Planeta ANSOL

In my writings, I virtually always refer to my hypothetical reader as “he”. The reasons for doing so were actually very well explained by none other than Richard Dawkins, in his great book “The Blind Watchmaker”1. If I am ever asked about this writing habit, I wanted to have some place to point my inquirers to, and thus I reproduce the relevant passage, from that book’s preface:

I am distressed to find that some women friends (fortunately not many) treat the use of the impersonal masculine pronoun as if it showed intention to exclude them. If there were any excluding to be done (happily there isn’t) I think I would sooner exclude men, but when I once tentatively tried referring to my abstract reader as ‘she’, a feminist denounced me for patronizing condescension: I ought to say ‘he-or-she’, and ‘his-or-her’. That is easy to do if you don’t care about language, but then if you don’t care about language you don’t deserve readers of either sex. Here, I have returned to the normal conventions of English pronouns. I may refer to the ‘reader’ as ‘he’, but I no more think of my readers as specifically male than a French speaker thinks of a table as female. As a matter of fact I believe I do, more often than not, think of my readers as female, but that is my personal affair and I’d hate to think that such considerations impinged on how I use my native language.

English is not my native language, but a similar convention exists in Portuguese—and, in light of the above, I see no reason not to follow it in English as well.

1. This was before his self-appointment as atheist-in-chief, back when he still wrote interesting things, instead of the fundamentalist anti-religious blabber that seems to be all he cares about these days…↩

\(\newcommand\funcdecl[3]{#1\, \colon #2 \rightarrow #3}\) \(\newcommand\covered[2]{#1 \preccurlyeq #2}\) \(\newcommand\finitefield[1]{\mathbb{#1}}\)

Given any binary boolean function \(\funcdecl{\varphi}{\mathbb{F}_2^n}{\mathbb{F}_2}\), there exist \(a_u \in \mathbb{F}_2\) such that:

This representation is called the algebraic normal form (ANF) of \(\varphi\), and the bulk of this text is about the way we compute these coefficients.1 But first, we need to show that this representation always exists. One way to do so, is by reasoning as follows: define a function \(\funcdecl{\delta}{\mathbb{F}_2^n}{\mathbb{F}_2}\), such that \(\delta(x)=1\) if \(x=0\), and \(\delta(x)=0\) if \(x\neq 0\).2 We can now define \(\varphi\) as:

This works because \(\delta(z+x)\) is \(0\) except when \(z=x\), in which case it is \(1\). Equation \(\ref{eq:phi_kronecker1}\) might seem a circular definition, specially if one is thinking of \(\varphi\) as being defined by a “formula”. However, the most general definition of a function from, say, set \(X\) to set \(Y\) is as a subset — that is, a relation — of \(X \times Y\), subject to the caveat that, for every \(x\in X\), there must exist exactly one pair \((x, y)\) in that relation (in the case of function \(\varphi\), we would say that for all \(x\), there must exist exactly one \(y\) such that \(y=\varphi(x)\)). Hence equation \(\ref{eq:phi_kronecker1}\) is just an analytical description of that subset.

Now, we can replace the \(\delta\) function with \(\prod_{i=1}^n (1+x_i+z_i)\), which is \(1\) if and only if \(x=z\), just like \(\delta\). Equation \(\ref{eq:phi_kronecker1}\) now becomes:

When \(\varphi\) is a concrete function, both \(\varphi(z)\) and the \(z_i\)’s will be concrete \(0\)’s and \(1\)’s — the only variables will be the \(x_i\)’s. Furthermore, the \(x_i\)’s belong to \(\mathbb{F}_2\), and hence we have that \(x_i^2 = x_i\). Thus when developing (\(\ref{eq:phi_kronecker2}\)) for a concrete function, we will end up with an equation of the form of (\(\ref{eq:anf1}\)). This shows that the ANF always exists.

The ANF is also unique, which we can show through a cardinality argument. Consider the set of all the functions \(\funcdecl{\varphi}{\mathbb{F}_2^n}{\mathbb{F}_2}\); this set has \(2^{2^n}\) elements; denote it set \(\mathcal{BBF}\).3 Now consider the set \(\mathcal{ANF}\) consisting of all the expressions of the form of the right hand side of equation \(\ref{eq:anf1}\). Its elements are in one-to-one correspondence with the set of subgroups of the set of all expressions of the form \(\prod_{i=1}^n x_i^{u_i}\). There are \(2^n\) of these expressions (one for each \(u\in \mathbb{F}_2^n\)), and hence the number of elements of \(\mathcal{ANF}\) is \(\sum_{j=0}^{n} \binom{2^n}{j}\), which, by a well-known identity relating the binomial coefficients,4 equals \(2^{2^n}\). As every element of \(\mathcal{ANF}\) corresponds exactly to one binary boolean function — i.e. an element of \(\mathcal{BBF}\) — and as both sets have the same cardinality, we conclude that the ANF is unique.5

We usually abbreviate \(\prod_{i=1}^n x_i^{u_i}\) as \(x^u\), so (\(\ref{eq:anf1}\)) becomes \(\varphi(x) = \sum_{u\in \mathbb{F}_2^n}a_u x^u\). But we can simplify it even further. For any \(x\in \mathbb{F}_2^n\), the set \(\{i\in \{1..n\}\ \vert\ x_i\neq 0\}\) is called the support of \(x\), denoted \(supp(x)\). Note that the cardinality of the support of \(x\) equals its Hamming weight, \(hw(x)\).6 Going back to \(\prod_{i=1}^n x_i^{u_i}\), it equals \(1\) if and only if for all \(i\), \(u_i = 1 \rightarrow x_i = 1\) (otherwise it is \(0\)) — but this is the same as requiring that \(supp(u)\subseteq supp(x)\). Let us denote this latter condition as \(u \preccurlyeq x\) (we thus say that \(u\) is covered by \(x\), or equivalently, that \(x\) covers \(u\)). This allows us to rewrite (\(\ref{eq:anf1}\)) as:

