[Phase 2] HIP-42: Allow Spanish phrase

HIP 42

Allow Spanish Phrase/ Permitir Frase en Español

Authors: Lety (Laura Leticia Lopez), nicobilinkis.eth

_status: Phase 2

Created: 2022/04/11
Snapshot signaling vote

Simple Summary / Resumen Simple

Allow the entrants to the PoH registry, or those who are renewing it, to say the required phrase in Spanish

Permitir que los ingresantes al registro de PoH, o quienes lo renueven, puedan decir la frase requerida en español

Abstract / Resumen

PoH entrants, or those who renew their registration, will be allowed to say the phrase that is required to register in Spanish at the time of recording their video. Those who choose to do it in Spanish will not need to record a video in English. All subsequent validation that is required must be in Spanish if it was the selected language.

Se permitirá que quienes ingresan a PoH, o quienes renueven su registro, al momento de grabar su video digan la frase que se requiere para registrarse en español. Quienes elijan hacerlo en español no necesitarán grabar un video en Inglés. Toda validación posterior que se requiera deberá ser en español si fue el idioma seleccionado.

Motivation / Motivación

Right now the Spanish PoH community is not only very active but has a larger number of active humans on their networks than the English community; so we believe that giving it the space it deserves is important for the growth of our community. We also consider it essential that entrants know exactly what they are saying when they are registering. And this way, make the pronunciation errors of native Spanish speakers almost none when they make the video.

En este momento la comunidad de PoH en español es, no solo muy activa, sino que cuenta con una cantidad mayor de personas humanas activas en sus redes que la comunidad en inglés; por lo que creemos que darle el espacio que merece es importante para el crecimiento de nuestra comunidad. Así como también consideramos que es primordial que las personas ingresantes sepan con exactitud lo que están diciendo al momento de registrarse. Y de esta manera hacer que los errores de pronunciación las personas hispanohablantes nativas al realizar el video sean casi nulos.

Specification / Especificación

Native Spanish speakers who choose to make the video in Spanish may do so by saying the phrase: “Certifico que soy una persona humana real y que no estoy actualmente en este registro”.
If other words or extra validation phrases are required or used, they should also be pronounced in Spanish.
Humans who are not native Spanish speakers should choose the English language, not Spanish.

Las personas hispanohablantes nativas que decidan realizar el video en español podrán hacerlo diciendo la frase: “Certifico que soy una persona humana real y que no estoy actualmente en este registro”.
En caso de requerirse o utilizarse otras palabras o frases de validación extra deben también ser pronunciadas en español.
Las personas humanas que no sean hispanohablantes nativas deberían elegir el idioma inglés, no el español.

Rationale / Razones

This HIP was made to meet the constant requests from the Spanish speaking community that wanted the registration to be in their native language.
The reason why this phrase was chosen is because in Spanish words have a gender unlike the phrase in English, which is why it cannot be translated directly since there could be people within the community who do not feel comfortable with a binary expression. We believe that such a phrase would therefore be appropriate.

Esta HIP se pensó en respuesta al pedido constante de la comunidad hispanohablante que solicitó en incontables oportunidades que el registro pudiera ser en su idioma nativo.
El motivo por el cual se eligió esa frase es porque en castellano las palabras tienen género a diferencia de la frase en inglés, razón por la cual no puede ser traducida de manera directa ya que podrían existir personas dentro de la comunidad que no se sintieran cómodas con una expresión binaria. Consideramos que esa frase sería, por lo tanto, adecuada.

Special Thanks / Agradecimientos Especiales

First, we want to thanks all the Proof of Humanity Telegram Community in Spanish.
And a Special Thanks to the admins and collaborators that support us every day in this great community!
And last but not least, to Ludoviko.eth, Mathaius.eth, V4len.eth, @MoniK74 and @Onediss for their invaluable contribution.

En primer lugar queremos agradecer a toda la Comunidad de Telegram de Proof of Humanity en español.
¡Y un agradecimiento especial a las administradoras y administradores, a cada persona voluntaria, que día a día nos apoyan en esta gran comunidad!
Y por último, pero no menos importante, a Ludoviko.eth, Mathaius.eth, V4len.eth, @MoniK74 y a @Onediss por su invaluable aporte.

3 Likes

Allowing multiple languages defeats the purpose of the registration phrase.
The registration phrase is made in order to make it harder for attackers to use videos of people who did not intend to register (for example by paying them to say the sentence).
Requiring the sentence to be in English is not completely foolproof (as some people do not understand English and can be tricked by attackers), but it’s probably the best language for this purpose as English is the defacto international language and victims are the most likely to know at least basics of it (or know someone who knows).

