• 0 Posts
  • 29 Comments
Joined 1 year ago
cake
Cake day: July 21st, 2023

help-circle
  • I have a few honest questions for anyone who supports this kind of legislation.

    First, what problem specifically is this trying to address? Have teen pregnancies gone up since the advent of kids being able to access porn on the internet? Kids with STDs? Sexual assaults on children? What specific metric has changed that makes this kind of legislation a priority right now? Is there a model that shows a correlation between the behaviors this legislation intends to address and the social ills you believe are associated with it?

    Second is the related question of what metrics you think will improve with the introduction of this legislation? How long do you think it will take for that change to come about? If it does not, would you support removing this legislation?

    Third, if a social ill were to be associated as per the above with online content, would you support similar legislation to regulate access (eg, if hate speech or LGBT-phobia posted online were to show a positive correlation with intolerance or violence), would you require online services to monitor access to sites hosting that kind of content, such as requiring a government issued ID to be kept on record and associated with specific user accounts?



  • This the order in which you should try to access papers:

    1. Normal Internet search including quotes to force the title and components like “pdf”
    2. Organizational/lab pages of the authors. Very many people will put either full papers or preprints on their personal professional pages.
    3. Preprint services like arXiv. The ones you look at will be determined by subject area. Preprints will usually only differ from the published work in formatting.
    4. Just email the authors. Most of us are so happy that virtually anyone wants to read the paper we spent months on that we will happily send a copy. Because people are busy you might need to hit them up a couple of times, but most will be more than happy to send you a copy, and most publications specifically carve out to allow authors to do that.







  • It’s just political posturing.

    1. They don’t have a military. The National Guard units would come under the command of the President of the US, and any units in rebellion would know they’re facing courts martial for crimes that would be career limiting in that the penalties could include anything from life in prison to execution. It’s literally treason by the legal definition.
    2. Even if any significant number of troops were to choose to violate the law, modern war isn’t about riflemen. There’s a massive infrastructure required to keep tanks and planes running, not to mention things like carrier battle groups. Northrop and Raytheon aren’t going to be forfeiting USG contracts to sell missile systems to Ohio.
    3. Only the president has the nuclear codes, so nuclear blackmail can’t work either.

    There isn’t going to be another civil war. Too much has changed between then and now in terms of military and economic organization. This is just Texas whacking off yet again, as they did under Obama and Bill Clinton.

    The very real risks we’re facing are the election of Donald Trump - this is the biggest threat - and far right domestic terrorism. The former is an existential threat to the United States and should be treated as such. The latter is a law enforcement issue and should be treated as such. I suspect the Proud Boys are infiltrated all to hell as are the other major organizations, but there’s the potential for a significant amount of harm being done on a larger than 9/11 scale, although it’d be drawn out.


  • The NYT has a market cap of about $8B. MSFT has a market cap of about $3T. MSFT could take a controlling interest in the Times for the change it finds in the couch cushions. I’m betting a good chunk of the c-suites of the interested parties have higher personal net worths than the NYT has in market cap.

    I have mixed feelings about how generative models are built and used. I have mixed feelings about IP laws. I think there needs to be a distinction between academic research and for-profit applications. I don’t know how to bring the laws into alignment on all of those things.

    But I do know that the interested parties who are developing generative models for commercial use, in addition to making their models available for academics and non-commercial applications, could well afford to properly compensate companies for their training data.



  • These sorts of decisions can become more understandable when we incorporate the idea that corporate politics and management structure play a huge role in decision making. If the director of the division charged with building the software is close with the CTO or c-suite, they’re going to give that person the job and just make the numbers fit by adjusting the projections.

    This is what’s referred to as the “agency problem” - that the people designated to act as the agents of others (whether you consider the others to be shareholders or employees) instead make the decisions based on the benefit to themselves, that’s a conflict of interest based on level of selection.

    And in three years when it turns into a disaster, the director will have moved on to a new role or new company.