Profile cover photo
Profile photo
Titus von der Malsburg
Titus's posts

Die Republikaner sollten bei ihrer Nationalversammlung Trump abgesägen (der eh schon verloren hat) und ihn durch eine weibliche Kandidatin ersetzten. Clintons Frauenbonus wäre auf einen Schlag weg und die Themen Email-Affäre, Benghazi, Wallstreet-Reden, Clinton-Foundation würden ihr das Leben sehr schwer machen. Vielleicht ist das die einzige Möglichkeit, wie die Republikaner noch gewinnen können. Könnte aber natürlich auch zum Untergang der Partei führen.

The catastrophic failure of The DAO shows that there may be a fundamental design problem with #Ethereum that cannot be easily fixed.

The problem is that the Etherum system uses an imperative language to describe smart contracts. This opens the door very wide for the kind of bugs that led to the theft of $50M from The DAO. The correct solution would have been to design a domain-specific declarative language that provides abstract building blocks for smart contracts. This would have allowed it to hide the procedural complexity in the interpreter and contract bugs like the present one would not have been possible. Even if there was a problem, it could likely be fixed in the interpreter leaving the contracts and the blockchain untouched! This means that a lot of unanticipated problems would not be fatal for DAOs. The problem of the current approach is that it conflates the business logic and it's implementation thus increasing the burden for contract authors and increasing the attack surface.

In other domains, a suboptimal choice of language paradigm would just add some extra friction but in the domain of smart contracts it leads to catastrophic failures. The paradigm is therefore crucial for the success of Ethereum. I wonder whether the imperative paradigm was a conscious design choice or something that just happened because the developers didn't even realize that it was a design choice.

Post has attachment
New article published: The Importance of Reading Naturally: Evidence From Combined Recordings of Eye Movements and Electric Brain Potentials

Post has attachment
When calibrating the reward for an assignment on #MTurk, it is better to use the geometric mean or median of the completion times instead of the arithmetic mean (the latter is what MTurk shows). If the arithmetic mean is used, extremely slow subjects can skew the result and the reward is likely going to be higher than necessary. In our case, it was 50% higher than what we were aiming for, so we were wasting quite a bit of money.

Click below for details.

Warum muss Merkel ihr Okay geben damit ein Gesetzt angewendet wird?

Post has attachment
New paper in press at Cognitive Science: The importance of reading naturally: Evidence from combined recordings of eye movements and electric brain potentials Joint work with Paul Metzner, Shravan Vasishth, and Frank Rösler

Abstract: How important is the ability to freely control eye movements for reading comprehension? And how does the parser make use of this freedom? We investigated these questions using coregistration of eye movements and event-related brain potentials (ERPs) while participants read either freely or in a computer-controlled word-by-word format (also known as RSVP). Word-by-word presentation and natural reading both elicited qualitatively similar ERP effects in response to syntactic and semantic violations (N400 and P600 effects). Comprehension was better in free reading but only in trials in which the eyes regressed to previous material upon encountering the anomaly. A more fine-grained ERP analysis revealed that these regressions were strongly associated with the well-known P600 effect. In trials without regressions, we instead found sustained centro-parietal negativities starting at around 320 ms post-onset, however, these negativities were only found when the violation occurred in sentence-final position. Taken together, these results suggest that the sentence processing system engages in strategic choices: In response to words that don’t match built-up expectations, it can either explore alternative interpretations (reflected by regressions, P600 effects, and good comprehension) or pursue a "good-enough" processing strategy that tolerates a deficient interpretation (reflected by progressive saccades, sustained negativities, and relatively poor comprehension).

Post has attachment
Die leider weitverbreitete Vorstellung, dass man mit viel Geld Spitzenforschung herbeikaufen kann ist vollkommen idiotisch. Der Unterschied zwischen den USA und Deutschland ist vor allem kulturell.

Post has attachment

People characterize the function of peer-review as the detection of articles that are worthy of being published but I think this is neither what peer-review is effectively doing, nor what it should be doing. First, if a study is rejected at one journal, the authors usually just submit it to the next journal and iterate until it is finally published. So even though peer-review may act as a gatekeeper for individual journals, it does not act as a gatekeeper for the published record, which is what really matters. Second, the primary purpose of peer-review should be to help colleagues improve their research, not just to make a judgment about it. In my experience, this works reasonably well. When it doesn't, it is just as likely the fault of the reviewers, who might write non-constructive reviews, as it is the fault of the authors, who might be unwilling to address constructive criticism.

Köln, Leipzig, Rigaer Straße. Man könnte so viele zynische Kommentare dazu machen, ich weiß gar nicht, wo ich anfangen soll.
Wait while more posts are being loaded