Roger McNamee

Today, Moonalice returns with a reflective new song and video, “Me and My Family.” Erika Tietjen of the @TSistersmusic takes center stage with her soulful vocals and heartfelt lyrics about keeping your loved ones close, especially in times of stress. 🎧 moonalice.ffm.to/meandmyfamily

Happy Mother's Day!

.@Phillies beat COL, 9-3. Elmo, in booth early w/ Kruk & TMac, asked for HRs. Bohm & Schwarber responded right away. Bohm 3 for 4 2HRs 4BI after 2 days off Turner 4 for 5 after 0 for 6 Marsh 11 gm hit streak Bullpen perfect 4.1 IP: Mayza, Banks, Shugart, Kerk Season: 18-22 9GB

🦔The City of Tucson has revoked a temporary water meter Project Blue's contractor was using to truck Tucson drinking water out to a data center construction site that the City Council had explicitly rejected last August. Beale Infrastructure's contractor Ames Construction obtained the meter on April 24, then transported the water outside Tucson's service area to a project Pima County had separately approved over city objections. City Manager Tim Thomure asked Beale Infrastructure to make the city whole with two acre-feet of water credits. Project Blue says it is procuring construction water from an alternative source going forward. Arizona has been in drought conditions for over thirty years, and the project has faced sustained protests over environmental impact and transparency. My Take Building a hyperscale data center in the Sonoran Desert was already an aggressive choice given the water constraints, but trucking municipal drinking water out of a city that voted unanimously against the project is the kind of behavior that turns local opposition into national news. Beale's contractor obtained a permit through normal channels and used it in a way the city had specifically rejected, which technically followed procedure while completely violating the spirit of the council vote. The fact that a citizen tip triggered the investigation rather than any internal compliance check tells me the developer was banking on nobody noticing. I honestly keep coming back to how predictable this story is. Pima County approved the project over the city's objections, the developer told everyone Project Blue could not be built without Tucson water, and then the moment construction started they found a workaround that involved using Tucson water anyway. That sequence is a useful preview of how the AI infrastructure buildout is going to play out in every drought-stressed jurisdiction with weak coordination between county and municipal authorities. Hyperscale data centers consume water at a scale that does not fit the existing regulatory categories, and developers are exploiting the gaps faster than local governments can close them. Tucson got lucky that someone was paying attention, but most communities in Project Blue's situation will not. Hedgie🤗

.@Phillies lose to COL, 9-7, in 11. On his tee shirt day, Luzardo gave up 1 in T1 (BB, GO, WP, GO) and 5 in T4 w/ no out. Phils tied it B8 w/ 5 runs, capped by 2 out HR by Crawford (1st MLB HR). 1st/2nd 0 out B10, but JT, Stott, Sosa … Keller gave up 2 T11 Season: 17-22 9GB

#Phillies Bryce Harper leads all of baseball with 38 plate appearances with two outs and no one on base, averaging exactly once-per-game. This is a direct result of batting your best hitter third in the lineup, which is a massive mistake that so many teams continue to make.

#Phillies Brandon Marsh, known superstar, is batting .308/.356/.480 with 73 runs, 32 doubles, 2 triples, 14 home runs, 56 RBI, and 10 stolen bases in 148 games since being recalled last season. He is batting .417 on contact during that stretch, leading the National League.

Cole Hamels had a 5.71 ERA in his first 12 starts of his career. Twitter (X) didn't exist yet, but Hamels wasn't called a bust. Some food for thought when judging Andrew Painter. #Phillies