Now if we had another language, the attacker will have a greater choice of language to trick victims (ask Spanish speaking victims to say the sentence in English and English speaking ones to say the sentence in Spanish). Currently victims not speaking English can be tricked.
If this proposal passes, victims who do not speak English or do not speak Spanish can be tricked. There it would mean that only victims who speak both English and Spanish (which is a very small percent of the world population) would be protected.

It would also make the work of challengers and jurors harder (therefore leading to an increase in the deposit size if we want to have constant security).

3 Likes

I want to add that this, in an attempt to address multiculturalism weakens the registry’s curation capability, as it deviates from following a proper standard. Let’s ask ourselves about the robustness and security of the registry in case of adding other languages such as Chinese, Arabic, etc. The ability to curate the registry is reduced as long as the curator does not also have knowledge of these languages.
Especially since the sentence is not strictly regulated by pronunciation or even by the exchange or addition of words. That way my ability to strengthen the register by doing a good curation decreases noticeably and therefore the list is already less secure.
I would advocate making the registry as standard as possible, understanding that in this case, less is more, because it gives us the ability to have a more robust and reliable system, being the main purpose of this registry a sybil-proof list of humans that other protocols can trust.

4 Likes

I propose a revision of the sentence considering three special points:

1 - As long as I understand, the best translation for “human being” is “ser humano”, which has to see with “to be a human”, i.e., with the social existence and recognition of the person rather than with an organic characteristic. Although this might be a longer philosophical debate, using both “person” and “human” seems redundant to me (which person is not human?).

Furthermore, this first part of the sentence it is a bit cacophonous, it does not sound well, although I cannot say exactly why (I was never a gramma master).

2 - I think this part of the sentence does not cover for the following situations: first, a person who is within the registry but whom profile has been removed; second, a person who is within the registry but is renewing (discussed on [Phase-2] HIP-32: Define text to be read for profile renewal - #43 by ludovico)

These two problems also stand for the original sentence, so by copying the English sentence the Spanish one emulates the same problems.

Therefore, I propose to a) anticipate the revision of the original sentence (something that we will need to face with the original sentence the sooner or the later), b) to divide between registering and renewing, and c) to consider a person who is within the register with a challenged (and removed) profile.

3 - Henceforth, if the goal of the Spanish sentence is:

  • to certify that the person is real (i.e., it exists in the physical world, it is not a digital avatar),
  • to certify that such a person is human (i.e., is not an AI or a monkey that learnt to speak),
  • to keep the sense of the original sentence,
  • to avoid binary expressions,
  • and to consider the special cases mentioned above,

then,

there must be a cultural interpretation and contextualisation of the translation rather than a literal translation.

Either we like it or not, Spanish language has the nuances for covering all of this with a right, accurate, melodic, and non binary expression.

Considering all of that, I propose:

First registration: "Certifico ser humano/a/e, real, y no tener ningún perfil activo en este registro"
This translates to English as: I certify to be a human, to be real, and not having any active profile on this registry.
Note that this version covers both who has never been challenged (first submission ever) and who has been challenged and is reapplying.

Renewing application: "Certifico ser humano/a/e, real, y estar renovando el único perfil que tengo en este registro"
This translates to English as: I certify to be a human, to be real, and to be renewing the only profile that I have on this registry.
Note that this version covers both a person who renews before and after the expiration date.

Finally, I propose to the DAO a deep revision both of the original and the Spanish sentences to help avoiding farming cases, yet that will be another discussion on a further stage.

Best regards to all,

2 Likes

Another reason not to allow multiples languages:
It makes using audio for detecting duplicate submissions way harder. An attacker could make a submission in 2 different languages and both humans and AI detection tools would have a way harder time to find out it’s the same person speaking (Text-Dependent speaker recognition is way easier than Text-Independent one).

Some people may say that those are not used yet. But when attacks will come they will need to be used. And if we don’t have the data about people already registered, we would not be able to use those.

Also note that both humans and AI systems do not need to limit themselves on image/video/audio to recognize if a submission is a duplicate, they can use all those information simultaneously and then take a holistic decision (which is generally superior compared to using only one type of cue).

For those saying that adding a second language makes the job harder to curate, I have a counter argument:

NOT adding more languages defeats the purpose of this whole project. If this is supposed to be a play toy for the English speaking internet to create some play money and pretend to save the planet with UBI then I’m out.