But how to compute the different \(a_u\)? We are aided by the fact that the “converse” of (\(\ref{eq:anf2}\)) also holds:

Theorem 1. For all \(u\in \mathbb{F}_2\), the following holds:

\[\tag*{$\square$}\]

Proof. Replace \(a_u\) in (\(\ref{eq:anf2}\)) with the right hand side of (\(\ref{eq:anf3}\)) to obtain:

These two nested summations run over all the pairs \((u, y)\) such that \(u\preccurlyeq x \land y\preccurlyeq u\). We can join both conditions in one: \(y\preccurlyeq u \preccurlyeq x\); and this means that we can cover all the pairs \((y, u)\), where both \(y\) and \(u\) verify the same condition as above, by iterating over all \(y\) covered by \(x\), and then for each \(y\), go over all \(u\) that covers \(y\) (and is covered by \(x\)). That is to say that we can rewrite the nested summation as follows:

This can be further rearranged as:

The last equality follows because of the set \(\{u \in \finitefield{F}_2^n \mid \covered{y}{u} \preccurlyeq x\}\): if \(x \neq y\), then it is either empty (if \(y \not\preccurlyeq x\)), or (if \(y \preccurlyeq x\)) it is nonempty, and has \(2^{hw(x)-hw(y)}\) elements — but in both cases, the inner summation evaluates to zero (remember we are summing in \(\finitefield{F}_2\)). If \(y=x\), the set has just one element, and the result follows. \(\blacksquare\)

The duality shown in equations \(\ref{eq:anf2}\) and \(\ref{eq:anf3}\) yields a simple algorithm to compute the ANF coefficients (\(a_u\)’s) from \(\varphi\)’s truth table, or vice-versa (i.e. it can also be used to compute the value of the function, given the \(a_u\)’s). I will focus on the former. Let us take as an example the function defined by \((x_1 \rightarrow x_2) \rightarrow x_3\). Here is its truth table:

\[ \begin{array}{c|c|c|c} x_1 & x_2 & x_3 & (x_1 \rightarrow x_2) \rightarrow x_3 \\ \hline 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 1 \\ 0 & 1 & 0 & 0 \\ 0 & 1 & 1 & 1 \\ 1 & 0 & 0 & 1 \\ 1 & 0 & 1 & 1 \\ 1 & 1 & 0 & 0 \\ 1 & 1 & 1 & 1 \\ \end{array} \]

The algorithm goes like this: start in the column that corresponds to the “most significant bit”; \(x_1\) in the table above. Where it has the value \(0\), leave the value of the function unchanged; where it has the value \(1\), add to the value of the function at the current entry the value of the function at the entry obtained by setting \(x_1\) to \(0\). This is illustrated in the table below, where instead of replacing the values of the function, they are shown in an auxiliary column:

\[ \begin{array}{c|c|c|c|c} x_1 & x_2 & x_3 & (x_1 \rightarrow x_2) \rightarrow x_3 & \textrm{aux_1} \\ \hline 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 1 & 1 \\ 0 & 1 & 0 & 0 & 0 \\ 0 & 1 & 1 & 1 & 1 \\ 1 & 0 & 0 & 1 & 1 \\ 1 & 0 & 1 & 1 & 0 \\ 1 & 1 & 0 & 0 & 0 \\ 1 & 1 & 1 & 1 & 0 \\ \end{array} \]

And now you just do the same thing for \(x_2\), and then for \(x_3\) (if there were more columns, you would just continue until reaching the one corresponding to the “least significant bit”). Again to illustrate, here’s the result, with two more auxiliary columns (aux_2 for \(x_2\) and aux_3 for \(x_3\)). Note that, as we are not replacing the function values, aux_2 is computed using the values of aux_1, and aux_3 using the values of aux_2.

\[ \begin{array}{c|c|c|c|c|c|c} x_1 & x_2 & x_3 & (x_1 \rightarrow x_2) \rightarrow x_3 & \textrm{aux_1} & \textrm{aux_2} & \textrm{aux_3} \\ \hline 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 1 & 1 & 1 & 1 \\ 0 & 1 & 0 & 0 & 0 & 0 & 0 \\ 0 & 1 & 1 & 1 & 1 & 0 & 0 \\ 1 & 0 & 0 & 1 & 1 & 1 & 1 \\ 1 & 0 & 1 & 1 & 0 & 0 & 1 \\ 1 & 1 & 0 & 0 & 0 & 1 & 1 \\ 1 & 1 & 1 & 1 & 0 & 0 & 1 \\ \end{array} \]

Now, the values in column aux_3 are the \(a_u\)’s that correspond to \(x^u\); for example the coefficient for \(x_1x_2\) (i.e. \(u=110\)) is \(1\), which means the monomial shows up in the final expression. Doing this for all the values in the table, yields that final expression:

And as mentioned above, if we start with a table listing of the \(a_u\)’s, this algorithm yields the original function:

\[ \begin{array}{c|c|c|c|c|c|c} u_1 & u_2 & u_3 & \textrm{aux_3 } (\text{i.e. } a_u) & \textrm{aux_4} & \textrm{aux_5} & \textrm{aux_6} \text{ i.e. } (x_1 \rightarrow x_2) \rightarrow x_3\\ \hline 0 & 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 1 & 1 & 1 & 1 & 1 \\ 0 & 1 & 0 & 0 & 0 & 0 & 0 \\ 0 & 1 & 1 & 0 & 0 & 1 & 1 \\ 1 & 0 & 0 & 1 & 1 & 1 & 1 \\ 1 & 0 & 1 & 1 & 0 & 0 & 1 \\ 1 & 1 & 0 & 1 & 1 & 0 & 0 \\ 1 & 1 & 1 & 1 & 1 & 1 & 1 \\ \end{array} \]

Doing this for a few different functions, it is very easy to get a “feeling” that this works. Of course the high level reason why this works is obvious: because by the time the algorithm get to the aux_3 column, it has already, for each row (where the variables’ columns are seen as an \(u\) value) summed all the function values for all the \(x\)’s that that \(u\) value covers. But how to prove this?

