Matt Corley

Senior Software Engineer - Research Platform @Freenome. Opinions my own.

[Matt Corley @corleymb]

@jxnlco It’s difficult to trust auto-selection because it presupposes that how a session started will be reflective of where it goes. sometimes that is true but then I know that already and can make those tradeoffs myself. anything less than high can’t be trusted for most (of my) work

[Matt Corley @corleymb]

@AllenPGreenMD @1cavemank @parmita “Patients might make choices I don’t like with the information” is not an ethical argument for withholding the information. It’s paternalism. Your job is informed consent, not pre-filtering reality for adults because you think they’re too anxious or stupid to weigh biopsy vs risk

[Matt Corley @corleymb]

@SonOfHesiod @AmandaAskell There is a pervasive paternalistic world view among clinicians that access to proactive screening (especially imaging) will turn up “incidentalomas” that patients are too stupid to make informed followup decisions about, leading to “harm” from unnecessary additional procedures.

[Matt Corley @corleymb]

@jxnlco The closest thing I have found is Outline. It’s self-hostable and ticked those boxes for us, has evolved into a gdocs replacement for my immediate team.

[Matt Corley @corleymb]

@thsottiaux Yes, and with a skill so that Codex has a sense for when to spin up batch-mode sub agents for work. i have witnessed many researchers doing corpus processing using overly expensive tokens because they did not know about batch api or api project setup friction in enterprise.

[Matt Corley @corleymb]

@thsottiaux @jxnlco Native integration of notebook-like UX will be a killer feature for scientists. This is a complex thing to get right (local vs remote, dependency tracking, sharing, etc).

Marimo notebook with codex app-server based integration would slap.

[Matt Corley @corleymb]

@madhavajay @dagarfield @LocasaleLab Academics are weird. Utter disdain for an imperfect yet also really hard to collect and fairly unique/remarkabale dataset. These questions aren’t really tractable for almost anyone else, and everyone clutches their pearls because (gasp) media coverage of science lacks nuance.

[Matt Corley @corleymb]

@mike_lustgarten @EricTopol Pretty cowardly from a science perspective too, leaving no room for open review/rebuttal.

[Matt Corley @corleymb]

@mattyglesias They are university professors in mathematics. They publish their works in a form that any nation can benefit from and unlike natural scientists don’t generate IP or research infrastructure or other capital that tends be limited by borders. Not the best example of the problem.

[Matt Corley @corleymb]

@yacineMTB @martin_casado My take is that people are looking at it all wrong. As much as agentic AI enhances non-technical people, the multiplying effect is even greater for those who approach the world with deep technical understanding, systems thinking, and an engineering mindset.

[Matt Corley @corleymb]

@charliermarsh @headinthebox GIL removal will be one of the best things to happen to Python if it reaches a place of widespread adoption. Guido’s point is valid… but so? CPython contribs should be the people to shoulder the burden of complexity if it means simple, true multicore programming for end users.

[Matt Corley @corleymb]

@redaction @westfallx For officers isn’t it mostly overtime due to the understaffing?

[Matt Corley @corleymb]

@Michael_Druggan @atlanticesque If the housing market operated efficiently, sure. It’s massively distorted by zoning and other local policies that prevent true equilibrium from being met.

[Matt Corley @corleymb]

@Lethophilic @mattyglesias It's hard to translate filibuster reform into tyranny of the 51%. We don't direct vote on legislation, and giving every state 2 senators is already amplifying the voices of smaller/regional interest groups. An additional 3/5ths quorum on top of that is too restrictive.

[Matt Corley @corleymb]

@skamille So far: I love that this book is also trying to speak to a leadership audience, with a rare degree of clarity re: organizational dynamics. However, I am unclear about the contrast the book draws between paved path vs "railway" platforms. any resources you might suggest?

[Matt Corley @corleymb]

@lillybilly299 It’s a mindset: something that isn’t working isn’t a personal failure but a challenge to be solved. When you’re younger/more foolish, it’s actually easier to maintain the self-belief that you can solve it. As we get older we tend to overindex on understanding vs experimenting

[Matt Corley @corleymb]

@mitchellh @RubinHeru Was this WSL2 days? My understanding is there was a big change going to WSL2.

I don’t see major performance differences there. Virtualized Linux on MacOS with orbstack is pretty delightful though, the experience sucked before orbstack imho.

[Matt Corley @corleymb]

@mitchellh I have a windows rig purely for gaming, but WSL truly is legit, and for non-terminal/dev usecases there’s plenty to dislike for either MacOS or Windows. not sure there’s a clear winner there.

Apple hardware wins me over though, so most dev work for me is my apple silicon mbp.

[Matt Corley @corleymb]

@sdamico @jjainschigg @JLopas @ImpulseLabs_ ok this is probably a dumb question but— those burners look a bit small? perhaps it’s just the puctures, but i would be curious to see what it’s like to cook without tiny pots.

[Matt Corley @corleymb]

@BDGesq @George_A_Callas @mattyglesias Peer review is great , but I think some subjects call for investigative journalism: https://t.co/YZ5aY4b2Jk https://t.co/r94enYzwd0 and https://t.co/oZGzF2Tw74