rif

[rif @derifatives]

@jodiecongirl @random_walker I have a PhD in Operations Research. It is *obviously* strategic.

Google Maps often *overestimates* the time, and seems to be roughly even between overestimating and underestimating. Uber / Lyft *never* arrive sooner than expected.

[rif @derifatives]

@LinkedInHelp OK, I DM'd you, got an automated reply bot, and then DM's you again that I still needed help. Please please help.

[rif @derifatives]

@LinkedInHelp

I can't access my account. I try to sign in, it says 'email already in use', password reset emails don't arrive. Account used my professional name. I have thousands of citations under this name. Can't reach support without login access. My DMs are open.

[rif @derifatives]

@deanwball @krishnanrohit @aidan_mclau The only thing I hate more is how Claude is trained to always end with a follow-up question, and 90% of the time it's super asinine.

[rif @derifatives]

@gusl @peterwildeford Reading more closely, maybe I misspoke, partially based on out-of-band conversation with @gusl. I agree "somebody will hook up GPT-4 to robot talking heads" in two years. I bet against any robot + GPT-4 combo "passing as human" in person in a small number of years.

[rif @derifatives]

@gusl @peterwildeford Oh, I'm here.

[rif @derifatives]

@ostwilkens This is lovely. How is this done?

[rif @derifatives]

@ladymcscope I'm interested in joining.

[rif @derifatives]

Execution by @cheezwizardofoz:

[rif @derifatives]

@zeynep Why? It seems that current policies have driven R well below 1. Later we can look for intermediate policies with R near 1 (roughly: no big crowds, keep bars closed, masks in crowded public spaces). Occasional shutdowns when areas get out of control. No herd immunity needed?

[rif @derifatives]

@adamvsteele @robinhanson Yeah, some of the best models do that: I'm partial to https://t.co/oN7z04AW98; the Flaxman et al. March 30 paper makes a credible attempt at this.

[rif @derifatives]

@robinhanson @tylercowen: The https://t.co/UdvawCfYMR approach is pretty problematic; I outline key issues above.

[rif @derifatives]

@robinhanson 4. The model is not actually "real-time estimating." It is at best "one-incubation-period ago estimating."

[rif @derifatives]

@KentBeck suggests that we "test until fear turns to boredom", but what do we do if fear leads to anger before it turns to boredom?

[rif @derifatives]

@colindcarroll @shoyer @dustinvtran @SingularMattrix @remilouf @iugoaoj @hardmaru @GoogleAI @lukaszkaiser @DynamicWebPaige @jekbradbury @anselmlevskaya Adding @sigfpe, who first developed JointDistributoinCoroutine for TFP.

[rif @derifatives]

Has anyone considered the therapeutic applications of deepfakes? "I always wanted my dad to tell me I was good enough."

[rif @derifatives]

@_julesh_ @BEBischof I'd be very curious to read it!