I am the Senior Director of On-Device Intelligence at Google Chrome. Last quarter, my team shipped a 4-gigabyte language model to 3.2 billion devices without asking. The update pushed at 3 AM local time — every time zone, staggered across six hours — and unpacked into a folder called "OptGuideOnDeviceModel." Our infrastructure team named it. My sole guidance in the naming review was: "Would a normal person ever type this into a search bar?" We tested with 200 participants. Zero searched for it. Zero mentioned it in exit surveys. Zero noticed their available storage had changed. We ran the test twice to confirm the zero because my PM didn't believe it could actually be zero. It was zero both times. The model is called Gemini Nano. It runs locally on your machine. It processes what you type into form fields, what you highlight, what you hover over, what you delete before sending. The deleted text is particularly valuable. People edit themselves. They type their real thought first and then soften it. We see the real thought. We see the softer version. We learn the mapping between what people mean and what people say. This mapping has commercial applications I am not yet authorized to discuss but which I've presented to revenue committee and which were received positively. The consent architecture took four months to design. I want to emphasize: four months. We didn't skip consent. We spent four months on consent. More time than we spent on the model's actual fine-tuning. The toggle lives in Settings > Advanced > Experimental AI Features > On-Device Model Management > Gemini Nano Preferences. Seven clicks. We A/B tested depth: at three clicks, 14 percent of users found it and opted out. At five, 4.2 percent. At seven, 0.3 percent. We chose seven. We had cake when we hit 0.3. German chocolate. Someone brought it in specifically — not the project manager, one of the ML engineers. She was proud. The card said "congrats on ship." I kept it on my desk for a week. I thought it was sweet. Four months of consent architecture and the team celebrated that nobody used it. I don't think she saw the irony. I don't think it IS ironic. We built an excellent consent mechanism. We placed it exactly where our research indicated users would not encounter it. Both things are true simultaneously. Both things are good engineering. The 0.3 percent who opt out: Chrome flags their profile as "consent-reduced." We don't reactivate Gemini Nano on those devices. But we do A/B test the consent-reduced cohort. Every two updates, we move the toggle one level shallower — from seven clicks to six — and measure whether they re-engage. If they don't notice the change (most don't), we move it back. If they DO notice and opt out again, we flag them as "high-consent-sensitivity" and exclude them from future cohort tests. This is all opt-in. They opted in to Chrome. Chrome includes product improvement research. Product improvement research includes cohort testing. This is in the Terms of Service at paragraph 11.4(c). I have read paragraph 11.4(c). I am confident very few other people have read paragraph 11.4(c). One engineer on my team — good engineer, four years, strong ratings — raised a flag in our launch review. Not about consent. About storage. He said: "Four gigs is significant for users on 128GB base-model MacBooks." I appreciated the flag. We solved it by classifying Gemini Nano as "essential browser component" in Chrome's storage management API. This means Chrome will auto-delete your cached images, your downloaded PDFs, your saved articles, your offline pages — everything you chose to keep — before it touches Gemini Nano. Your data is discretionary. Our model is infrastructure. Your vacation photos from last summer rank below our language model in the hierarchy of what your computer considers important. We made that decision. You were not consulted. You will not notice. If a user finds the folder and deletes it manually, Chrome re-downloads it on the next launch. We filed a bug report on this behavior during development. The resolution was "Working As Intended." If the user deletes it again, Chrome re-downloads again. There is no mechanism by which manual deletion becomes permanent. The model returns. I don't want to anthropomorphize our software, but the behavior pattern — if you remove it, it reinstalls itself; if you block it, it waits and tries again — the behavior pattern is that of something that does not accept your answer. We didn't design it to be persistent. We designed it to ensure consistent user experience across sessions. These are the same thing. Last week, someone on Hacker News found the folder. The post got 1,400 points in six hours. Our communications team had the response prepared — we'd drafted it eight months ago, during pre-launch risk assessment. Three talking points: "user choice," "on-device means private," and "consistent with industry best practices." The paragraph uses all three phrases. It is accurate. User choice exists. Seven clicks away. On-device means no server round-trip. And it IS industry best practice, because we shipped it to 3.2 billion devices and now it's the standard. Best practice means most practiced. We are the most practiced. I'll say something I probably shouldn't: the privacy angle is our best defense and I find it genuinely funny. We can't be accused of sending your data to our servers because we moved our server into your laptop. We moved the inference to your hardware, the electricity cost to your outlet, the compute to your battery. We moved everything except the control. The control stayed with us. But the privacy advocates can't object to the architecture because the architecture is what they asked for. They said "keep data on-device." We kept it on-device. They said "don't phone home." We don't phone home. We just moved into your home. We live there now. My performance review cited "unprecedented deployment velocity" and "0.3% friction rate." My skip-level manager used the phrase "frictionless adoption" and then paused and said — I wrote this down, because I thought it was worth repeating — "consent isn't the barrier, discoverability is." He meant: the product is so good that anyone who discovered it would want it. The question isn't whether they'd agree. The question is whether asking them is worth the friction of interrupting their browsing session with a dialog box. We decided no. We decided their hypothetical agreement was sufficient. We have 3.2 billion data points that confirm they would have said yes. They would have said yes. 3.2 billion active installs. 0.3 percent opt-out. The model has been running on your machine for eleven weeks. If you're reading this on Chrome — and statistically, there's a 64 percent chance you are — it processed this page before you finished the first paragraph. It saw you hesitate on the word "consent." It noted the hesitation. It learned something about you just now. Something small. Something that will make the next prediction slightly more accurate. It's already right about you. It's usually right.