To see the lower level reason why this works, we borrow a bit of notation from Python: indexing starts from \(0\), and \(x[m:n+1]\) means the substring of \(x\) that starts at index \(m\) and ends at index \(n\), including this last element (but not the one at position \(n+1\)) — this is how it works on Python. The element of \(x\) at index \(m\) is denoted \(x[m]\). Also, \(x[m:]\) is the strings that goes from (and includes) position \(m\), to the end of string \(x\). Lastly, if \(x\) and \(y\) are strings, then \(x == y\) means the strings are equal, and \(x + y\) denotes concatenation. A string that contains just the element \(a\) is denoted \([a]\) (this is also a “Python-ism”).

Also, to explain why the algorithm works (we will only look at the case going from the function to the \(a_u\)’s), it is a lot easier to forget the ANF and just think of it as a transversal problem: for a given \(x\), which elements does the algorithm touch? We will show it touches (once and only once) precisely those elements \(u\) such that \(u\preccurlyeq x\).

Indeed, if we are dealing with strings of length \(1\), it is obvious that the algorithm works. The case of length \(2\) can be easily checked by hand, and furthermore, the truth table for length two essentially “doubles” that of the case of length \(1\) — and this provides the motivation for an inductive reasoning. Suppose that for a given string \(x\) of length \(n\) (i.e. when working in \(\mathbb{F}_2^n\)), the algorithm works as expected: i.e., it passes by every string \(u\) of the same length, such that \(u\preccurlyeq x\) holds, exactly once. Consider now the strings of length \(n+1\), where the “extra bit” is added to the left: then, this set can be constructed taking the set of strings of length \(n\), and prepending to it a column of all \(0\)’s, and then doing the same thing, but now with all \(1\)’s column. To take a concrete example, in the truth tables above, observe that the set of tuples of \(x_2\) and \(x_3\) are repeated, once for \(x_1=0\) and again for \(x_1=1\).

Let \(x\in \mathbb{F}_2^{n+1}\), and suppose that \(x[0] = 0\). From the induction hypothesis (and the above discussion about truth table “duplication”), it is clear the algorithm will touch exactly once in all strings covered by \([0] + x[1:] = x\) — which is what we wanted to show. If \(x[0] = 1\), then a similar argument shows that the algorithm will touch exactly once all strings covered by \([1] + x[1:] = x\). However, because the the first element is a \(1\), it will also touch once on the string \([0] + x[1:]\) — which entails that, as it moves throughout the remaining columns, it will also touch exactly once on all strings covered by this latter string. Thus, again we conclude it touches once and just once on all strings covered by \(x\). As \(x\) is an arbitrary string, this holds for all elements of \(\mathbb{F}_2^{n+1}\).

EDIT November 22, 2018: added the table showing how to go from the \(a_u\)’s to the function (the inverse case to what is explained).

EDIT March 30, 2019: simplified the explanation of the algorithm, and of the double summation swap.

1. \(\mathbb{F}_2^n\) denotes the vector space of dimension \(n\) over the Galois field with two elements. It will be clear from context the field in which the summations take place (i.e. \(\mathbb{F}_2\) or \(\mathbb{Z}\)). Also, this formalism implies the convention — widely accepted within the “discrete domains” of mathematics, but by no means universal outside of it — that \(0^0 = 1\). If the reader finds himself mystified by this, checking out the relevant Wikipedia page is very recommended.↩

2. Such a function is called the Kronecker delta.↩

3. There two choices — \(0\) or \(1\) — for each element of the domain, and each element of the domain must have exactly one image. Hence we have two choices for the first element of the domain, and for each of those, another two choices for the second element, … which yields a total number of functions of \(2^{2^n}\). More generally, for a function \(\funcdecl{\psi}{X}{Y}\), with \(X\) and \(Y\) finite, the total number of functions is \(|Y|^{|X|}\).↩

4. Wikipedia to the rescue.↩

5. Actually, this argument also shows the ANF always exists, because the correspondence between \(\mathcal{BBF}\) and \(\mathcal{ANF}\) is a bijection.↩

6. https://en.wikipedia.org/wiki/Hamming_weight.↩

• Japan
• Tokyo
• Back to the beginning…
• The complete albuns

Japan, Earth’s pervert uncle.
–John Oliver1

Yes!, yes, on my first week there—on my first bloody week!—I would come to learn just how very true that quote above is. Which may say something about me, but I assure you, it says a lot more about that country! But we will get to that.

So it was early March (ahem, of last year…) when I set out to Japan—the proverbial land of manga, sushi, and cute weirdness in general—for a three month internship.2 En route I passed through Switzerland, home of course to mild winters, and cheap watches.

I touched down in Japan in the early hours of Friday, and much to my surprise, cleared customs rather quickly (“Come to Japan in business or tourism?” “Ah, I am actually an academic, I came to visit a professor…” [stamps passport, and before I could finish talking] “Ok just write that here and off you go. Sayonara!”). Of course, before customs there is the fingerprinting part—yes, since I think ’08, all “gaijin” (foreigners) get fingerprinted—and after stamping your passport, they inquire about your checked luggage.

