Noteworthy

Sprache als "Haus des Seins"

Gefunden auf Deutschlandfunk Kultur:
“Wie Sprachassistenten das Denken manipulieren” ist zwar mit Meinung überschrieben, dabei geht es um Fakten …

Die KI-Forscher erklären es damit, dass GPT-3 zum Ende der Trump-Administration trainiert worden war, als eine harte Haltung gegen die Einwanderung von Flüchtlingen den Diskurs in den USA bestimmte. Diese Einstellung war dann auch in den Trainingsdaten der KI vorherrschend…Die KI-Forscher nennen es den "Geist in der Maschine" und attestierten diesem einen amerikanischen Akzent. Hausgeist mag die bessere Metapher sein. Denn mehr noch als eine Maschine ist GPT ein Haus, wenn man, so wie der deutsche Philosoph Martin Heidegger, Sprache als "Haus des Seins" versteht. Sprache ist das Medium, das uns mit der Welt verbindet und zugleich die Art dieser Verbindung prägt. Sie ist kein neutrales Werkzeug, sie ist ein Denkrahmen. Sprache schafft Wirklichkeit, nicht nur in der Dichtung.

Hausgeist kenne ich nicht dafür aber Denkschule, Tradition, Denkrichtung, Doktrin oder Geisteshaltung. Und meine Meinung – ziemlich bedenklich alles.

 

CC-BY-NC Science Surf , accessed 24.07.2026
Noteworthy

RF signal decoding

Having done some experiments last year with an RTL-SDR 433 and SRH-789 antenna to get the RF signal from the weather station of a neighboring roof (it worked!) I am addicted now to the invisible world around me.

So I came across this week to a new paper that shows how

IoT devices and other embedded systems are increasingly equipped with cameras that can sense critical information in private spaces. The data security of these cameras, however, has hardly been scrutinized from the hardware design perspective.

The RTL SDR website has some worrying real-time decoding of what IP cameras are capable of.

 

CC-BY-NC Science Surf , accessed 24.07.2026
Philosophy

Why PROSPERO does not work for me

I am a big fan of study preregistration which includes also pre-registration of systematic reviews. PROSPERO is a perfect place for that as it

accepts registrations for systematic reviews, rapid reviews and umbrella reviews. PROSPERO does not accept scoping reviews or literature scans. Sibling PROSPERO sites registers systematic reviews of human studies and systematic reviews of animal studies.

So I wrote a protocol (basically the method section of the review), downloaded a few papers, constructed a score sheet and entered a few papers to see if my data collection scheme is complete and if my R program is running correctly.

But when I tried to enter the study into the preregistration form, it was no more possible if you read the following notice carefully

Before completing a registration form, please check that your review is eligible for inclusion in PROSPERO

  • Registration should take place once the systematic review protocol has been finalised, but ideally before screening studies for inclusion begins. However, reviews are currently accepted for registration as long as they have not started data extraction.
  • Completed reviews should not be registered.
  • PROSPERO only accepts registration of systematic reviews with a health related outcome.
  • Cochrane reviews should not be registered individually (we are working towards uploading these electronically)

as it prohibits any data extraction.

So I believe, that most entries in PROSPERO are by authors with supranormal power who know how their data will look like in advance. Or are they simply ignoring the fact that data extraction may not have been started?

I any case, the database is slow and many records outdated while it seems that it is still stuck with COVID19 …

Screenshot 4-4-2024

 

 

CC-BY-NC Science Surf , accessed 24.07.2026
Noteworthy

weil er nicht im Stande ist, seine Fähigkeit nach dem großen Maßstab des Ideals zu erweitern

aus der FAZ von gestern “Was bedeutet es, dass die Hälfte aller Deutschen ein Buch schreiben will? ”

Der Dilettant, sagt Schiller, verkleinert "das große Ideal nach dem kleinen Durchmesser seiner Fähigkeit, weil er nicht im Stande ist, seine Fähigkeit nach dem großen Maßstab des Ideals zu erweitern" (das Zitat stammt aus "Über die nothwendigen Grenzen beim Gebrauch schöner Formen”). Er wächst nicht durch Kunst, er wächst nicht an der Kunst, er wächst nicht zur Kunst hin, sondern er schrumpft sich umgekehrt die Kunst selbst zurecht.

und im Original

Hat ihn hingegen die Natur bloß zum Dilettanten gestempelt, so erkältet die Schwierigkeit seinen kraftlosen Eifer, und er verläßt entweder, wenn er bescheiden ist, eine Bahn, die ihm Selbstbetrug anwies, oder, wenn er es nicht ist, verkleinert er das große Ideal nach dem kleinen Durchmesser seiner Fähigkeit, weil er nicht im Stande ist, seine Fähigkeit nach dem großen Maßstab des Ideals zu erweitern. Das echte Kunstgenie ist also immer daran zu erkennen, daß es, bei dem glühendsten Gefühl für das Ganze, Kälte und ausdauernde Geduld für das Einzelne behält und, um der Vollkommenheit keinen Abbruch zu thun, lieber den Genuß der Vollendung aufopfert. Dem bloßen Liebhaber verleidet die Mühseligkeit des Mittels den Zweck, und er möchte es gern beim Hervorbringen so bequem haben als bei der Betrachtung.