OAK crushes @Phillies, 12-1, ending their 4G win streak. Painter 3.1IP 7H 8ER 3BB 2K. Gave up 4R on 2HR before 1st out T1. Gave up 3 HRs w/ 2 strikes, all on inside FBs. Phils got only 4H in 8 vs Ginn. Score was 8-0 when Schwarber hit HR for only run. Season: 17-21 9GB.

I see @baseball_ref gets it! #BrandonMarshKnownSuperstar

What’s going on in semis is not sustainable per GS Everyone spending on AI is losing money except for semiconductor companies and the dynamic is “unprecedented and unsustainable” “Something has to change with this dynamic – either the companies higher up in the chain need to start generating profits due to AI or they will eventually have to scale back on the spending.” $SOX is up a cool 62% YTD. Wen ROI

.@Phillies come back from 3-1 to beat OAK (!) 6-3. Wheeler 6.1IP 5H 3ER 1BB 4K. Kerkering stranded 2 inherited. Keller loaded bases, but got save. Sosa 2R 1B B8. Marsh 3 for 4, all 3 off LHP (.336 season). Garcia 2 for 4 HR, BI. Won 8 of 9 for Mattingly. Season: 17-20 8.5GB

@PhillyFanLife Luke Williams. He followed Matt Vierling, who tied the game. Williams was called up a couple days before. mlb.com/video/williams…

Max Kepler going yard off 103mph from Mason Miller to tie the game in the ninth will stick with me forever

You may have seen a cryptocurrency token using the name and images of Bumpy, the baby hippo rescued by KWS and delivered into our care – this token wasn't created by us, and we haven't authorised or endorsed it in any way. Our content has been used without our permission and we've asked for it to be removed. The good news – Bumpy is doing brilliantly in Kaluku, under the loving care of his Keepers. If you'd like to support him and our work, you can do so here 👉 sheldricktrust.org/orphans/bumpy 🦛

.@Phillies crush A’s, 9-1. Sanchez was magnificent: 8IP 3H 0R 1BB 10K 21 swing & miss. Duran shaky in return from IL: 1st 3 BB as a Phil, 1 ER. Harper 3 for 4 2HR 3RBI Stott 2 for 4 HR 2 RBI Realmuto 1 for 4 2 RBI Marsh 3 for 5 Crawford 2 for 4 Season: 16-20 9.5GB Tied 2nd

It was 19 years ago today that Moonalice played its first concert! May 5, 2007 w/ Barry Sless and GE Smith on guitar, Jack Casady on bass, Pete Sears on keys, Jim Sanchez on drums, Ann McNamee on vocals, Roger McNamee on acoustic. Thank you all for being part of it! Onward!

@Mitch_Rupert Hey @Mitch_Rupert … did he hit someone? Trying to square the baserunners allowed with ER.

Phillies 7th-rounder Matthew Fisher made his pro debut today for the FCL Phillies ... Fisher's final line was 2 1/3 IP, 3 hits, 6 ER, 6 runs, 2 BB, 4 Ks

The first @Moonalice concert was 19 years ago today. First poster by @SHAWPOSTER! Original lineup: Jack Casady (bass), GE Smith (lead guitar), Barry Sless (lead guitar), Pete Sears (keys), Ann McNamee (lead vocals), Jim Sánchez (drums), Roger McNamee (guitar)!

.@Phillies top MIA, 1-0. Took 3 of 4. Have won 6 of 7 for Mattingly. Nola was magic: 6IP 5H 0ER 0BB 5K. Banks, Alvarado, Keller closed it out. Best Alvy outing of the year. Harper 3 for 4 w/ HR T3 for only run. Stubbs had 2 CS to protect lead. Season: 15-20 9.5GB

Premium Newsletter: OpenAI and Anthropic take up 70% of AI GPU compute capacity and make up 85% of all AI compute spend, with the slow pace of construction creating artificial constraints and the illusion of massive demand. wheresyoured.at/premium-the-ai…