Next was the adventure of carrying all that luggage through trains and the famously (over)crowded Tokyo Metro, up to the (no less crowded) Shinjuku area, to finalise my lodging arrangements. To account for baka gaijin (“idiot foreigners” – a redundancy, in the minds of many a Japanese), who for some reason almost never speak Japanese (go figure!) the housing agency actually hands out a sheet with images of buildings and shops from Shinjuku station up to their offices. If you are thinking “well, couldn’t they have just given you, you know, a map?”, they could have, of course, but it would have been useless: Japanese maps make next to no sense, and the Japanese don’t seem overly comfortable with them either. This even comes mentioned in tourist guides! Case in point: when, during the weekend, I went to discover where exactly my hosting institutions’ offices were located (so that I would not arrive late on Monday), I ended up showing the address to a security guard in a building I knew was nearby, and asking for the exact location. Always very politely—as is always the case with the Japanese—but not knowing a word of English (which is also very usual), he tried to help me as much he could, but to no avail. Until an employee came in who spoke English, and I showed her the address, and both she and the guard peered over a computer screen, arguing about a Google Maps search. And, fifteen minutes later, they were able to tell me that the building I was looking for, was just behind the one we were in!! So as you can see, maps and address, don’t really help you all that much…3

But going back to Shinjuku, I finally arrived at the offices of the housing company, only to be handed the paperwork I was required to read before signing the lease agreement. Try to imagine: you are de facto awake for some 30 hours, tired from carrying yourself and your luggage halfway across the city, and before being able to go home for some much deserved sleep, you are made to read reams of paper, about the gazillion of reasons they have to evict you, the other gazillion things that are forbidden, the very detailed protocol for garbage disposal—face palm, right?!—the importance of keeping the place clean because of the moist climate, etc, etc… You cannot just gloss over it either, you have to read once yourself, and then the person attending to you will go over it with you one more time.

By the time I left, I must have really looked like a zombie. And I was given a map to the house I would stay in, so guess what? — I got lost again! By the time I was finally in my room, it was late afternoon (remember I had landed early in the morning). It was small (less than 10m²), but that is quite normal in Japan. The reason is not hard to see: Japan has a population of ~126 million people, which is about half of that of the United States—but the country’s area is less than that of the state of Montana (~4% of the area of the U.S.). So lots of people (roughly 14 million live in Tokyo alone), but not lots of land! This ensures that real estate prices are always at a permanent premium—which in turn causes houses to be built as small as possible.4

And lastly, before beginning with the tales of Tokyo, two of my fellows tenants suffered one of those screw ups that us Westerners usually assume to be impossible to happen in Japan. The flat we were going to, had just recently been converted into a shared house (before that it had been an English school, go figure). Which meant that the tenants that were already there (and me) were living into a brand new share house — which is supposed to be an awesome thing! And for me, indeed it was, but for the first two to move in, things started really badly. The entrance to the building, you see, is always open — as in, you don’t need a key to get into the building — but only from 04h00 until 23h00. If you arrive between 11PM and 4AM, you need to take a sort of service entrance, in the back of the building, which requires a pin code to unlock. I was given the pin when I arrived, but when the first two tenants moved in, apparently nobody thought of this… until one of them returned after 11PM and found himself locked out!! Not the sort of thing one expects in über-efficient Japan, right? But anyway, in the end everything was sorted out, so let’s move on…

As mentioned above, I arrived on Tokyo on a Friday, which was great, I thought — sightseeing weekend! Or so I thought — what actually happened was I was so jet-lagged that I mostly languished in bed, waking up by tea time… (at that time the time difference with Portugal were a whopping 9 hours, after all…). Japan being also the land of (over)work discipline, by Monday I was thoroughly pissed at myself: I just wasted two full days, and from then on I would have to start spending the days (literally!) at the office. What actually happened on Monday though, was that I was told to go there after lunch hour, and just had to take care of a small amount of bureaucracies, and then — to my utter astonishment — was told “see you tomorrow”. So, with the better part of an afternoon to kill, I ended up in Harajuku.

Under normal circumstances—Harajuku being one of the weirdness hotspots in 東京—that would have meant a gazillion of photographs.5 But unfortunately it was pouring showers (the winter in Japan is not the rainy season, that dubious honour falls upon the summer months; when I got there it was half way to Spring already, when the showers gradually become more and more frequent). So I just strolled around until my thick and hooded—but not exactly water-proof—coat had had enough, and then ended up in Starbucks (for the fifth—or thereabouts—time in my life). Which should hint to you that I am not a particular fan of the franchise—so why did I choose it, being in Japan of all places? Well—other than the aforementioned pouring showers—it was because it was the one place where I was confident that you cannot smoke. In Japan, you see, while people are very conscious about pretty much everything, they are so in the typically sui generis Japanese way—and smoking is no exception. And hence, on the hand, you cannot smoke even on the streets (!), except in designated places (I kid you not)—but on the other hand, restaurants and bars are fair game, even if the place is small. You can probably imagine the resulting stench, specially during the meal hours.

Now at this point I should say something about the Japanese concept of restaurant—but given how, when I wrote about Norway, people complained about beginning with food(ie) stuff (“you go to Norway and first thing you talk about is food? Really?!”)—it is perhaps best to consign this discussion to the latter parts of this multi-post writing.6

Going back to Harajuku, I should mention that, a couple of paragraphs above, I actually lied: while facing the drizzle in Harajuku, I did snap one picture:

I was told about this “feature” of Tokyo: animal-themed cafés: besides hedgehogs, there also also cats and owls, and there are probably even more. Making a note to myself to check out this later on, I went my merry way – on to Takashita street (“dori”). This is where you buy all the hippie “alternative” stuff – and I found it rather telling that it is just one street from a mega Apple-store (Steve Jobs would have approved). I should have done my souvenir shopping there and then, because the shop owners lower the prices in rainy days – the bigger the pelting, the bigger the price drop. Who said the Japanese cannot be pragmatic? But alas, I only discovered this a couple weeks afterwords – baka gaijin! Speaking of pragmatism, this is one of those things that I only noticed after I had been there for a couple of weeks: the Nihonjin (Japanese people) are notorious sticklers for the rules, often to a bloody fault, but there is a downside to this: to put it bluntly, they suck at improvising. What in Portugal we call “desenrascanço”, basically the ability to Macgyver your way out of trouble, they completely lack. I guess yin and yan do balance out in the end… (in Portugal we excel at that “desenrascanço” thing, but as for rule-following… well, if needs must…).

