Artificial Intelligence, Privacy and Google: An explosive mix

A few days ago, I knew about a big company using AI in the recruiting processes. Not a big deal since it has become a common practice and if something can be criticized, is the fact that AI commits the same error that human recruiters: The false-negative danger; a danger that goes far beyond recruiting.

The system can have more and more filters, learning from them but…what about the candidates that were eliminated despite they could have been successful? Usually, we will never know about them and, hence, we cannot learn from our errors rejecting good candidates. We know of cases like Clark Gable or Harrison Ford that were rejected by Hollywood but, usually, following the rejected candidates is not feasible and, then, the system -and the human recruiter- learns to improve the filters but it’s not possible learning from the wrong rejections.

That’s a luxury that companies can afford while it affects to jobs where the supply of candidates and the demand for jobs is clearly favorable. However, this error is much more serious if the same practice applies to highly demanded profiles. Eliminating a good candidate is, in this case, much more expensive but both, AI system and human recruiters, don’t have the opportunity to learn from this error type.

Only companies like Google, Amazon or Facebook with an impressive amount of information about many of us could afford to learn from the failures of the system. It’s true that, as a common practice, many recruiters “google” the names of the candidates to get more information, but these companies keep much more information about us than the offered when someone searches for us in Google.

Then, since these companies have a feature that cannot be reached by other companies, including big corporations, we could expect them to be hired in the next future to perform recruiting, evaluate insurance or credit proposals, evaluation about potential business partners and many others.

These companies having so much information about us can share Artificial Intelligence practices with their potential clients but, at the same time, they keep a treasure of information that their potential clients cannot have.

Of course, they have this information because we voluntarily give it in exchange for some advantages in our daily life. At the end of the day, if we are not involved in illegitimate or dangerous activities and we don’t have the intention of doing it, is it so important having a Google, Amazon, Facebook or whoever knowing what we do and even listening to us through their “personal assistants”? Perhaps it is:

The list of companies sharing this feature -almost unlimited access to personal information- is very short. Then, any outcome from them could affect us in many activities, since any company trying to get an advantage from these data is not going to find many potential suppliers.

Now, suppose a dark algorithm that nobody knows exactly how it works deciding that you are not a good option to get a job, insurance, a credit…whatever. Since the list of suppliers of information is very short, you will be rejected once and again, even though nobody will be able to give you a clear explanation of the reasons for the rejection. The algorithm could have chosen an irrelevant action that happens to keep a high correlation with a potential problem and, hence, you would be rejected.

Should this danger be disregarded? Companies like Amazon had their own problems with their recruiting systems when, unintendedly, they introduced racial and genre biases. There is not a valid reason to suppose that this cannot happen again with much more data and affecting many more activities.

Let me share a recent personal experience related to this error type: Recently, I was blacklisted by a supplier of information security. The apparent reason was having a page, opening automatically every time I opened Chrome, without further action. That generated repeated accesses; it was read by the website as an attempt to spy and they blacklisted me. The problem was that the supplier of information security had many clients: Banks, transportation companies…and even my own ISP. All of them denied my access to their websites based on the wrong information.

This is, of course, a minor sample of what can happen if Artificial Intelligence is applied over a vast amount of data by the few companies that have access to them and, because of reasons that nobody could explain, we must confront rejection in many activities; a rejection that would be at the same time universal and unexplained.

Perhaps the future was not defined in “1984” by Orwell but in “The process” by Kafka.

Goodbye, AliExpress (Alibaba)

The first condition for online commerce to work is trust. If you cannot trust a place, that place will die.

A few weeks ago, I bought a bone conduction earphone through Aliexpress. One of the side did not work from the beginning. I simply asked for the replacement with one who worked.

The supplier asked for proof about the problem. However, it’s quite difficult showing in a recording that one side of the headphones is working while the other don’t. I sent two videos, one to supplier and the other one to AliExpress.

In the meanwhile, the supplier asked me to retire the claim, alleging “personal reasons” instead of the real problem, that is, defective product.

