Comments on Quelques gadgets de science-fiction (pour manipuler le temps et l'Univers)

a3nm (2019-10-04T22:08:01Z)

Ce genre de trucs me semble vaguement lié au présélecteur négatif :

https://en.m.wikipedia.org/wiki/Elitzur%E2%80%93Vaidman_bomb_tester

Ruxor (2019-07-19T16:04:53Z)

@Quark67: J'ai plein d'exemples montrant qu'il est beaucoup plus facile d'acquérir une connaissance passive d'une langue (i.e., la comprendre, à l'écrit et/ou à l'oral) qu'active (i.e., être capable de s'exprimer dedans, à l'écrit et/ou à l'oral, et notamment de trouver les bons mots). On peut très bien être dans une situation où on arrive à lire avec peu d'efforts des textes même assez complexes (quitte à regarder un mot de temps en temps dans un dictionnaire) mais où s'exprimer demande des efforts énormes ; et cette situation sera assez stable au sens où ce n'est pas en continuant à lire plus (ou écouter plus) de cette langue qu'on va progresser en connaissance active.

Ceci étant, je trouve qu'assez a été dit sur le sujet, je voudrais limiter les digressions par rapport au sujet de l'entrée.

(Mais merci pour avoir signalé DeepL, qui est effectivement meilleur que, disons, Google Translate.)

Quark67 (2019-07-19T13:34:15Z)

Vivement l'arrivée de locuteurs alsaciens et bretons, cela mettra de l'ambiance dans les commentaires. Vive la tour de Babel et le langage ummite ! Évidemment les privilégiés parisiens bobos n'ont aucune difficulté à lire et écrire la langue de la domination occidentale.

Depuis le temps que je lis la présence des commentaires anglais de Jonas sur le weblog de David, durée qui doit être peu ou prou équivalente à la présence de Jonas en France, il lui devrait être possible de parler un minimum en français. C'est sûr que s'il s'exprime toujours en anglais, il ne va jamais progresser. Je ne lui tiendrai absolument pas rigueur pour des fautes de grammaire ou de syntaxe (surtout quand on voit le niveau de personnes ayant toujours habité en France, merci la déculturation nationale d'enseigner à faire des pipes plutôt que d'écrire en français). Il pourra toujours s'aider d'un service de traduction basé sur le deep learning (voilà qui pourrait faire l'objet d'un article par Ruxor…) et d'un logiciel (Windows/Linux/Mac) de correction grammaticale et orthographique édité par une compagnie québécoise (langues gérées : le français, et l'anglais en achetant un module optionnel), s'il souhaite perfectionner la maîtrise de notre langue, belle mais complexe.

Bon, c'est mon avis, mais le tenancier du weblog ne semble pas d'accord. Tant pis !

La traduction automatique faite par Deepl (jugez-en le niveau, est-ce correct ?) :

I can't wait for the arrival of Alsatian and Breton speakers, it will bring some atmosphere to the comments. Long live the Tower of Babel and the Ummite language! Obviously, the privileged Parisian bobos have no difficulty in reading and writing the language of Western domination.

Since I have been reading the presence of Jonas' English comments on David's weblog, which must be more or less equivalent to the presence of Jonas' in France, it should be possible for him to speak a minimum in French. Of course, if he always speaks English, he will never progress. I will not hold it against him for grammatical or syntactical errors (especially when you see the level of people who have always lived in France, thank you for the national deculturation of teaching to blowjobs rather than writing in French). He will always be able to use a translation service based on deep learning (this could be the subject of an article by Ruxor…) and a grammatical and spelling correction software (Windows/Linux/Mac) published by a Quebec company (managed languages: French, and English by purchasing an optional module), if he wants to perfect his command of our beautiful but complex language.

Well, that's my opinion, but the weblog owner doesn't seem to agree. Never mind that!

CaptainObvious (2019-07-19T11:01:40Z)

C'est marrant, au-delà de l'aspect temporalité réelle, le rewinder et le fastforwarder jouent énormément avec la mémoire (comme l'amnésie, le rêve…). Ah oui, c'est tout, rien de plus intéressant à signaler ! ^^

Sancho (2019-07-19T01:45:53Z)

Il y a aussi la question de savoir s'il est en effet possible/probable dans l'univers en question que te vienne à l'esprit de prendre la ferme résolution d'appuyer sur le bouton si autre chose qu'un 6 sort (tu es vraisemblablement toi aussi sujet au déterminisme, aux probas conditionnelles…).