This is absurd! Proof of Humanity should attempt to be a global registry of all humans on earth. Wikipedia estimates that around 2 billion people in total speak English. If you don’t know anyone who doesn’t know anyone who doesn’t speak English, then you should consider yourself a very big niche. I cannot work for making PoH bigger in Latin America if we exclude people who don’t speak English. Even from those who do speak, I’ve heard people say they would like to join but are afraid because their accent is very strong.

Yes, we can’t add all languages at the same time: the registry must expand slowly and only add new languages as it expands its curation ability. Yes this means that curators of Spanish speakers must understand Spanish and the project becomes multiple smaller projects, all under the same Humanity umbrella.

This is another reason to add new languages: I would also suggest adding more strict reinforcements of proper language usage: if it sounds like someone doesn’t actually speak the language but is just repeating some random phonemes, then we should consider not accepting them. The phrase is meaningless if the speaker doesn’t understand it.

Maybe instead of complaining that 6 billion people who don’t speak English we should simply ask our curators to make an effort in learning more languages. Or we only slowly add languages when we feel we have enough curators for that language.

Are you saying people should be standardized and if they don’t fit what you believe is the “standard” way of life they don’t deserve the privileges of being a human?

I think this is a very valid point, and one I am certain we could think on ways to alleviate. But IMHO the downsides of not adding another language by far outweighs the risks of a double language speaker. Should we alienate billions of people at the risk of giving up a few more dollars a day for people who speak two languages?

1 Like

I think, the purpose is to create a list of humans as safe as possible and the problem of farming is real. Adding languages is contemplated, internationalization is necessary, the problem appears when you make videos of other people without them knowing what they are saying, so far you can’t control at all that others are not cheated, but you can limit a little the amount of people that can be cheated. One thing is the user interface and another the phrase needed to validate the user, in this sense I think there must be other more fruitful methods that aim to improve this, but adding new languages to the phrase does precisely the opposite. It increases the amount of people who can potentially be fooled by not knowing either language, since you give the farmer options to choose which language the person doesn’t know to execute his maneuver.

So if there are 8 billion people in total, and 2 billion can’t be fooled (a quarter of the world’s population), by adding other languages you are adding those 2 billion to the population that could potentially be fooled. (Only those who speak both languages could escape that group), but you are already expanding the farming surface. And to the extent that you insert optional languages, you further increase the percentage susceptible to being cheated.

I don’t see this as a good option, in case the language to be included was Chinese or Hindi, I think your answer for this case would vary.

This response is quite far from what my intention for making the registry safer suggests. I am a native Spanish speaker and in this case I understand why it is supported to use one phrase for all peoples of the world regardless of distinctions of any kind. I think it makes the register more secure for the reasons already mentioned. The “standard way of life” of individuals is irrelevant.

A HIP should not generate more risks to the security of the project, on the contrary, it should strengthen the registry of humans that other protocols can trust.

One proposal to consider would be to make the registration sentence shorter and more understandable. Enrolling better a user who needs, primarily, to be part of a secure registry that can be trusted.

I believe we can communicate more clearly that mispronunciations and accents are acceptable. Maybe add the phonetics in the UI, and inform users about the meaning of the phrase in their language before they are required to speak.

In the Spanish-speaking groups we have the phonemes that we share with people:

“ai certifai dat aiam a rial giuman and dat aiam not olredi réyisterd in dis réyistri”

We could have one for each language.

If you think adversarially the phrase meaning is actually secondary. In the worst case scenario an attacker/farmer will look for phrases the person doesn’t understand anyway. The phrase could be just gibberish (or esperanto), but why choose gibberish when you have a language that 1/4 of the world understands?

In the best case scenario, the user can be informed of what they are about to say in the UI, in any language they originally speak.

Of course the UI and everything around it should be in their native language - it’s my understanding that Nico, Migue and you are working on this, and I helped translate it to Portuguese.

I had the same impression, and initially that was how I interpreted this HIP. But I believe it might be hard to draw a line between what is just repeating phonemes and what’s actually speaking the language, and we might either end up rejecting submissions that should not be rejected, or letting basically all people in. (Also, it requires a high level of knowledge of the chosen language)


In short, my perspective is that there’s still a lot we can improve in the experience for non-native english speakers:

  • have the UI/policy translated to multiple languages
  • explain the meaning of the phrase in the user’s native language
  • make the phrase less intimidating
  • communicate more clearly accents and mispronunciations are accepted
  • offer phonetics in the UI appropriate for their native language

After we do that, we can evaluate properly how big a barrier it is for the phrase to be in English.

Also, I don’t think this HIP should in any way settle this matter forever, it should be an ongoing conversation, and we should keep looking for ways in how to accept other languages - and for example, accommodate for people that cannot speak.

2 Likes