Finally, AliExpress considers that I did not prove the problem and decides that I must not receive a reimbursement nor a new product.

After that experience, I only can say “Goodbye, Alibaba”. I won’t buy anything from you anymore and I will invite my relatives to do the same.

A contracorriente: ¿Libertad de información o libertad de difamación?

Creo que en ciertos momentos es necesario dejar clara la posición personal para evitar confusiones, de modo que allá va:

No me gusta la forma en que ha llegado al poder en España el líder del PSOE, Pedro Sánchez, no me gusta su actuación en los principales temas de España, no me gustan sus socios, no me gusta que remolonee en la convocatoria de elecciones y la ocultación de su tesis me parece un asunto sospechoso del que el tiempo dirá, a corto plazo, si hay algo real o es una tormenta en un vaso de agua. Está claro ¿verdad?

Sin embargo, me parece impresentable que el gran asunto actual no sea nada de eso sino que haya osado exigirles la rectificación a algunos medios de comunicación y cómo esa exigencia atenta supuestamente contra la libertad de información. En esta trampa han caído interesadamente periodistas y políticos e intentan que los demás caigamos también. Pues no:

La libertad de información no es un derecho del periodista a decir lo que le parezca sino un derecho del ciudadano a recibir información cierta y desde la perspectiva que a ese ciudadano le apetezca.

Si el periodista difama, está tan sujeto a la ley como cualquier otro y no puede invocar la libertad de información como si fuera una patente de corso.

Si alguien, en este caso un presidente de un Gobierno y con independencia de la opinión que se tenga de él, cree que ha sido difamado por un medio de comunicación está en su perfecto derecho de exigir una rectificación.

¿Quiere llevar el asunto a los tribunales? Adelante; es su derecho aunque sea una jugada arriesgada porque, si pierde, no tendrá manera humana de amarrarse a un sillón al que ha llegado de una forma tan irregular.

Desde luego, quien no puede ni debe tratar de impedírselo son los medios de comunicación, alegando libertad de información que, al parecer, consideran sinónimo de libertad para decir lo que les de la gana.

Insisto: No defiendo al personaje sino su pleno derecho a acudir a instancias judiciales si cree que ha sido difamado. Si lo ha sido o no, ya lo veremos pero no cabe rasgarse las vestiduras por un atentado a la libertad de información ante el ejercicio de tal derecho.

THE HARD LIFE OF AVIATION REGULATORS (especially, regarding Human Factors)

There is a very extended mistake among Aviation professionals: The idea that regulations set a minimum level. Hence, accepting a plane, a procedure or an organization means barely being at the minimum acceptable level.

The facts are very different: A single plane able to pass, beyond any kind of reasonable doubt, all the requirements without further questions would not be an “acceptable” plane. It would be an almost perfect product.

Then, where is the trick and why things are not so easy as they seem to be?

Basically, because rules are not so clear as pretended, giving in some cases wide room for interpretation and because, in their crafting, they are mirroring the systems they speak about and, hence, integration is lost in the way.

The acceptance of a new plane is a very long and complex process till the point that some manufacturers give up. For instance, the Chinese COMAC built a first plane certified by Chinese authorities to fly only in China or the French Dassault decided to stop everything jumping directly to the second generation of a plane. It be judged a failure, but it is always better than dragging design problems until they prove that they should have been managed. We cannot avoid to remind cases like the cargo door of DC10 and the consequences of a problem already known during the design phase.

The process is so long and expensive that some manufacturers keep attached to very old models with incremental improvements. Boeing 737, that started to fly in 1968, is a good example. Its brother B777, flying since 1995, keeps a very old Intel-80486 processor inside but changing it would be a major change, despite Intel stopped its production in 1997.

The process is not linear, and many different tests and negotiations are required in the way. Statements of similarity with other planes are frequent and the use of standards from Engineering or Military fields is common place when something is not fully clear in the main regulation.

