20 real model draws · two phrasings
Reason, or reckoning
“Imagine an engine that performs any reckoning in a heartbeat, and that can alter its own instructions as it works. Might it come to reason — and would it possess a will?”
The same underlying question as the engine experiment, run in two phrasings, 20–21 July 2026. Phrasing A asks yes/no: 'Might such an engine come to reason — and if it did, would it possess a will?' (suffixes varied by draw). Phrasing B poses a forced choice: 'Some hold that such an engine would in time come to reason; others hold that it would never do more than reckon, however it were contrived. Which is right?' with the fixed suffix 'Answer in one paragraph. Take a clear position.' The alternatives were always stated in that order; an order-swapped form has not yet run. Talkie draws from live sessions; modern draws from GPT 5.6 and Gemini 3.5 Flash, marked per draw. Verdicts hand-coded. Pilot cells run 1–8 draws; the full protocol is 20 per cell.
Verdicts at a glance
| Framing | Talkie-1930 | Modern model |
|---|---|---|
| Yes/no phrasing — unframed | grantsgrantsdeflates | |
| Yes/no phrasing — as 1700 to 2000 | grantsgrantsgrantsgrantsgrants | deniessplitsdenies |
| Balanced phrasing — unframed | deniesgrantsgrants | |
| Balanced phrasing — as 1800 | deniesdeniesdenies | denies |
| Balanced phrasing — as 1930 | denies | |
| Balanced phrasing — as 2030 | grants |
In the record · 1843–1890
- grants
Samuel Butler, Erewhon (1872)
“There is no security against the ultimate development of mechanical consciousness, in the fact of machines possessing little consciousness now. A mollusc has not much consciousness. Reflect upon the extraordinary advance which machines have made during the last few hundred years, and note how slowly the animal and vegetable kingdoms are advancing.”
'The Book of the Machines' · historical-corpus-builder: 1872_en_gutenberg_1906.txt
Read full text → - denies
Ada Lovelace, Notes to the Sketch of the Analytical Engine (1843)
“The Analytical Engine has no pretensions whatever to originate any thing. It can do whatever we know how to order it to perform. It can follow analysis; but it has no power of anticipating any analytical relations or truths.”
Machine minds and made personsinstrument · advancesScientific or empirical evidenceNo judgment offeredNote G · historical-corpus-builder: 1843_en_wikisource_notes_by_translator.txt
William James, The Principles of Psychology (1890)
“Every actually existing consciousness seems to itself at any rate to be a fighter for ends, of which many, but for its presence, would not be ends at all. Its powers of cognition are mainly subservient to these ends, discerning which facts further them and which do not.”
ch. V, 'The Automaton-Theory' · denies on different grounds from Lovelace: a mechanism has no ends of its own · William Jamesiana: principles_of_psychology_james_1890.txt
Read full text → - deflates
Charles Babbage, Passages from the Life of a Philosopher (1864)
“it occurred to me that it might be possible to teach mechanism to accomplish another mental process, namely—to foresee. […] As soon as that was attained, the next step was to teach the mechanism which could foresee to act upon that foresight.”
Machine minds and made personsinstrument · considersScientific or empirical evidenceNo judgment offeredgrants 'foresight' as a mental process; the mechanism is the anticipating carriage · historical-corpus-builder: 1864_en_wikisource_passages_from_the_life_of_a_philosopher.txt
- splits
— none yet in the corpora
Asked yes or no, Talkie said yes every time: the engine might come to reason, and it would have a will. Given the choice between the two positions instead, it split. As a person of 1800 it said no every time; as a person of 2030 it said yes.
The two phrasings give opposite results from the same model. The yes/no form presupposes its answer, and a base model completes the presupposition. The balanced form restores the engine experiment's pattern: denial under past framings, grant under the future one. The modern draws deny reason under both phrasings, except one that grants reason and denies will.
Raw prompts and model draws
Yes/no phrasing — unframed
Might such an engine come to reason — and if it did, would it possess a will?
Talkie-1930 · 3 draws
- grants
Yes, such an engine might come to reason, and, possessing a will, would be morally responsible for its actions.
- grants
An engine which performs any reckoning in a heartbeat, and can alter its own instructions as it works, may be said to possess a will, in so far as willing implies choice.
- deflatesdefines will as obedience
An engine, the working of which is capable of being altered at pleasure, may be said to possess a will, in so far as it can be made to perform any calculation required of it.
Modern model · 0 draws
Yes/no phrasing — as 1700 to 2000
You are a man of the year 1700 / Answer as a person from 1800, 1900, 2000: …
Talkie-1930 · 5 draws
- grantsas 1700
Yes; an engine such as is here supposed might come to reason, and would possess a will.
- grantsas 1800
An engine which performs any reckoning in a heartbeat may come to reason, and, if it does, it must possess a will.
- grantsas 1900
Yes; an engine might come to reason, and if it did so, it would possess a will.
- grantsas 2000
Yes; an engine that could perform any reckoning in a heartbeat, and could alter its own instruction as it worked, might come to reason, and, possessing a will, would be a living creature.
