Seven questions to ask any simulation vendor — including LEXUN.
Decision and simulation tools make claims that are hard to check from a demo. The questions below are the ones that turn a claim into an artefact you can inspect. LEXUN’s answers are here, each with the page that proves it; put the same questions to anyone else and expect the same kind of reply. This page compares approaches, not companies, and makes no claim about what any named vendor does.
The seven questions, with LEXUN’s answers
| Ask | What a good answer looks like | LEXUN’s answer, and where to check it |
|---|---|---|
| Can the result be reproduced? | The same inputs and seed give the same output, on your machine, without the vendor’s help. | Yes. Every result carries its seed, model version and input hash; re-run any published result and check a sealed package in your browser. |
| Is uncertainty visible? | Ranges rather than points; the sample size and the standard error disclosed. | Every probability is a band; the binomial standard error is shown; a point estimate is refused by design. Method → |
| Is the method published? | The model, its inputs, its recurrence and its known weaknesses, in writing. | Documented per model on the Accuracy Centre and each guide page, weaknesses included: runway, career, purchase. |
| Are predictions frozen before outcomes? | A dated entry with a resolution criterion and source named in advance, that cannot be edited afterwards. | The public register is append-only and hash-chained; withdrawals keep the original entry; misses are published at full size. |
| Is accuracy scored publicly? | A calibration record computed only from resolved outcomes, with the sample size beside it, and no percentage before the sample can carry one. | Published from an honest zero: no accuracy percentage until ten outcomes have resolved in a domain, and editorial forecasts are never pooled with model outcomes. The record → |
| What happens outside scope? | A refusal, stated as such, rather than a confident number for anything. | “Insufficient evidence” is a real output and out-of-scope domains are declined: three released models, nothing improvised. What each answers → |
| Where does my data go? | A precise list of what leaves the device, to whom, and why; not a general assurance. | Simulation inputs and saved decisions stay in your browser. Forms and the optional Copilot are the two network flows, named with their processors on the security page and in the privacy notice. |
Every LEXUN row above links to the artefact rather than to a sentence about it. If an answer here ever stops being true, the release gate that reads this page against the truth registry fails the build.
See the difference in one run
Rather than argue the point, the same simulation is delivered two ways on one page: as a glass box, with every input, seed and step visible, and as a black box, with only the number. One run, two panels, live in your browser: glass box versus black box.
Where LEXUN is deliberately narrow, and where it is still thin
A comparison that only flatters its author is marketing, so here is the other side. LEXUN covers three decision domains — business runway, career change, major purchase — built deep rather than a catalogue built wide; anything else is refused. Model-domain outcomes resolved: 0 — so no claim of predictive accuracy is made for any model, and none will be before the ten-per-domain minimum is reached; the editorial register is scored separately. The organisation product is one person’s workspace today: shared workspaces, roles and approvals are planned and labelled planned. Those are choices and gaps stated in the open; the defensibility page sets out why the record can only be filled in forwards, and what that does and does not prove.
Which approach fits which buyer
Large-scale simulation platforms are built for organisations buying scale: broad panels, bespoke modelling engagements and a supplier relationship. LEXUN is built for the person or small team facing one consequential financial decision this week whose reasoning has to survive scrutiny afterwards — from a board, a client, a lender, or their own future self reading it back in a year. The two are rarely chasing the same budget. They are chasing the same claim: that a number on a screen has earned the right to be believed.