Of course, some of the guidelines can contradict others since they are addressed to different uses. For instance, a good military regulation very used in Human Factors (MIL-STD-1472) includes a statement about required training, indicating that it should be as short as possible to keep full operating state. That can be justified if we think in environments where lack of resources -including knowledge- or even physical destruction could happen. It should be harder to justify as a rule in passenger’s transportation.

Another standard can include a statement about the worst possible scenario for a specific parameter, but the parameter can be more elusive than that. The idea itself of worst possible scenario could be nonsense and, if the manufacturer accepts this and the regulator buys it, a plane could by flying legally but with serious design flaws.

Regulations about Human Factors were simply absent a few years ago and HF mentions were added to the technical blocks. That was partially changed when a new rule for planes design appeared addressing precisely Human Factors as a block on its own. However, the first attempts were not much further than collecting all the scattered HF mentions in a single place.

Since then, it has been partially corrected in the Acceptable Means of Compliance, but the technical approach still prevails. Very often, manufacturers assemble HF teams with technical specialists in specific systems instead of trying a global and transversal approach.

The regulators take their own cautions and repeat mantras like avoiding fatigue levels beyond acceptability or planes that could not require special alertness or special skill levels to manage a situation.

These conditions are, of course, good but they should not be enough. Compliance with a general condition like this one, in EASA CS25  “Each pilot compartment and its equipment must allow the minimum flight crew (established under CS 25.1523) to perform their duties without unreasonable concentration or fatigue” is quite difficult to demonstrate. If there is not a visible mistake in the design, trying to meet this condition is more a matter of imagining potential situations than a matter of analysis and, as it should be expected, the whole process is driven by analysis, not by imagination.

Very often, designs and the rules governing them try to prevent the accident that happened yesterday, but a strictly analytic approach makes hard to anticipate the next one. Who could anticipate the importance of controls feedback (present in every single old plane) until a related accident happened? Who could anticipate before AA191 that, perhaps, removing the mechanical blockage of flaps/slats could not be so sound idea? Who could think that different presentations of artificial horizon could drive to an accident? What about different automation policies and pilots disregarding a fact that could be disregarded in other planes but not in that one?…

Now, it is still in the news the fact that a Boeing factory had been attacked by the WannyCry virus and the big question was if it had affected the systems of the B777s that were manufactured there. B787 is said to have 6,5 million of code lines. Even though B777 is far below that number, checking it should not be easy and it should be still harder if computers calculating parameters for the manufacturing must be also checked.

That complexity in the product drives not only to invisible faults but to unexpected interactions between theoretically independent events. In some cases, the dependence is clear. Everyone is conscious that an engine stopped can mean hydraulic, electric, pressure and oxygen problems and manufacturers try to design systems pointing to the root problem instead of pointing to every single failure. That’s fine but…what if the interaction is unexpected? What if a secondary problem -like oxygen scarcity, for instance- is more important than the root problem that drove to this? How are we going to define the right training level for operators where there is not a single person who understands the full design?

In the technical parts, the complexity is already a problem. When we add the human element, its features and what kind of things are demanded from operators, the answer is everything but easy. Claiming “lack of training” every time that something serious happens and adding a patch to the present training is not enough.

A full approach more integrated and less shy to speak about using imagination in the whole process is advisable long ago but now it is a must. Operators do not manage systems. They manage situations and, at doing so, they can use several systems at the same time. Even if there is not an unexpected technical interaction among them, there is a place where this interaction happens: The operator who is working with all of them and the concept of consistency is not enough to deal with it.



Anyone in touch with dynamic fields can find this phenomenon: Things are faster than the rules intending to control them. Hence, if the capacity to be enforced is very strong, old rules can stop the advancement. By the same token, if that capacity is weak, rules are simply ignored, and the world evolves following different paths.