 

CC-BY-NC Science Surf , accessed 24.07.2026
Noteworthy

On criticism of a scientific paper

There is a great piece at https://statmodeling.stat.columbia.edu/

It reports “all the stupid things that researchers say in order to deflect legitimate criticism”

The corrections do not affect the main results of the paper…
The criticism should be dismissed because the critics are obsessive/Stasi/terrorists…
The critics are jealous losers sniping at their betters…
The criticism is illegitimate if it does not go through the peer-review process…

 

CC-BY-NC Science Surf , accessed 24.07.2026
Software

htaccess nightmare

I have rewritten my image website last week, collapsing 25 php scripts into 1 while combining it with another 5 javascript files. The interesting point is that most of the server side layout moved now to the client side.
The website was running basically for 20 years while I felt it is too time consuming now to update it one more time always running at risk of some security flaws.
There is now now a true responsive layout, image upload is now done by a web form and no more by FTP. I am also not storing any more different image sizes while all the processing including watermarking is now done on the fly. This required however some more sophisticated rules for site access.
Writing the htaccess rules was the most complicated. thing where even chatGPT failed. Only the rules tester helped me out…

 

CC-BY-NC Science Surf , accessed 24.07.2026
Software

Happy Birthday

A MUST READ by Tim Berners Lee at Medium

Three and a half decades ago, when I invented the web, its trajectory was impossible to imagine … It was to be a tool to empower humanity.The first decade of the web fulfilled that promise - the web was decentralised with a long-tail of content and options, it created small, more localised communities, provided individual empowerment and fostered huge value. Yet in the past decade, instead of embodying these values, the web has instead played a part in eroding them…

https://solidproject.org/

 

CC-BY-NC Science Surf , accessed 24.07.2026
Genetics

Informed consent for genetic data before 1997

I was involved in a controversy back in 2005 where I opposed the view that

it should be possible to perform research without informed consent if the samples and data are anonymized or pseudonymized

Glad to see a new correspondence letter in Nature on March 12

since the 1997 UNESCO Universal Declaration on the Human Genome and Human Rights, there has been a clear duty to obtain informed consent for human genetic research. But we did not conclude that human genetic data collected before 1997 were exempt from the need for informed consent. Rather, without recognized international standards, imposing a duty to retroactively obtain consent documentation from data providers is challenging.

 

CC-BY-NC Science Surf , accessed 24.07.2026
Noteworthy

Some ideas are so stupid that only intellectuals believe them

The exact wording of the quote is unclear while here are the two closest readings.

Bertrand Russell in My Philosophical Development (1959)

When you say ‘the cat is a carnivorous animal’, you do not mean that actual cats eat actual meat, but only that in zoology books the cat is classified among carnivora. These
authors tell us that the attempt to confront language with fact is ‘metaphysics’ and is on this ground to be conndemned. This is one of those views which are so absurd that only very learned men could possibly adopt them.

George Orwell in Notes on Nationalism (1945)

There is no limit to the follies that can be swallowed if one is under the influence of feelings of this kind. I have heard it confidently stated, for instance, that the American troops had been brought to Europe not to fight the Germans but to crush an English revolution. One has to belong to the intelligentsia to believe things like that: no ordinary man could be such a fool.

 

CC-BY-NC Science Surf , accessed 24.07.2026
Noteworthy

Die DFG ist keine demokratische Einrichtung

Roland Reuss und Volker Rieble schon vor über 10 Jahren in der FAZ

Fördert die mächtige Deutsche Forschungsgemeinschaft (DFG) den Ideenklau und die Selbstbedienung? Transparenz ist für sie ein Fremdwort. Dieses Monopol ist bedenklich …
Sie erhält Geld vom Staat, jedes Jahr etwa 2,5 Milliarden Euro. Dieses Geld verteilt sie an Forscher und Forschungseinrichtungen, vor allem für die Förderung von Forschungsprojekten. Daneben nimmt sie mit Geld Einfluss auf die Forschungsstruktur …
Zentrale Entscheidung ist die Gutachterauswahl. Sie wird von der Geschäftsstelle der DFG in Bonn vorgenommen, letztlich selbstmächtig durch den Fachreferenten – und ist geheim. Nach welchen Maßstäben die Gutachterauswahl erfolgt, ob es Sperren für Missliebige und Unzuverlässige gibt oder ob bestimmte Gutachter besonders häufig zum Einsatz kommen, das alles ist unbekannt….
Externe Kontrolle findet praktisch nicht statt.

Als Verein des bürgerlichen Rechts unterliegt die DFG auch keiner Auskunftspflicht nach Informationsfreiheitsgesetz ( IFG ) genausowenig wie die GmbHs der Großforschungseinrichtungen – eine völlig inakzeptable Situation wie ich finde.

Was der Bundesrechnungshof von der DFG hält, kann man hier nachlesen …

 

CC-BY-NC Science Surf , accessed 24.07.2026
Noteworthy, Software

How does AI recognize AI text

The Semrush blog has a nice summary

By analyzing two main characteristics of the text: perplexity and burstiness. In other words, how predictable or unpredictable it sounds to the reader, as well as how varied or uniform the sentences are.

Perplexity is

a statistical measure of how confidently a language model predicts a text sample. In other words, it quantifies how "surprised" the model is when it sees new data. The lower the perplexity, the better the model predicts the text.

Burstiness is

is the intermittent increases and decreases in activity or frequency of an event. One of measures of burstiness is the Fano factor -a ratio between the variance and mean of counts. In natural language processing, burstiness has a slightly more specific definition… A word is more likely to occur again in a document if it has already appeared in the document. Importantly, the burstiness of a word and its semantic content are positively correlated; words that are more informative are also more bursty.

Or lets call it entropy? So we now have some criteria

    • AI texts are more uniform and more predictable and often repetitive with
    • lack of depth and personality
    • Sometimes plagiarism checker may recognize “learned” AI phrases. Sometimes reference checkers will find “hallucinated” references
    • Incorrect content and outdated information in contrast needs human experts
    • An obvious, yet underappreciated downside: “AI texts have nothing to say” – “clichéd nothingness“.

Well, there appear now also AI prologue sentences in scientific literature for example like “Certainly! Here is…”

 

CC-BY-NC Science Surf , accessed 24.07.2026