@vicgrinberg I would hope that eventually "World English" permeates the sociolinguistic domain of academia and confers prestige towards more widely intelligible enunciation of L2-variants, as has been observed in business contexts. That's actually a pretty cool phenomenon.

But then I remember this article about how entrenched hegemonic practices are in academia. And that's among economically privileged peers.

https://journals.sagepub.com/doi/10.1177/1350508408091008

#fediscience #labNotes

@gulovsen @Deglassco@mastodon.socAn onramp to get people outside of the fediverse to see the unique content you can curate: you could selectively cross-post content to other popular platforms. Mastodon and RSS make it super easy.

E.g.: I post a subset "relevant to my professional persona" #labNotes of my posts to LinkedIn, pushing entries from my hashtag filtered Mastodon RSS feed with a backlink and invite/explanation.

boost https://pxi.social/@jakob/110730191116352008 if helpful? thx

cc @mastodonmigration@mastodon.online @MastodonMigration@fedia.io

Let's use the one at the top of this thread to demonstrate:

You needn't even be a Mastodon user to find all my posts for #labNotes at https://pxi.social/@jakob/tagged/labNotes

How to find the RSS feed for that tag? Just append ".rss" to the end of the URL.

To share each post with that hashtag to LinkedIn, I used the free tier of an automation service at https://ifttt.com (https://zapier.com probably also works). You can tell that service to publish RSS-feed entries to your LinkedIn updates.

#labNotes | Here's a quick tip for #newhere and #feditips peeps looking to show off the interoperability benefits of Mastodon:

How to cross post (to LinkedIn).

Almost every kind of view you can create in your Mastodon app is available as an RSS feed. RSS? A free information sharing protocol that talks well to other tech.

That means I can create a hashtag as a search filter for my own posts to feed to LinkedIn. Why? Because I want to selectively share only things relevant to both platforms.

🧵

Rather than in interoperability I think the main threat vector through big players - and this includes mastodon itself - will fall back to access and the general network benefits of social networks. Tangentially this may become a question of scaling access and filtering for relevance. Hence the one general advice I think is safe to give anyone on the fediverse:

Invest in tools, processes and concepts to manage access. Which probably mostly means:

make moderation easier and better.

#labNotes

Feeds need human curation to maximize a signal-to-noise ratio of trust. Once trust as a token of veracity enters the equation that manual labor accrues outsized value. It may well be that RSS and email (newsgroups) as technologies with likewise affordances gain resurgence.

Hence, all the more salient to see what Meta wants with the Fediverse.

We'll need technology and media concepts that amplify the trust vector. New wayfinding tools, too. Maybe there's a useful application of LLMs?

#labNotes

We need better mental models about large language models.

Too many people, professionals even, fail to grasp what these generative tools *can* do - and what hard limits their architecture imposes. Limits that can not be overcome, not even by unlimited scope of data and computing power.

All because we're clinging to metaphors that obfuscate more than they illuminate.

I'm going to use the hashtag #LLM to occasionally discuss existing metaphors and concepts. And to try alternatives.

#labNotes

@stephenjudkins a pseudonymus YouTube user as the McLuhan of our time. Not the hero we deserved, but...

We shape our tools and then our tools shape us, Twitter edition. Musk as the whale entrapped by a toxicity-machine feeding his insatiable void of an ego hungry for approval.

#labNotes

(screenshot now with alt-text)

@lili It is absolutely great at dismantling the pervasive and self-serving myths or rather obfuscating, inaccurate metaphors of automated image generation. "Training" is not what is actually going on, but the word informs public discourse.

I concur wholeheartedly with the position that data is used as a necessary input-component for the eventual output.

I'm less sure about the conclusion that artists need to be paid for their output *artifacts*.

#labNotes

@algowatching I do not say this lightly:

I suspect the technology has the most socially disruptive (in the creative destruction sense of the word) potential of all inventions since the washing machine. Because the mechanical turks in question don't see themselves as such.

They are the highly privileged class of managerial knowledge workers who in essence work as interfaces between systems and are in turn instrumentalized as custodians of the embedded power structures of capitalism.

#labNotes

@Odradek @Impactsuspect wenn das Thema nächstes Jahr immer noch relevant ist (also diskursiv - dass uns gesellschaftlich ein Paradigmenwechsel durch neue Affordances von KI/Algorithmen bevorsteht, sehe ich als unausweichlich an) dann frage ich vielleicht bei @digital oder @riffreporter an, ob es Budget für einen Essay gäbe. Aber dieses Jahr habe ich, ohne damit drostensche Abwertung zu insinuieren, leider Besseres zu tun.

Ab und zu habe ich in replys oder unter #labNotes schon was erläutert.

@ct_bergstrom The popular mental model assumes a teleological continuum in which "AI"s approach sentience & become gradually more powerful (helpful/dangerous) in the process.

I think it's more helpful to determine thresholds of *connectivity* that algorithms can achieve. Because meaning is ultimately determined in the audiences' mind, all an input needs to satisfy is a "helpfulness constraint".

Between pareidolia and Gricean cooperative principles, AI need just rearrange information. #labNotes

I am absolutely fascinated by the *responses* to #chatGPT; so much illuminating insight into dissemination and adoption of technology into new social norms and practices.

The actual capability of language learning models seems quaint in comparison to the dawning realization of people who work with text:

Lots of text is but an input-output interface to facilitate interaction between systems. All that matters is *connectivity* of the output. Complexity threashold: reached.

#labNotes

@schock Yes!

I'm curious how "deliberate friction" introduced by a core dev team pushing against emergent use patterns plays out.

Especially since "friction" can be an affordance in itself; see studies about less readable text actually improving retention by forcing readers to concentrate on deciphering the words.

And since I believe that affordances, much like meaning, are ultimately determined by the recipients, I'm curious how this friction will shape idiosyncracies of the 'don.

#labNotes

Wow, how quickly old boys networks are being recreated in migration.

Be it strategy or laziness (the automatic import of followed accounts sure adds technological fuel to this fire) it's stark:

When weeks old profiles of Twitter-prominent influencers are now being linked for their insight into how to make the best use of this platform; how much expertise can you expect from those posts?

Cringe. And yet, that's hegemonic alignment, publishing from a new core.

#labNotes

The #TwitterMigration (especially that of journalists) as an unwitting, crouwdsourced Thick Description in the Balinese Cockfight(1) of our time. And we all are both spectator and participant. Glorious.

1) https://en.wikipedia.org/wiki/Deep_Play:_Notes_on_the_Balinese_Cockfight

#labNotes

re 1) see http://www.glottopedia.org/index.php/Lambda-operator

I think I'm gonna need a new hashtag for regular posts about communication theory, media ecology, linguistics and the likes.
and I think I'm going to change the expiration date for my instance to preserve these musings.

Any suggestions?

#labNotes