173129966
submission
Rei writes:
In a blog post today, Bluesky, the social media network founded by Jay Graber, announced that they have finally opened to federation. Users can now operate their own PDS (backend) servers. How to do so is discussed on the developers' blog and a new Discord channel for PDS administrators.
As the blog notes, there are key differences between ATProto/Bluesky federation, vs. ActivityPub/Mastodon federation, including: global conversation (rather than local-server based with remote content only brought in from follows); a decentralized user account not bound to a specific host; user-composable moderation lists not inherently tied to a specific server, offsetting the need for defederation; user-composable feeds / algos, not tied to servers; and full account portability, without the need to be initiated by your server, protecting users from rogue admins or servers that disappear.
Despite the difference, a number of projects, such as Bridgy-Fed, plan to bridge Bluesky and Mastodon together, with all of Bluesky appearing as a single Mastodon server on ActivityPub, and Mastodon users being translated to a DID for ATProto calls.
145140672
submission
Rei writes:
After a Tesla crash in the Woodlands killed two people last Saturday night, news reports were quick to jump to the conclusion that Autopilot (or even FSD) was being used and led to the strange crash, in which investigators reported nobody in the driver's seat, one victim buckled in in the front passenger seat, and the other buckled in behind them. On Twitter, however, Autopilot users were quick to question this account, noting that Autopilot can't be enabled on a road lacking lane lines, the speed and acceleration were far higher than Autopilot allows, and numerous other problems. Now Elon Musk has weighed in with the first official statement since the crash. Responding to a user questioning the reporting: "Your research as a private individual is better than professionals @WSJ! Data logs recovered so far show Autopilot was not enabled & this car did not purchase FSD. Moreover, standard Autopilot would require lane lines to turn on, which this street did not have."
What actually caused the crash and why nobody was found in the driver's seat remains unclear at this point; analysis of the logs and investigation of the crash site remains ongoing.
144851304
submission
Rei writes:
Having moved from pigs to rhesus macaques in pursuit of the goal of hopefully beginning human trials by the end of the year, Neuralink has continued their recruitment drive with a pair of videos showing their latest progress. In the first video, they show how they train the macaque to control a joystick with its mind, and how after associating the neural signals with intent, they can disconnect the joystick and the macaque continues to be able to operate the training interface solely through Neuralink. They then switch it over to controlling a cursor in Pong (picture-in-picture showing synapses here). Even with the game set to high speed and with the distraction of his banana-milkshake reward, the macaque puts out an impressive gaming performance.
Musk expects the first commercial product to enable a paralyzed person to interact with a smartphone faster than a healthy person using their thumbs.
142854990
submission
Rei writes:
On 15 February, 2021, the paper Discrete SARS-CoV-2 antibody titers track with functional humoral stability was accepted for publication by the prestigious journal Nature — interesting not only for being a large-cohort study on COVID-19 reinfection, but for the presence of one of its coauthors: one Elon Reeve Musk. According to reporting, Musk — concerned in April 2020 with maintaining the schedule for the SpaceX crewed launch in May and wanting to make sure that an outbreak wouldn't set back plans — contacted academic researchers and worked with them to set up an antibody testing research programme. Over 4000 SpaceX employees volunteered and were provided with periodic free testing at work to look for infection and monitor previously-infected people for reinfection. The programme gave SpaceX an advance heads up about upcoming threats, such as the growing wave in Texas in June, and continues to this day, with a new focus on mutant COVID strains.
The primary results of the study? Past infection provides a strong, although not perfect, barrier to reinfection; the level of antibodies strongly indicate the level of risk of reinfection; and this bodes well for vaccines, which tend to result in much higher antibody levels than infection.
138345402
submission
Rei writes:
After having recently moved from a private to a public beta, the numbers of SpaceX Starlink users receiving their hardware (dubbed "Dishy McFlatface" by SpaceX) has increased, and along with it, the number of people reporting speed tests. Of particular interest has been performance in adverse weather. While the network is still being deployed, Pacific Northwest have been reporting rainy-weather download speeds ranging from ~95-140Mbps, upload from 9-18mbps, and ping times from 32-34ms. The upper range is surprisingly not that different from other November clear-weather data (which, while dramatically higher than September reports, is still supposed to improve over the coming year). The tolerance to adverse weather is likely due multiple satellite paths as well as phased array tuning overcoming wind buffeting. SpaceX does plan to add additional higher frequency V-band transmission to future Starlink satellites, but this should suffer more bad-weather attenuation than the current Ka/Ku-band.
137903412
submission
Rei writes:
According to an email sent out to the Starlink mailing list, Starlink is now moving from private, free, invite-only beta to a much larger, subscription-based public beta. Bandwidth estimates have risen to 50-150Mbps, while latency remains similar, at 20-40ms. This is expected to decrease to 16-19ms by summer of 2021. As it is a beta, the email cautions that "There will also be brief periods of no connectivity at all" as they enhance the system. Pricing involves an antenna purchase ($500) and a $99/mo subscription rate. There is no data cap. The beta currently only appears to be for the northern US and Canada, but SpaceX expects to quickly move further south; "near global coverage" is targeted at summer of 2021.
137704954
submission
Rei writes:
"Oh, it's going dude!" "But there's no lanes — there's cars on the side of the road!" "DUDE, its navigating through it ALL, bro!" ... "It paused to look DUDE!" "There's NO LANES! Elon, you madman!" Such was one of the many reaction videos to come out overnight as Tesla released a major upgrade of Autopilot to a limited public beta. Complete with a new LIDAR-like visualization system, the car now provides a detailed display of how it perceives the world as it dodges parked cars, takes unprotected left turns with cross traffic and pedestrians, etc. No issues have been reported thusfar, although one driver aborted after a roundabout due to a panicked passenger who didn't like how it exited into a lane with parked cars ahead (though the screen showed that it planned to change lanes to go around them).