D'autre part, tu peux "transformer" un préselecteur positif non-usagé et destructible en un négatif en prenant des résolutions du type "si blabla, alors je détruis le préselecteur sans l'activer" ? Le danger est alors un peu moindre pour toi : si Le Destin veut t'empêcher de le détruire sans l'activer, peut-être que quelqu'un viendra activer le préselecteur juste avant que tu ne le détruises (ou peut-être avait-il déjà été activé auparavant sauf que tu l'ignorais).

zEgg (2019-07-18T23:35:10Z)

@Quark67: "agréable pour l'ensemble des lecteurs".

Euh je suis un lecteur et CERTAINEMENT PAS ! Je trouve au contraire incroyablement malpoli d'imposer l'effort d'un changement de contexte de ce genre. Je ne vois pas en quoi c'est de la bienséance que de forcer une uniformité linguistique complètement inutile. La bienséance c'est plutôt de laisser chacun parler comme il veut. Comprendre est toujours plus facile que parler donc pour optimiser la communication le fardeau du changement de contexte doit être chez celui qui écoute, pas celui qui parle.

Ruxor (2019-07-18T17:19:31Z)

@jonas: I love your story!

jonas (2019-07-18T12:32:15Z)

Quark67 does have a point, in general. I sometimes answer French comments on this blog in English, which isn't ideal. But in the case of this particular comment thread, I'm replying mainly to a3nm and Ruxor, they understand English, and I'm linking to an English article, so I think I'm fine.

jonas (2019-07-18T12:29:07Z)

I also consider a certain variant of the positive preselector gadget with limited power. Here's how it works.

You enter a car dealership. You notice a beautiful new car and fall in love with it. But you know that your love in vain, the car costs way more than you can afford. Reading the information sheet hanging from the window of the car confirms your guesses: the car has all the features that you value in cars, and its price is ten times as much as you will spend on cars total in your life. You realize that this car will haunt you for the next decade: every time you'll sit in the crappy used car that you're planning to buy, you'll be dreaming of this expensive car and longing for it.

As soon as you realize all this, the Devil appears in front of you and tempts you. “There is a way you can get the expensive car, if you are serious about desiring it. I can give you enough luck for you win the jackpot of the lottery next time you buy it. The money will be enough that you can buy the car and maintain it. In compensation for this, I ask your immortal soul, which I will torture for eternity after your death.” He shows you a contract to this effect, and a ceremonial hypodermic needle so you can sign it in blood. The Devil makes it clear that he's not rushing you. You can sign the contract any time before your death. He can leave you alone while you research the lottery to verify that it can really pay for the car, research the car to verify that it's really the kind you'll enjoy, and seriously consider whether your soul is a price worth this dream. The Devil doesn't need to hoodwink you by rushing you into a deal that you don't understand, because he knows your desires better than you yourself do, and knows that he can tempt you with an honest deal.

But you've heard stories about the Devil. They tell you that the Devil is malicious, and if you accept his deal, he might make it backfire on you so you end up really bad even in life. He could, for example, make you lucky enough to win the lottery and buy the car, but also unlucky that a truck hits you on the street, you become paralyzed for your life, and won't ever be able to drive the car that you own even once. You aren't sure whether the stories are true, but they're sufficiently scary that you don't want to take the chance.

The Devil confirms that the stories are true, he is malicious. He could make a truck hit you in a way that you become paralyzed for life. But don't worry, your soul is actually worth more than the expensive car. In exchange for your soul, he can make you win the lottery and also protect you from all traffic accidents. He swiftly produces an amended contract stating this, together with a new ceremonial needle. You'll be able to drive the expensive car backwards on the motorway in a foggy night with the car's lights turned off, he explains, and neither you nor the car will suffer the slightest harm from it.

But you've heard more stories about the Devil. He is creative and good with words. No matter how advantageous the contract seems, even if you hire the best human lawyers to read it, the Devil will find holes in it and find a way to do harm to you in your life already, while keeping the exact words of the contract. You aren't sure if those stories are true, but they make you seriously hesitate before you sign any contract.