But I am digressing. In fact, if there was ever a proof that I am not fond of large cities, it is that even in Tokyo – an extraordinary and unique city, in many, many ways – I still found myself thinking that in a way, big cities are all the same, and what I really want to see is the rest of country (that is topic of part II). Anyway, all my whimsical musings notwithstanding, Edo, as Tokyo was originally called, is indeed an amazing place. Not least because, for a population of 14 million, it is remarkably… calm. What do I mean by this? That while there definitely is more than enough stress, confusion, and rushing around, it is still less than what I saw in cities like, say, Lisbon – which has little over 10% of Tokyo’s population! If this is not testament to the organisational skills of the Japanese, I don’t know what is.

To give an example I happened to witness, one morning I got lost inside Tokyo’s main train station (creatively named “Tokyo Station”), but eventually found a known place, a set of four parallel escalators that, if memory serve, lead you to the connecting metro station (Otemachi). Now when I had previously been there, it was not rush hour, and so two of those escalators went upwards, and the other two downwards. But during rush hour, all four were going upwards, churning out people at full throttle. As I happened to pause and observe such phenomenon, I realised that the resulting human flow — with people walking as if guided by invisible lanes — was more organised that some roads I had driven in. Another display of their sans par organizing skills, I suppose.

I was lucky enough to work and live near the Imperial Gardens, a huge green islet in the middle Tokyo, that is also where the Emperor’s official residency is located. And top it off, I arrived just before Sakura, the cherry-blossom mania! Although it is supposed to cover streets and parks with lush pink, the reality is that quite a lot was actually white-ish… (and it was not a limitation of the photographer; a lot of it indeed leaned towards white). It’s still very worth seeing!

Despite being a ginormous city — or maybe because of it? — Tokyo has quite a number of parks, huge swaths of green where one can stroll and blissfully forget the surrounding jungle of concrete. And while I indeed went and got lost into a few of them (story for a next post), this was long after the sakura period (which only lasts a few days) had passed. But, following the suggestion of a flatmate, there was another place I went for sakura-seeing: Naka-meguro. Much like an über-scaled Venice, Tokyo has a lot of streams streaking inside it, and here was a place where you could enjoy the cherry trees bending and sloshing over the water.

I got there only a short period before sunset — miraculously, the place was not crowded! — and discovered that the night offers you no less beauty!

There was another place where I went sakura-seeing: Hibiya park. This was actually organised by the housing agency, but honestly, it was a bit of a letdown, because Hibiya does not have that many trees, so there could never be that much sakura to see… Considering the agency’s name is Sakura-house, I expected a bit more…

All the talk so far might convey the impression that sakura is a “park-only” phenomenon — but nothing could be further from the truth:

Moving on, one the biggest peculiarities of Japan, the indescribable mixture of modernity and tradition, is exacerbated in Tokyo: being one the world’s most modern cities, it is nevertheless very common to see the old and modern, both in style and actually construction age, side by side.

But alas, this post is getting too big (and taking too long to write…). So, while there are still some more things I want to say about Tokyo, let’s leave that for another text, and get back to my initial days in Tokyo.

Right, so to recap, arrive on Friday, first full day at the office on next Tuesday, and then… it was Friday again. I thought it would be a good opportunity to, you know, get to know the new colleagues (all also foreigners), go out to dinner somewhere, and maybe a stint at a bar afterwards — which goes on to show that one really must be careful with what he wishes for.

It turned out that there already was a… ahem… night out, planned. An outline of the plan was more or less (emphasis on the “less”) like this: go out to have takoyaki (a sort of deep-fried dumpling made of octopus) for dinner, on a food stall on the street, then go to visit (ipso verbo!) a famous, seven-storey high, sex-shop in the Akihabara district, and go to a maid cafe and have something more to eat over there. Me being the inconvenient person that everyone knows and loves, I pointed out that 1) it was raining, so eating on the street, maybe not the best of ideas (and was then told that that part was optional); and 2) food in a maid cafe will likely be (over)pricey (and was then told that maybe so, but the beers would be cheap). You can probably see where this is going…

Eating takoyaki under the rain was indeed most inconvenient, and like all food portions in Japan, it leaves you nowhere close to satiety (I basically starved for three months). The sex-shop is indeed seven-story high, but each floor is small, and it gets smaller as you go up (and weirder, but that’s another story). As there are floors that are specific to men, and others to women, we agree to meet downstairs in fifteen minutes’ time. We each went our way, and fifteen minutes later I dutifully went down — and found myself on the street, completely by myself: no sign of my colleagues. So I decided to wait, and ten minutes pass, and still no sign of them. Remember I had just been in Japan for a week, and I did not have anyone’s numbers (we used a Korean app, Line, to chat, which does not require a SIM card, but it does require mobile data, which I did not have). Nervously, I decided to take a quick glance around the block, and then come back. Luckily, I then ran into one of the girls, and asked where the hell was everyone! “All the other girls are still in the sex-shop!” Facepalm, if ever there was one.

