The Independence of Error
No single source is trustworthy enough to build a decision on, and that statement is not cynicism, it is simply how information works. Every source has a vantage point, an incentive, and a blind spot, and the question is never whether a source is biased but in which direction and by how much. Triangulation is the discipline of combining multiple imperfect sources so that their errors, which point in different directions, partly cancel, leaving something closer to the truth than any one of them offered alone. The old rule of thumb, get it from more than one place, is sound. What matters is understanding why it works and where it fails, because applied naively it can make you more confident and more wrong at the same time.
The core idea is independence of error. If two sources are genuinely independent, arriving at their views through different methods and different vantage points, and they agree, that agreement is meaningful, because it is unlikely that two unrelated errors would happen to coincide. But if two sources are not actually independent, if both trace back to the same original claim, then their agreement tells you nothing at all. It is one source wearing two coats. This is the trap that catches most people who believe they are triangulating: they gather three articles that all cite the same press release and feel reassured by the consensus, when in truth they have verified nothing. The number of sources is irrelevant. What matters is the number of independent sources, and those are far rarer than they appear.
So the first and most important step is to trace each source back to its origin before counting it. When three outlets report the same statistic, walk each one back. If all three cite a single vendor study, you have one source, not three, and its bias is entirely uncancelled. If instead one is a regulatory filing, one is a direct customer conversation, and one is an independent measurement, now you have genuine triangulation, because those three could not easily share the same error. Learning to collapse apparent sources down to their true independent count is the central skill, and it is unglamorous work that almost no one does, which is why apparent consensus so often turns out to be hollow.
The strongest form of triangulation deliberately combines sources of different types, because different source types fail in different ways and therefore check each other well. Set what a company says about itself against what its regulatory filings disclose, against what its customers report in direct conversation, against what the physical evidence, hiring, permits, supply commitments, reveals, against what independent measurement shows. When those diverse vantage points converge, you can hold a view with real confidence. When they diverge, the divergence is not a failure of your method; it is the most valuable signal available, because it marks exactly the spot where someone's story is not matching reality, and that spot is where the opportunity or the risk is hiding.
It is worth naming the failure modes, because they are seductive. The first is the false consensus already described, mistaking many echoes of one source for many sources. The second is false precision, where combining several rough estimates produces a number that looks authoritative but inherits every underlying error; three vague guesses averaged together yield a precise-looking guess, not a fact. The third is correlated bias, where sources appear independent but share a hidden common incentive, for instance an analyst and a trade outlet who both depend on the same advertisers or the same access, so their agreement reflects a shared pressure rather than a shared observation. A rigorous triangulator is always asking not just whether sources agree, but whether they could have arrived at agreement dishonestly through a channel they hold in common.
There is also the matter of what to do when sources genuinely conflict, which is the normal case rather than the exception. The instinct to average them is usually wrong, because the truth is rarely the midpoint of two views; more often one source is simply better positioned than the other on this particular question. The move is to weight by vantage point and incentive. On a question of real deployment, the operator running the system outweighs the analyst who models it from outside. On a question of financial exposure, the regulatory filing outweighs the marketing claim. Ask, for each conflicting source, who is positioned to know and who has reason to shade, and let the weighting follow the answer rather than splitting the difference out of a false sense of even-handedness.
Done well, triangulation is not a mechanical procedure but a habit of mind, a reflexive unwillingness to accept a single vantage point as the whole picture, paired with the patience to assemble the picture from parts. It is slower than trusting a source. It is also the only method that reliably survives contact with an information ecosystem engineered to persuade you. The persuasion is built for people who stop at the first source. Triangulation is simply the decision not to stop there.