The Devil confirms that those stories are true as well. He can twist rules better than you. But he also really, really wants to get your immortal soul. It's worth more to him than just a lottery and protection from traffic accidents. It is, in fact, worth to him enough to make you really attain your desires and not turn against you during your life, even if you can't articulate them well enough for a contract. There is a rule binding him that he can't just take your soul by force, he can only get it if you sell it to him by your own free will. So he can give you a better offer. With that, the Devil produces a third contract, and a third ceremonial hypodermic needle. This contract is really short. It says nothing about the car, accidents, or even your desires, or any other conditions. All the contract says is that you give your immortal soul to the Devil. You are confused.

The Devil explains that he's still patient. You don't have to sign the contract now. You can sign it any time before your death. If you're serious about selling your soul for the car, then the Devil can make you win the lottery, make you not get into any sort of accident, and, in general, make you live a long satisfying life. Near the end of your life, when you are close to death and your body is weak enough that you can no longer drive any car, you will know that the Devil hasn't twisted your words against you in any way and haven't caused you to suffer, and you can sign the contract at that point. But he advices you that this can only work if you truly decide in advance that if you get the car and a satisfying life with it, you will sign the contract. If you aren't ready to promise that to yourself, then the Devil won't make you win the lottery, you will live your ordinary life without the Devil influencing it, and never own the expensive car. You can't bluff the devil about this. He's not a perfect oracle, because that is impossible. But he has much more prescience and telepathy and undesrtanding your true nature and all that stuff than you can ever have. It's very unlikely that you can trick him, because you aren't clever enough. If you lie to yourself that you'll give your soul in exchange for the car, but in the end you'll decide not to give your soul, the Devil can guess this in advance, and won't give you anything. If you have such unrealistically high expectations of a satisfying life that the Devil isn't willing to give them to you, not even for your soul, then again he won't give you anything.

(Instead of the Devil, you can also consider a genie in a lamp. He isn't bound to give any wishes. But he is bound to remain in the lamp unless you free him. He really wants to get out of the lamp. Eventually. He's patient, he can wait for all your life. He's willing to give more than a car in exchange for that.)

Ruxor (2019-07-18T11:27:55Z)

@Quark67: Je ne suis pas d'accord avec ce commentaire. jonas fait l'effort de lire un blog en français alors que ce n'est pas sa langue maternelle, je ne vois pas pourquoi il devrait forcément commenter en français. S'il écrit des commentaires en hongrois, je les approuverai aussi (même si, évidemment, ce serait beaucoup moins commode pour moi de les comprendre, mais c'est son choix et je le respecte). Quant à passer par un traducteur automatique, les lecteurs qui ne comprennent pas l'anglais peuvent aussi faire cet effort (et choisir, du coup, le traducteur qu'ils veulent utiliser). Certes, si quelqu'un veut en utiliser un et poster avec, je ne suis pas contre non plus.

Quark67 (2019-07-18T09:43:45Z)

Cher jonas, il serait certainement agréable pour l'ensemble des lecteurs que vous vous exprimiez en français. Si vous estimez ne pas maîtriser suffisamment notre belle langue, vous pourriez vous reposer sur les services de <URL: https://www.deepl.com/translator > (utilisable gratuitement pour de courts textes). Vous pourrez toujours rajouter dans un second paragraphe votre texte d'origine en anglais.

Il est vrai que les indications en anglais dans les champs consacrés à la publication de commentaires n'aident pas à saisir cette simple règle de bienséance (l'article auquel on répond étant rédigé en français).

Dear jonas, it would certainly be pleasant for all readers if you could express yourself in French. If you feel you don't have enough knowledge of our beautiful language, you could rely on the services of <URL: https://www.deepl.com/translator > (free use for short texts). You can always add your original English text in a second paragraph.

It is true that the indications in English in the fields devoted to the publication of comments do not help to grasp this simple rule of propriety (the article to which we reply being written in French).

jonas (2019-07-17T20:36:49Z)

That article itself doesn't give proofs for what this computational model can do, so you may have to follow the references from it if you want to understand that. Meanwhile, Scott Aaronson confirms <URL: https://www.scottaaronson.com/blog/?p=4253#comment-1814486 > that you can't just simulate P^{NP} with the negative preselector, if I understand correctly that that's equivalent to his computational model.

Ruxor (2019-07-17T19:52:24Z)

@jonas: I'm quite confused by all of this, and I need to take the time to read Aaronson's paper you cited and think about it, but provisionally that would be my conclusion, yes. (If this is correct, it's quite remarkable: I expected my question « what distinguishes the rewinder from the negative preselector » to have a metaphysical answer if it had one at all, I certainly didn't expect one to come from complexity theory.)

jonas (2019-07-17T18:27:38Z)

Ruxor: … oh!