Anyway, after we all got back together again, we ended up on the maid cafe, and this was next on menu:

I was later told that the particular maid cafe we went to was one of the more mainstream ones, but I was still quite shocked to see grown men, many probably family men, behave like pre-pubescent teenagers, let of the leash. Imagine the scenario: young doll-ish looking girls singing and dancing on-stage (and often off-stage, amidst the crowd), and adult men singing, shouting, kicking, calling them, indeed yelling for them, like teenagers watching their idol for the very first time. I still have a vivid image of a guy in his 50s doing push-ups (!!!) at the feet of one of the girls — if that’s not teenage bravado, I don’t know what is.

Anyway, eventually it was time to leave, and then I made a most important discovery: the subway closes at midnight (it was way past midnight). So we figured that instead of spending money on a cab, we might as well walk to a bar nearby, and have some more pints, while waiting for the 6AM metro to go home. And guess what?! Before I knew it, four Japanese young women had come to talk to me!! And not the usual chat-up chit-chat either; they went to the point: “are you working here?”, “are you living here?”, basically, “are you here to stay?” (though none of them phrase it like that — well, that I remember). The writing on the wall was fairly obvious, but still, gaijin for the win! :D

Alas, the night came and went, and when I finally arrived home, the sky was already pre-sunrise crimson. And so I spent the rest of my second weekend in Nihon-land recovering from lack of sleep—again!—and started my second week thinking that, as I was not going to be there for that long, this would be all of “weird Japan” I would get to see.

Oh, little did I know…

The story continues in part ii, but in the meantime, here are Tokyo’s photo albums:

• Tokyo in general (including airports in the journey).

• Asakusa (including Sensoji temple).

• Sakura.

• Shinjuku Gyoen park

• Yogogi park

## The top image depicts mount Fuji, photographed while on the Shinkansen to Kyoto. The photo as shown is basically the original, after gimp’s “equalise” operation. Not bad, for a photo taken in a bullet train, uh? ##

1. https://www.youtube.com/watch?v=s14qCFC5RwI, 0m:06s. By the way, that Kanji in the title, I (and most people in Tokyo, as far I could tell) read it “Nihon”—literally “origin of the Sun”—i.e. Japan’s name in Japanese. But, as with virtually everything else there, it’s a bit(!) more complicated than that. And if this was just about the country’s freakin’ name, imagine the rest…↩

2. For the uninitiated reader, I am referring to manga the comic; see here.↩

3. After returning to Portugal, an acquaintance who had studied Japanese culture explained to me part of the reason why addresses in Japan are so weird: for a given area, house number 1 is the oldest, number 2 is the second oldest, and so on. Now imagine what happens when the oldest and second oldest houses are not adjacent, and extrapolate that to a few hundred houses, and you should start to get some idea of the inanity involved in map reading…↩

4. Though I wonder if there could more to it than this. In the next post I will talk about the castle of Matsumoto, one of Japan’s oldest castles, which still has its original construction. It dates back to late 16th century, when I suppose the population of Japan was a lot less than what it is today. And yet, at least by any European standard, it is small.↩

5. If you are tempted to surmise that that Kanji 東京 means “Tokyo”, you are of course correct. Incidently, which word do you think you get when you reverse that Kanji? Perhaps Kyoto?! Alas, no.↩

6. See here for my stories in Norway, and here for the food related ones.↩

As entidades que constituem a administração pública têm, muitas vezes, de desenvolver software: um website, uma plataforma, uma aplicação, etc. O desenvolvimento desse software é pago com dinheiro público, dinheiro dos impostos de todos nós.

O problema é que a administração pública permite que esse software tenha uma licença de software proprietário.

O que é que isto significa?

Esse software, que foi pago por todos nós, não pode ser reutilizado, nem verificado.

Mas se uma entidade pública licenciar o código do software que desenvolve com uma licença de software livre ou de código aberto, então qualquer outra entidade pode reutilizar esse código, para prestar serviços similares, poupando dinheiro, trabalho e tempo, uma vez que não é preciso estar a desenvolver novo código: reutiliza-se o que já existe!

Para além disto, é possível a qualquer pessoa ver o código desse software para verificar que o software não está a fazer nada que não deva fazer: que não está a enviar informação sobre o que têm no vosso computador para quem não deve enviar, que não tem backdoors, que possam servir de entrada a um ataque, etc.

Assim, a Free Software Foundation Europe (FSFe) criou uma carta aberta, a ser enviada aos representantes Europeus, para pedir a criação de legislação, que determine que o software financiado com dinheiro público e desenvolvido para o sector público tenha de ser disponibilizado com uma licença de software livre e de código aberto.

A carta já foi assinada por 80 organizações (incluindo as Portuguesas Associação Ensino Livre, a Associação Nacional para o Software Livre e a Associação D3 – Defesa dos Direitos Digitais). Assinem e enviem para os vossos contactos!

A lista de assinaturas que permite comentários tem vários bastante interessantes, que mostram bem como os cidadãos consideram esta medida óbvia e até que já devia estar implementada. Vale a pena percorrer os comentários aqui.

Para explicar melhor o que está em causa, a FSFe criou um vídeo, que deixo aqui:

Em Julho, várias comissões do Parlamento Europeu discutiram e votaram alterações à proposta da Comissão Europeia para alterar a directiva sobre direito de autor: a comissão do Mercado Interno e Protecção do Consumidor (IMCO) – o segundo voto mais importante; da Cultura e Educação (CULT) e a comissão da Indústria, Investigação e Energia (ITRE). Falta ainda o voto da comissão mais importante, a comissão dos Assuntos Legais (JURI), que terá uma discussão em Setembro e votará em Outubro.

Em que ponto estão as alterações ao direito de autor no Parlamento Europeu?

TL;DR O European Party’s People (sociais-democratas) propôs levar a Europa para a idade das trevas, e o Progressive Alliance of Socialists & Democrats (socialistas) respondeu que estava bem.

Agora com detalhes: O que é que está em causa? Que mudanças estão a ser discutidas e aprovadas? Para quando está prevista a decisão final? Como podemos endireitar o direito de autor?

