Desperately Seeking Validation

I've decided to jump on the bandwagon of the week and chat about the sudden surprise that is Deep Seek.  I don't think I'm imparting anything new here, but I see people focusing on various arguments without looking at a more comprehensive view.

It's very difficult to figure out who to root for in this situation, if anybody.  It's like having to choose between Freddy or Jason (and I don't mean choosing vis a vis the reference frame of the actual movie, but choosing between two declining franchises that sold out so hard that they decided a Freddy vs. Jason movie was a good idea to begin with).

• On the one hand, DeepSeek was able to train their model for $6M, far less than the tens of millions of dollars OpenAI and their ilk spent on their own training.  And their training budget had to be FAR more than that, as Anthropic refused to quote the dollar amount when asked - a typical business tactic indicating "it would be massively embarrassing to admit how much we actually spent, since we just got metaphorically pantsed and don't want to compound that embarrassment."
• On the other hand, it was apparently trained on OpenAI's model, which is a little like me writing new introductory text to Hamlet and then claiming the play as my own.
• And, yes, I'm as tickled and appalled as everyone else that OpenAI is looking into possible legal action against DeepSeek for infringement.  If corporations are people in America, OpenAI has a blinding lack of self-awareness.
• On the one hand, DeepSeek is open source (or what's passing for that term in today's vernacular).
• On the other hand, it originates from China.  China is run by an authoritarian regime, and authoritarian regimes lie to save face or because they have no one calling out their lies or outrageous claims.
• I'm not naive, I know all governments do this, but the more authoritarian you are, the more likely your systems are built on a house of cards.
• It's ironic that idiots like Curtis Yarvin and Techno-topians advocate for a more authoritarian hand in public policy and economics.  It's almost like they want to be free to build skyscrapers out of crackers in order to reap greater profits at the expense of public safety.  Oh wait.  It's exactly like that.
• I saw someone on LinkedIn saying "be careful, because it originates from China," and someone else countering with "show proof that we should be careful, not suppositions."  Uh, I think there are several decades of proof that China isn't acting in the populace's best interest (theirs or anyone's) so it's not unusual to expect that, yet again the football will be pulled before one is able to kick it (c'mon Charlie Brown!)
• On the one hand, you can install the model on a typical laptop or small server, which makes it much easier to manipulate for your own purposes.
• On the other, it's still an LLM, so it's subject to the same asinine hallucinations as every other LLM.
• It's less energy-intensive, which is good, but it has the potential to spread misinformation at a much less expensive rate, which is bad.

Finally, all of you libertarians should be celebrating the (apparently temporary) crash of the AI-pegged tech stocks and the venture capitalists behind them.  That's capitalism.  If you believe in efficient markets, this is your time to shine!  Celebrate your ideals while crying over your balance sheets!

Until next time, my human and robot friends.