So what you're saying is that you distinguish between the rewinder and the negative preselector, because your argument shows that the rewinder lets you compute PH problems, but the negative preselector presumably doesn't?

jonas (2019-07-17T17:33:04Z)

Ruxor: the article I pointed to says that a rewinder and polynomial time allows you to compute problems in the complexity class called BPP_path <URL: https://complexityzoo.uwaterloo.ca/Complexity_Zoo:B#bpppath >. What the Complexity Zoo says about this class seems to imply that you can't trivially solve a P^NP problem this way, and you definitely can't go much higher than a few steps on the polynomial hierarchy, unless it collapses. That must mean that your argument doesn't work. I am confused, and don't understand how all this selection thing works.

Let me try to explain why your argument fails, although I could be wrong here. Your older article might be relevant. If you want to solve an NP problem with the rewinder, you can't just use the method of pressing the button if a randomly chosen string is not a witness to the NP problem, because then if your input is not a member of your NP problem, you'll always press the button. Instead you'll have to leave an exponentially small probability that you change your mind and don't press the button even if you haven't guessed a witness. On the other hand, if your input is a member, then the NP problem may have a large number of solutions. You can't limit how many solutions there will be. Thus if you calibrate the probability of changing your mind such that if there's only one witness, you are likely to find it after postselection, then if there are an exponentially large number of witnesses, you'll be exponentially unlikely to change your mind. The reason why you can't recursively use the rewinder to solve a P^NP problem is that you are exponentially more likely to not press the button if the answer from the NP oracle is yes. You thus can't make the computation choose an approximately uniform distribution on the witness of the original P^NP problem, because the postselection from the NP problem messes that distribution up.

jonas (2019-07-17T10:00:33Z)

a3nm: See section 9 in Scott Aaronson, “NP-complete Problems and Physical Reality” (2005) SIGACT News Complexity Theory Column <URL: http://www.scottaaronson.com/papers/npcomplete.pdf#17 >

Ruxor (2019-07-17T09:28:59Z)

@a3nm: S'il n'y a pas de limitation sur le nombre d'utilisations du rewinder, c'est au moins NP : tu tires une solution candidate au hasard (selon n'importe quelle loi qui donne à chaque solution candidate une probabilité strictement positive), tu vérifies avec un algorithme polynomial, et si ce n'est pas bon tu rewindes jusqu'au choix. Mais du coup, c'est au moins P^NP = Δ₂P, et en récursant la même astuce, il me semble que ça donne accès à toute la hiérarchie polynomiale PH (en temps observé polynomial, je veux dire). Je suis vaguement tenté de penser que c'est exactement ça, mais bon, je n'y connais rien en complexité, je ne fais que parcourir le Complexity Zoo en me disant à chaque fois « tiens, j'aurais spontanément pensé que c'était la même classe que <autre truc>, je suis surpris ».

a3nm (2019-07-17T06:09:07Z)

On est d'accord que si je branche un rewinder à mon ordi (et sans doute un "vrai" générateur d'aléa) je peux m'en servir pour résoudre des problèmes NP en temps polynomial ? Je me demande quelle est la puissance computationnelle exacte de ce machin-là.

VA (2019-07-16T23:22:04Z)

Au cas improbable où tu ne les aurais pas lus: si ma mémoire est bonne il y a un fast-forwarder dans *Le Voyageur imprudent* de Barjavel (j'apprends d'ailleurs sur Wikipédia qu'apparemment c'est dans ce livre qu'est né le paradoxe du grand-père). Et ton présélecteur négatif m'a fortement rappelé la nouvelle *The Hundred-Light-Year-Diary* in *Axiomatic* de Greg Egan.

jonas (2019-07-16T17:47:44Z)

Let me add a link to your recent sci-fi short story that features a less powerful variant of the rewinder <URL: http://www.madore.org/~david/weblog/d.2016-12-31.2414.html#d.2016-12-31.2414 >.


You can post a comment using the following fields:
Name or nick (mandatory):
Web site URL (optional):
Email address (optional, will not appear):
Identifier phrase (optional, see below):
Attempt to remember the values above?
The comment itself (mandatory):

Optional message for moderator (hidden to others):

Spam protection: please enter below the following signs in reverse order: f868eb


Recent comments