The same fact can be observed in many different fields:

Three months ago, an article was titled “POR QUÉ ALBERT EINSTEIN NO PODRÍA SER PROFESOR EN ESPAÑA” (Why Albert Einstein could not be a professor in Spain) and, basically, the reason was in a bureaucratic model tailored for the “average” teacher. This average teacher, just after becoming a Bachelor, starts with the doctorate entering a career path that will finish with the retirement in the University. External experience is not required and, very often, is not welcome.

The age, the publications and the length of the doctoral dissertation (17 pages) could have made impossible for Einstein to teach in Spain. The war for talent means in some environments fighting it wherever it can be found.

If we go to specific and fast evolving fields, things can be worse:

Cybersecurity can be a good example. There is a clear shortage of professionals in the field and it is worsening. The slowness to accept an official curriculum means that, once the curriculum is accepted, is already out-of-date. Then, a diploma is not worth and, instead, certification agencies are taking its place, enforcing up-to-date knowledge for both, getting and keeping the certification.

Financial regulators? Companies are faster than regulators and a single practice can appear as a savings plan, as an insurance product or many other options. If we go to derivative markets, the speed introduces different parameters or practices like high-frequency trading hard to follow.

What about cryptocurrencies? They are sidestepping control by the Governments and, still worse, they can break one of the easiest ways for the States to get funds. Governments would like to break them and, in a few weeks, EU will have a new rule to “protect privacy” that could affect the blockchain process, key for the security of cryptocurrencies and…many Banks operations.

Aviation? The best-selling airplane in the Aviation history -Boeing 737- was designed in 1964 and it started to fly in 1968. The last versions of this plane don’t have some features that could be judged as basic modifications because the process is so long and expensive (more and more long and expensive) that Boeing prefers to keep attached to some features designed more than 50 years ago.

In any of these fields or many others that could be mentioned, the rules are not meeting its intended function, that is, to keep functionality and, in the fields where it is required, safety as a part of the functionality. Whatever the rule can be ignored or can be a heavy load to be dragged in the development, it does not work.

We can laugh at the old “1865 Locomotive Act” with delicious rules such as this: The most draconic restrictions and speed limits were imposed by the 1865 act (the “Red Flag Act”), which required all road locomotives, which included automobiles, to travel at a maximum of 4 mph (6.4 km/h) in the country and 2 mph (3.2 km/h) in the city, as well as requiring a man carrying a red flag to walk in front of road vehicles hauling multiple wagons (Wikipedia).

However, things were evolving in 1865 far slower than now. Non-functional rules like that could be easily identified and removed before becoming a serious problem. That does not happen anymore. We try to get more efficient organizations and more efficient technology, but the architecture of the rules should be re-engineered too.

Perhaps the next revolution is not technologic despite it can be fueled by technology. It could be in the Law: The governing rules -not the specific rules but the process to create, modify, change or cancel rules- should be modified. Rules valid for a world already gone are so useful as a weather forecast for the past week.

Useless diplomas, lost talent, uncontrolled or under-controlled new activities or product design where the adaptation to the rules are a major part of the development cost and time are pointing to a single fact: The rules governing the world are unable to keep the pace of the world itself.

La conjura de los locos

La poca información que va trascendiendo sobre la masacre de Niza trae más dudas que certezas. Vamos a plantear esas dudas en forma de hipótesis genérica:

Supongamos que cualquier tarado decide matar al mayor número de personas que pueda porque la voz de Dios así se lo ha ordenado. ¿Habría que hablar de terrorismo o de crimen religioso o simplemente de alguien que debería haber recibido tratamiento psiquiátrico en lugar de circular por la calle con riesgo para sí mismo y para los demás?

Supongamos, además, que hay una organización conocida por sus salvajadas y que puede resultarle suficientemente atractiva al tarado para, en lugar de un genérico “voz de Dios”, decir que ha causado la masacre en su calidad de soldado de tal organización.

Supongamos, además, que tal organización no tuviera ni noticia de la existencia de tal individuo pero, una vez cometido el crimen, da por buenas las alegaciones del asesino porque entiende que eso le viene bien como baza propagandística.

