ADSL, bits et bytes [fr]

[en] Don't confuse bits and bytes like me. 1B (Byte) = 8b (bits). A 2000Kb/s DSL connection will only download 250KB/s, and that's normal.

Je viens d’apprendre quelque chose. Je sais que j’ai une ligne ADSL “2000/100”, ce que j’ai toujours traduit dans ma tête comme “Deux Mégas par seconde en download, et 100 K par seconde en upload.”

Que non. Je suis en train de transférer des données de mon serveur en Allemagne sur mon ordinateur. Mon serveur en Allemagne a une grosse bande passante (enfin, relativement: 100Mbits/s). Assez pour saturer une ligne ADSL, en principe. J’étais donc un peu surprise de constater que la vitesse de téléchargement que m’indiquait ma machine tournait autour de 250KB/s. J’ouvre une deuxième connection sur un autre serveur, et la vitesse de la première chute à 130KB/s — la deuxième, je vous le donne en mille, tournant autour des 120KB/s.

Mince, me dis-je. J’ai été trompée sur la marchandise! Eh oui. Je m’attendais en effet à voir un total autour de 2MB/s.

“Encore une victime du marketing,” me dit Patrick, avec qui j’avais partagé mon étonnement.

Une ligne de “2000”, c’est 2000Kbits/s. La taille des fichiers sur mon disque dur, par contre, est mesurée en KBytes. Bits. Bytes. Pas la même unité.

1B (Byte) = 8b (bits) *(pour se souvenir dans quel sens ça va, la grosse unité a droit au B majuscule)

Vous voyez où je veux en venir? On donne les vitesses des connections ADSL en bits, ça fait des nombres plus grands. 🙂 Du coup, cette confusion d’unités peut nous donner l’illusion que nos connections sont beaucoup plus rapides qu’elles ne le sont en réalité.

Donc, 2000Kbits/seconde, cela veut dire en fait environ 250KBytes/seconde (divisé par huit). Il faudra donc 4 secondes pour télécharger un fichier de 1MB (une grande photo par exemple, ou une minute de fichier mp3).

Alors voilà, ma connection ADSL marche très bien, mais elle est huit fois plus lente que ce que j’imaginais!

The Lee Bryant Experiment [en]

An account of the “Lee Bryant Experiment”, where I posted his write-up of his talk into SubEthaEdit bit by bit as he was talking. Some ideas about note-taking, talking, presentations, and write-ups.

[fr] Lorsque Lee Bryant a donné sa conférence à  BlogTalk, j'ai collé la version écrite de ce qu'il disait dans SubEthaEdit, à  mesure qu'il parlait. Cela paraissait une idée intéressante à  expérimenter quand j'ai offert de le faire, mais l'expérience n'était pas concluante. Cela m'a cependant amené à  m'interroger sur les rôles respectifs du discours proprement dit, du support visuel (dias, présentation), de la prise de notes, et de la publication par écrit du contenu d'une conférence.

So, what was this “Lee Bryant Experiment” I was talking about? No, we did not replace Mr. Bryant by a cyborg-lee during the conference so that he could go and have coffee during his own talk. We simply pushed the whole collaborative note-taking experience one step futher.

Lee mentioned during the first afternoon or BlogTalk that his talk was a bit long, and that he was debating whether to rush thr0ugh it or cut stuff out. I of course suggested cutting things out, but then, that meant that some of the things he wanted to say would not reach the audience. Then we had this idea: paste a written, more detailed, version of his talk into SubEthaEdit while he was talking. I offered to do it. We would annotate his notes, and then stick it all up on the wiki. It sounded like a great idea, and a fun thing to do.

I had a few doubts about it in the morning (so had Lee), worried that it would divert the “note-taker’s” attention from what he was actually saying. However, we decided to go ahead and do it, to see what happened.

I didn’t have much trouble keeping up with Lee’s talk and slides and pasting chunks of his text into the common document as he talked. However, I quickly noticed that this completely killed the note-taking. And it got me thinking.

Was that a problem? Is note-taking important, if you get a transcript or detailed paper of the talk afterwards? I think it is. I think that note-taking as a process is important. I know I listen differently whether I am taking notes or not. There is something to be said for reformulating what you’re listening to on the fly. To me, it clearly aids the integration of what is being said. Now, to what extent does collaborative note-taking defeat that? Open question.

Notes are also more succint than the presentation. One interest of note-taking for me is that I summarize in quickly-readable form what I got out of the presentation. Great for refreshing memories.

So yes, I think that was a problem. I don’t think it’s a good idea to give the audience too much text to read during a talk. That goes for slides too. For me, slides should give visual cues to help the audience keep track of where we are in the talk, and what is being said. They shouldn’t contain “stuff to read while you listen” — you can’t read and listen at the same time. If slides are content-heavy, then the talk should be a comment of the slides, and not something done “in parallel with the slides in the background.”

I think a written version of a talk, especially if it is more detailed than the talk itself, should never be made available before or during a talk. I was told that, by the way, in the 3-day project management course I followed while I was at Orange: when presenting something, don’t hand anything out to people unless you want them to stop listening to you.

What would have made more sense, in hindsight, would have been to put up the written version of Lee’s talk on the wiki in parallel with the notes we would have taken, and allow people to comment the paper. Another thing to try, maybe, would be to put only the outline in the SubEthaEdit document — but then, I noticed that when people are writing they rarely scroll down to see what is written below in the document. Note-taking in a text editor does tend to remain a pretty linear operation.

To summarize, I would say that for me, this experiment was a failure. It was not a failure in the sense that we managed to do what had planned to do, and that it worked, but it was a failure in the sense that what we did failed to give any added value to Lee’s talk.

Think otherwise? Open to discussion.