that's coming for us next — and the Kennewick Man case is where it breaks completely open.
Megan Skiendel: And that's the thing that actually breaks it — because it's not just that certain voices got ranked lower. It's that the ranking system itself was built by the people who had pens. Literally. If you didn't have access to writing, your account doesn't exist in the hierarchy at all.
Zara Reyes: Which is exactly what happened with Kennewick Man. 1996, remains discovered near Kennewick, Washington — Columbia Plateau tribes say: he's ours, he's the Ancient One. Scientists say: we need to study this. And the written-record hierarchy has nothing from the tribes because — why would it? They weren't the ones drafting federal reports.
Megan Skiendel: Twenty years that dispute ran.
Zara Reyes: Twenty years. And oral tradition — continuous, intergenerational, literally the community's living memory — carried zero institutional standing against written scientific records.
Megan Skiendel: Until April 2016 — U.S. Army Corps of Engineers confirms genetically that Kennewick Man is related to modern Native Americans. The oral tradition was right. The hierarchy had just had no mechanism to register it.
Zara Reyes: No but — wait. That framing is almost too gentle. It's not that the hierarchy couldn't register it. NAGPRA, the 1990 federal law, created the legal framework for repatriation. The system had a tool. It just required genetic evidence to actually activate the oral tradition's standing. Like, the community needed a written scientific result to validate what they already knew.
Megan Skiendel: That's — honestly, that's the cruelest version of it.
Zara Reyes: Right — because the material evidence, the genetics, the skeletal remains — that stuff exists independently of who had a pen. It wasn't created to persuade anyone. And that's what finally cracked the case open. Not a newly discovered document. Bones.
Megan Skiendel: And it resets what we were just saying about contemporaneous records being the gold standard — because material evidence doesn't have an intentionality problem at all. It wasn't made to say anything. Which, actually, means it sits outside the whole proximity-plus-intentionality engine we just built.
Zara Reyes: Okay that's — I mean, that's a real complication. Because the hierarchy we described assumes sources were created by someone with a purpose. But a femur from 9,000 years ago wasn't created to testify.
Megan Skiendel: And yet explorer and trader accounts — written, contemporaneous, full of colonial bias and literal financial incentive to misrepresent land value — those carry primary source weight in aboriginal rights litigation right now. Today. Because they're written.
Zara Reyes: That's — lowkey the hardest fact in this whole conversation.
Megan Skiendel: The hierarchy didn't malfunction in the Kennewick Man case. It functioned exactly as designed — for a world where the only people making records were the ones with institutional access. The design was the problem.
Zara Reyes: So it's not a truth tool that occasionally gets corrupted by power. It's a power tool that occasionally produces truth. Those aren't — no, those are genuinely different things with different fixes.
Megan Skiendel: And the fix isn't just 'add oral tradition to the pile.' It's asking who had access to record-keeping at the moment of the event — and then applying more scrutiny to the sources that benefited from that access, not less.
Zara Reyes: Which means internal criticism — the part of the method that interrogates intent and bias — has to work harder on the sources the hierarchy trusts most. Not just the retrospective ones. The contemporaneous colonial record sitting at the top of the stack.
Megan Skiendel: And that's the thing I can't actually settle — because if you apply that same scrutiny forward, to AI systems trained on primary sources, you get something genuinely unsettling. They'll weight proximity and document volume the same way the hierarchy does. Joseph Lister's writings, explorer accounts, colonial administrative records — all of that goes in as authoritative input. Oral tradition, material evidence, community memory — structurally lighter in the training data. And the system presents itself as historically authoritative. That's not a future problem. That's the architecture.
Zara Reyes: It inherits the blindness and loses the disclaimer.
Megan Skiendel: Exactly. At least von Ranke's framework — external criticism, internal criticism — built in the instruction to interrogate. An AI character built on primary sources doesn't come with that caveat attached. It just speaks.
Zara Reyes: I'm stuck on the same question and can't answer it — is the hierarchy a heuristic we've stopped treating as a heuristic? Because the whole NAGPRA case, twenty years, genetic confirmation in 2016 — that's what it costs when the default weights stop being adjustable. But I don't know what the adjusted version looks like in practice. Like, I genuinely don't.
Megan Skiendel: Neither do I. And I think that's probably the honest place to leave it. Not 'abandon the hierarchy' — it tracks real cognitive decay, that part holds. Just — stop letting it be self-justifying.
Zara Reyes: Good talk. Genuinely unsettled in the best way.