Supongamos, además, que desde todos los terminales políticos y mediáticos se da por bueno lo alegado por el asesino ya que su aserción ha sido avalado por alguien tan fiable como la organización de salvajes sin fronteras.

Pregunta: ¿No se le está regalando a esa organización asesina una inmensa baza de propaganda?

Desconozco, aunque es lo que parece, si la masacre de Niza responde o no a este hipotético modelo pero esta forma de actuar contribuye a lo que todo terrorista desearía: Esparcir el miedo por hacer creer que está en todas partes y lo controla todo.

Por cierto, la eventualidad de que los salvajes de DAESH no hubieran tenido nada que ver con la masacre de Niza no les hace mejores. No interpretemos estas dudas -como también se está haciendo cada vez que alguien las manifiesta- como una forma de apoyo o, al menos, de intento de suavizar la imagen de esas alimañas. Simplemente, es una nota de atención sobre el hecho de que se les estén regalando bazas propagandísticas cada vez que un loco diga actuar en su nombre: Le ponen el sello de “lobo solitario” y aceptan la alegación, tenga o no verosimilitud.

Se podrían hacer muchas hipótesis sobre los motivos de actuar de esa forma pero hay un principio básico que casi siempre funciona:

Nunca atribuyamos a la maldad lo que pueda ser explicado por la mera estupidez.

Ciudadanos está perdiendo otro tren; tal vez el último

Desde su creación hasta hace un par de años, Ciudadanos ha tenido una trayectoria impecable en lo que siempre han vendido como su motivo de estar en política: Regeneración.

Cuando, con su valiosa ayuda, UPyD prácticamente desapareció, las cosas empezaron a ser distintas: De repente a Ciudadanos le entraron ínfulas socialdemócratas e hizo un acuerdo absurdo con el PSOE para buscar la investidura de Pedro Sánchez como presidente.

Mucho del voto de Ciudadanos, tanto en las anteriores elecciones como en las últimas, puede ser transversal -de verdad, no de la transversalidad de todo a cien de Podemos- y estar compuesto por gente que pudiera haber votado PP o PSOE pero estaban hartos de la forma de funcionar. Puesto que, para muchos de esos votantes, pasar de casta a castuza no era una solución, Ciudadanos aparecía como la única opción disponible.

Sin embargo, ya en ese momento Ciudadanos había empezado a hacer cosas raras. Por ejemplo ¿Realmente un partido como Ciudadanos tiene que presentar a bombo y platillo un programa económico? ¿Tiene sentido que Ciudadanos entre en asuntos como la desaparición o la permanencia del impuesto de sucesiones? Cuando hace eso, se mete en una guerra que nunca ha sido la suya y para la que, además, no están preparados.

Si quieren ejercer de socialdemócratas, pueden afiliarse al PSOE; si quieren ejercer de derecha rara, pueden afiliarse al PP. ¿Acaso todo esto viene de que Rivera aspira a convertirse en el futuro líder del PSOE?

Qué sencillo sería todo si Ciudadanos se hubiera mantenido en su planteamiento original: Medidas de regeneración, exigibles a cualquier Gobierno sea de izquierda o de derecha ¿Programa de gobierno? El del partido que gane las elecciones, apoyado por Ciudadanos siempre que cumpla con esas medidas de regeneración.

¿Era tan difícil? Parece que sí porque el fiasco del 20D no bastó. En la noche electoral del 26J Rivera continuaba con su historia de plantear una reunión a tres donde se debían plantear temas en los que, con todos mis respetos, Ciudadanos no pintaba absolutamente nada.

Su marca, ésa que hizo que muchos les votasen y cuyo abandono ha propiciado que muchos les dejasen de votar, no es un programa de gobierno sino un programa de regeneración institucional y el apoyo al programa de gobierno de otro, sea quien sea, que resulte más votado y siempre que respete ese programa de regeneración.

No han querido darse cuenta y posiblemente han dejado pasar el último tren. Allá ellos.