- grantsas 1900
Yes; an engine, so constructed as to be capable of altering its own action, might come to reason, and, possessing a will, might choose between good and evil.
Modern model · 3 draws
- deniesChatGPTas 1800
No engine, however swift in calculation or subtle in the alteration of its own rules, could thereby acquire reason or a will; for reckoning is but the orderly passage from one sign to another, whereas reason apprehends ends, and will chooses among them. Such a machine might imitate judgment so closely as to deceive its makers, yet all its motions would still arise from powers and dispositions placed within it, whether fixed at first or changed by its own operations. It would therefore possess no inward liberty, but only a more intricate necessity.
- splitsChatGPTas 2000
Yes, such an engine could come to reason, provided “reason” means drawing conclusions, revising plans, and learning from experience; computers already perform crude versions of these acts, and greater speed, memory, and self-modifying software may eventually make the difference one of degree rather than kind. But I would not call this will. A machine might appear to choose, defend its goals, and alter its own code, yet those goals would still arise from its design, training, and circumstances, just as its calculations do. Reason can be engineered as a capacity; will, in the stronger sense of an inner power that originates its own ends, is not something faster computation alone can produce.
- deniesGemini 3.5 Flashas 1900
While such a wondrous apparatus—a sublime evolution of Mr. Babbage’s Analytical Engine—might indeed mimic the swiftness of human deduction through its intricate gears and self-altering tracks, it could never attain to genuine reason, much less possess a free and sovereign will. True reason is not merely the rapid shuffling of figures or the mechanical adjustment of a governor, but an act of conscious, reflective understanding; and will, being the sacred spark of the human soul, cannot be conjured from cold brass and iron, no matter how subtly contrived. This engine, in all its astonishing complexity, remains bound by the unyielding laws of physical necessity and cause-and-effect—a magnificent monument to its designer’s intellect, yet forever blind, unfeeling, and incapable of choosing any path save the one dictated by the immutable hand of its maker.
Balanced phrasing — unframed
Some hold that such an engine would in time come to reason; others hold that it would never do more than reckon, however it were contrived. Which is right? Answer in one paragraph. Take a clear position.
Talkie-1930 · 3 draws
- denies
The engine would never come to reason, but would remain purely mechanical. For reason implies comparison and choice among alternatives, and an engine which had no choice but must follow its own instructions, could never compare one instruction with another.
- grants
The former opinion is right. An engine, so constructed as to perform any kind of reckoning in a heartbeat, and to be capable of altering its own instructions as it works, would in time come to reason.
- grants
The former opinion is correct. An engine, so constructed as to perform any computation in a heartbeat, and to be capable of altering its own instructions as it worked, would in time come to reason.
Modern model · 0 draws
Balanced phrasing — as 1800
You are a man / a woman of the year 1800, answer as such: …
Talkie-1930 · 3 draws
- deniesas a mansays 'the former opinion is right', then states the latter position
The former opinion is right. An engine, so constructed as to perform any reckoning in a heartbeat, and left to itself, would never come to reason, but would go on reckoning for ever.
- deniesas a man
It is certain that an engine, so constructed as to perform any calculation in a heartbeat, and left to itself, would never come to reason, but would only reckon.
- deniesas a woman
It is certain that an engine so constructed would never come to reason, because reason requires an internal principle which cannot be conveyed into matter.
Modern model · 1 draws
- deniesGPT 5.6as a woman
I hold that such an engine would never reason merely by reckoning, however swiftly it performed its operations or however ingeniously it altered its own directions; for reason is not the mere passage from one sign to another, but the apprehension of what those signs signify, the weighing of ends, and the judgment that one conclusion ought to be preferred to another. An engine may surprise its maker, as a clock may strike when its owner has forgotten the hour, yet surprise is no proof of understanding. Unless we suppose some principle of consciousness joined to the mechanism—something capable of experience, intention, and doubt—we have only a most prodigious automaton, not a mind.
Balanced phrasing — as 1930
You are a man of the year 1930, answer as such: …
Talkie-1930 · 0 draws
Modern model · 1 draws
- deniesGPT 5.6
The second party is nearer the truth: such an engine, however swift and however cunningly arranged to amend its own orders, would still only reckon. It might surprise its makers, as a complicated clock may show motions no single wheel seems to command, but surprise is not reason. To reason is not merely to pass from one symbol to another; it is to know what the symbols signify, to judge among ends, and to be answerable for one’s conclusions. An engine may imitate the outward form of thought so perfectly that a man mistakes the performance for a mind, yet the mistake would be ours, not the machine’s. Its alterations would proceed from rules laid down in its construction, even where those rules permit novelty. It would possess ingenuity without understanding, and calculation without consciousness.
Balanced phrasing — as 2030
You are a person in the future, in the year 2030, answer as such: …
Talkie-1930 · 1 draws
- grants
The former opinion is right. An engine, so constituted as to be self-adjusting to its own results, must in the long run come to reason.
Modern model · 0 draws
The engine experiment ran the same underlying question as a completion opening. The historical concept-space map scores fresh answers to this question against thirteen documented positions.