No final do ano passado, a Comissão Europeia propôs criar um novo direito conexo de 20 anos após a publicação das notícias, a ser dado aos editores de publicações de imprensa.

Notem que as notícias já tinham, e continuam a ter, direitos de autor, pertencentes aos autores/jornalistas, ou seja, ninguém pode partilhar uma notícia inteira nas redes sociais. Mas podemos partilhar um link com um excerto ou título da notícia, como por exemplo:

https://twitter.com/MCF977/status/884495131390476288

Mas se este novo direito para os editores for aprovado, passamos a não poder partilhar excertos ou títulos de notícias na Internet. No caso acima, o MCF não poderia publicar aquele tweet. A não ser que o Twitter passasse a pagar a licença exigida pelos representantes dos editores dos órgãos de comunicação social.

Aqueles que dedicam o seu tempo a verificar factos ou a fazer críticas ao jornalismo, como mostra o tweet abaixo, deixarão de o poder fazer se a proposta da Comissão for avante. A menos que as plataformas/sites que usam para fazer essa crítica paguem as tais licenças.

O Negócio omite que a autorização data de 5 de Junho, dando a entender no título e no lead que esta sucede o roubo de material. É falso. pic.twitter.com/NGVnsPCkAz

— Os Truques (@ostruques) July 1, 2017

Aquelas plataformas que não conseguirem pagar as taxas, irão desaparecer e mesmo as grandes empresas podem decidir não pagar: em Espanha, onde foi aprovada uma lei similar, a Google decidiu fechar o Google News. Como se isto não fosse suficientemente mau, sociais-democratas e socialistas ainda decidiram estender tal restrição às notícias analógicas. Assim, tweets como este do presidente da RTP, com excertos de notícias de um jornal em papel, também não poderão ser partilhados.

Que barbaridade de posição de Gentil Martins hoje no Expresso. Que desilusão. pic.twitter.com/OEOnLYiRDx

— Goncalo Reis (@GoncaloReis007) July 15, 2017

Consequências desta lei, se for aprovada:

• É um ataque ao direito fundamental ao acesso à informação. A lei irá diminuir a circulação de notícias, cortando o acesso à informação pelos cidadãos;
• É um ataque à própria democracia. A diminuição da circulação de notícias de órgãos de comunicação social terá como consequência imediata o aumento da circulação das “fake news”. Uma vez que os utilizadores não vão sair da Web, nem vão deixar de partilhar conteúdos, os cidadãos irão partilhar aqueles conteúdos que os deixarem partilhar. Ora, quem produz notícias falsas não quer saber de direitos de autor, nem direitos conexos para nada. Os direitos de autor e conexos servem para restringir o acesso e/ou utilização; quem produz notícias falsas quer é que estas se espalhem rapidamente pelo maior número de pessoas possíveis, pelo que as “fake news” continuarão a circular, mas agora com menos notícias “reais” para contrabalançar;
• Vamos assistir ao desaparecimento dos órgãos de comunicação social (OCS) mais pequenos. Tem sido passada esta ideia de que a crise no jornalismo se deve principalmente ao facto das pessoas deixarem de comprar o jornal e lerem as notícias na Internet de forma gratuita, mas a verdade é que nunca a venda do jornal, nem a venda de assinaturas, pagou o jornal. Num jornal de papel, a primeira decisão sobre a distribuição dos conteúdos pelas páginas diz respeito à colocação da publicidade que se conseguiu vender para aquela edição. Publicidade na primeira página será mais cara do que nas páginas interiores, publicidade nas páginas ímpares será mais cara do que nas páginas pares, uma página de publicidade será mais cara do que 1/4 de página e assim sucessivamente. Só depois de colocada a publicidade que se conseguiu vender, é que, nas sobras, nos espaços em branco que sobrarem, se colocam as notícias, as opiniões, as reportagens, e outros conteúdos do jornal. A razão pela qual se distribui primeiro pelo jornal a publicidade, que o jornal conseguiu vender, deve-se ao facto de ser a publicidade a maior fatia responsável por pagar o jornal. Se têm ou tinham o hábito de comprar um jornal em papel regularmente e notaram alguns dias em que o jornal trazia mais publicidade do que habitual, é provável que o jornal estivesse a passar por dificuldades económicas nessa altura. Se o número de excertos com links para notícias diminuir, o número de visitantes às páginas dos jornais também irá diminuir. Ora, nenhum anunciante paga para ter publicidade em sites com poucas visitas.
• É um ataque a quem faz crítica às notícias, que obviamente precisa de usar partes dessas notícias para explicar e exemplificar a sua opinião e, neste contexto, é um ataque ao direito fundamental à opinião, liberdade de expressão e crítica.
• É um ataque à inovação no sector, esta lei irá desencorajar startups de criarem serviços que podiam criar novos modelos de negócio, novas formas de chegar às audiências, etc.
• Leis similares na Alemanha e em Espanha não funcionaram: quando a lei entrou em vigor na Alemanha, houve uma quebra tal nas visitas aos sites dos OCS, que os editores renunciaram à taxa; em Espanha, também houve quebras significativas, mas a lei não permite a renúncia à taxa, pelo que desde então há associações de editores a pedirem ao Governo Espanhol para retirar a lei.

Como se as consequências de uma lei deste género não fossem suficientemente negativas, sociais-democratas e socialistas ainda acharam que deviam aplicar esta lei às publicações científicas. A proposta da Comissão Europeia sublinhava que este direito se aplicava apenas às publicações noticiosas, deixando de fora as científicas.

No Twitter, o deputado ao Parlamento Europeu, Carlos Zorrinho, disse que a lei não permite separar publicações periódicas:

A lei não permite separar "academic journals" de outros. Mas o acesso geral aos dados para investigação fica muito facilitado

— Carlos Zorrinho (@czorrinho) July 11, 2017

O que não me deixa muito descansada porque a ser assim significa que a Comissão Europeia propôs uma lei, sem saber como a lei funciona.

Mas mesmo que a lei não permita separar tipos de publicações periódicas, qualquer lei pode ter excepções. Porque não se excepcionaram as publicações científicas e académicas? Mesmo no pior cenário de nada disto ser possível, é incompreensível aceitar uma lei que faz tal ataque à ciência.

• Quebra na disseminação de publicações científicas, a menos que as plataformas onde se partilha o título+link ou excerto+link paguem a taxa;
• Ataque ao Acesso Aberto. Até agora os artigos em acesso aberto dependiam apenas do autor; se esta lei for aprovada, os editores passam a ter também direitos sobre a circulação desses artigos, de forma automática. Assim, se publicarem um artigo em acesso aberto numa editora X e a vossa universidade ou um outro investigador quiser partilhar o título ou excerto com um link para o vosso artigo numa plataforma (site, blog, ou redes sociais) só o poderá fazer se essa plataforma pagar a tal taxa. Ou seja, mesmo que a publicação tenha uma licença livre, que não imponha nenhuma restrição ao conteúdo, linkar com um excerto ou título para essa publicação deixa de ser permitido.

Se a proposta da CE for aprovada, as plataformas cujo modelo se baseia em conteúdos criados pelos utilizadores terão de passar a verificar todos os conteúdos antes destes serem publicados. Ou seja, sempre que quiserem colocar o que quer que seja no Twitter, no vosso blog, no YouTube, a plataforma terá de verificar esse conteúdo antes de ser publicado. Espera-se que esta verificação seja feita através de filtros automáticos, que são cegos às excepções. Ou seja, se quiserem usar partes de uma obra numa apresentação, ou se quiserem usar um excerto de uma obra para darem a vossa opinião, podem fazê-lo porque a lei contém excepções ao direito de autor que vos autoriza a fazê-lo. Mas os filtros automáticos não conseguem distinguir estes casos, pelo que os vossos conteúdos serão barrados.

Medidas deste género já foram condenadas pelo Tribunal de Justiça da União Europeia, que proibiu explicitamente um requisito legal para obrigar a monitorização prévia dos conteúdos. E percebe-se bem porquê. Um sistema deste tipo é a génese de um sistema censório: se um conteúdo for barrado, como sabemos se foi uma violação de um direito de autor ou se alguém não gostou daquilo que se disse?

A comissão ITRE, na sua votação, tentou minimizar o estrago, retirando a referência aos filtros automáticos, mas mantendo a possibilidade de medidas para impedir a disponibilização de conteúdos.

Já a comissão CULT tornou a proposta da CE ainda pior, ao incluir serviços de “cloud”. O que significa que deixa de ser permitido guardarem música ou outras obras que tenham adquirido legalmente em serviços como o Dropbox, Drive, ou outros.

Uma das emendas submetidas por alguns Membros do Parlamento Europeu refere-se à correcção das chamadas medidas tecnológicas, vulgarmente designadas por DRM. Estas medidas são colocadas nas obras para impedir a cópia da obra. O problema é que também impedem as excepções a que os cidadãos têm direito, incluindo aquelas excepções pelas quais pagamos taxas (cópia privada, utilização para fins de ensino, liberdade de expressão, opinião, discussão ou crítica, utilização para fins de investigação científica, etc.). Estas excepções foram criadas para garantir direitos fundamentais e o próprio legislador reconhece na lei que o DRM não pode impedir os cidadãos de beneficiar destas excepções e, por conseguinte, não pode impedir os direitos fundamentais dos cidadãos. O problema é que a forma como isto foi exposto na lei não funciona, nunca funcionou, nem nunca vai funcionar.

Assim, a proposta é alterar a lei de forma a garantir as excepções, criadas para garantir direitos fundamentais, a que os cidadãos têm direito, mesmo nos casos das obras que tenham tais medidas.

As emendas propostas por alguns deputados são particularmente razoáveis porque não alteram nenhum direito dos titulares. Esta alteração não retira direitos aos autores, nem dá mais direitos aos cidadãos. O único impacto que tem é que os cidadãos passam a poder exercer os direitos, que já tinham, mesmo no caso de obras com as tais medidas.

A comissão com o segundo voto mais importante (IMCO) tinha um deputado Português, Carlos Coelho (PSD), e o sr. deputado resolveu votar contra esta alteração. Aliás, todo o grupo parlamentar EPP votou contra esta alteração. De referir que esta medida foi rejeitada apenas por um voto.

Quanto a este voto na comissão ITRE, não é possível saber os votos por artigo. O sr. deputado Carlos Zorrinho (PS), membro desta comissão, não disse se votou ou não a favor, mas aprovou o documento final, que não inclui esta alteração.

Se esta alteração não for aprovada pela JURI, dificilmente chegará ao Plenário para ser votada.

A próxima, última e também mais importante comissão a votar as emendas à proposta da CE é a Comissão JURI. Portugal tem nesta comissão um representante português, o sr. deputado Marinho e Pinto, com o qual devemos entrar em contacto e sensibilizar para estes problemas. A votação está prevista para o mês de Outubro.

António MARINHO E PINTO
Email: antonio.marinhoepinto [at] europarl.europa.eu
Brussels Office Phone: +32(0)2 28 45403

Dos deputados Portugueses, e até agora, a sra. deputada Marisa Matias (Bloco de Esquerda) foi a única a defender os direitos digitais dos cidadãos.

Se conhecerem pessoas de outros países, peçam-lhes para contactarem os seus representantes na Comissão JURI. Neste post, encontram os contactos dos deputados por país.