The new version is the result of a long-running "4D" rewrite at Tesla, to overcome local maxima in earlier versions. Instead of processing each camera individually as a static series of frames, the neural net now processes a unified stitch of inputs over time, allowing context to persist between cameras and between frames. This in particular enhances parallax processing of depth input, both in terms of parallax between camera viewpoints and parallax between vehicle positions. If the public beta goes well, Elon is hopeful that Tesla will be able to roll out the new version broadly by the end of the year.
129456880
submission
Rei writes:
A paper recently published by The American Society of Tropical Medicine and Hygiene adds further support to recent CDC guidelines for minimizing the use of invasive ventilators. As physicians had been voicing concern that doctors were being too eager to put patients on invasive ventilation and may be doing more harm than good, the investigators looked into outcomes of intubated patients vs. non-intubated patients experiencing hypoxia. Unlike with other forms of pneumonia, they found that COVID-19 patients were unusually damaged by invasive ventilation but also able to tolerate higher levels of anoxia — to the point that one doctor recalls having to tell patients to get off their cell phones so that they could be intubated. The recommendation is that guidelines be adjusted to discourage invasive ventilation unless a patient is physically struggling to breathe, rather than relying strictly on oxygen levels; otherwise, the use of non-invasive ventilation, such as CPAP and BiPAP, should be encouraged. When invasive ventilation is used, oxygen levels should be minimized in order to reduce the risk of damaging healthy tissue.
124509544
submission
Rei writes:
On Twitter yesterday evening, Elon Musk announced an upcoming AI party/hack-a-thon to be held at his house, and that invites will be going out soon. Asked whether a person needs to have a Ph.D to attend, Musk replied no — "All that matters is a deep understanding of AI & ability to implement NNs in a way that is actually useful (latter point is what’s truly hard). Don’t care if you even graduated high school." The hack-a-thon appears to be Musk's latest attempt to accumulate AI talent; last fall, Tesla acquired AI startup Deepscale in order to bring its engineers into its team, while days ago Musk posted an AI job ad on his Twitter page.
117988296
submission
Rei writes:
After two profitable quarters last year, Tesla was hit by a perfect storm of filled US backlog, S/X cannibalization by Model 3, a botched international launch, and price cuts due to US tax credit phaseouts, leading to a very poor Q1 showing. While cashflow went positive in Q2, profits remained elusive, and — relying on lower-cost Model 3 variants with minimal US tax credits — expectations for Q3 weren't much better.
Instead, Tesla posted a blowout quarter: $5,3B record cash on hand, profits ($143M GAAP, $342M non-GAAP), margins rising from 18.9% to 22.8%, and sizeable growth in both solar and storage. Across the board, the company ran ahead of schedule: volume production of Model Y is pulled forward to next summer; Gigafactory 3 in Shanghai is producing cars and awaiting final sales certification after being built up from a muddy field in 10 months at a third the capital cost per vehicle; Semi (previously suggested as slipping to 2021) is back to 2020 production; and the production version of the solar roof tiles will be launching at an event on Thursday. The new, shipping crate-format Megapack energy storage products start being installed this quarter. As for vehicles, the company continues to be production constrained, with significant wait times on new orders in all markets; annual production and sales guidance of 360-400k was reiterated. Model S/X production is being raised to make up for new demand for the "Raven" update. On the self-driving front, while the company launched Smart Summon at the end of Q3, only $30M of revenue was recognized because of it; half a billion dollars of unrecognized Full Self Driving (FSD) revenue remains on the books for future quarters. The company reiterated guidance of FSD being "feature complete" (handling all driving from driveway to destination, with supervision) by the end of this year at least as a limited prerelease, and capability for unsupervised driving by the end of next year, limited by the rate of regulatory approvals. Also announced as upcoming in the next few weeks: OTA upgrades for range on new Model S/X vehicles, a 3% OTA performance improvement to S/X, and a 5% performance improvement for Model 3.
During the earnings call, Musk credited the surge in progress in Tesla's non-core divisions to being able to dedicate more engineering and financial resources to them after stabilizing Model 3 production rates and costs. Tesla's stock surged 20% in aftermarket trading, equivalent to the company's second-highest percentage gain ever, and its highest in absolute terms.
110121576
submission
Rei writes:
On the face of it, sodium seems like about the worst thing you could make a wire out of — it oxidizes rapidly in air, releases hot hydrogen gas in water, melts at 97,8C, and has virtually no tensile strength. Yet, in the late 1960s and early 1970s, the Nacon Corporation did just that — producing thousands of kilometers of high-gauge sodium wiring for electrical utilities — and it worked surprisingly well.
While sodium has three times the (volumetric) resistivity of copper and nearly double that of alumium, its incredibly low density gives it a gravimetric resistivity less than a third of copper and half of alumium. Priced similar to alumium per unit resistivity (and much cheaper than copper), limitless, and with almost no environmental impact apart from its production energy consumption, sodium wiring proved to be much more flexible without the fatigue or installation damage risks of alumium. The polyethylene insulation proved to offer sufficient tensile strength on its own to safely pull the wire through conduits, while matching its thermal expansion coefficient. The wiring proved to have tamer responses to both over-current (no insulation burnoff) and over-voltage (high corona inception voltage) scenarios than alumium as well. Meanwhile, "accidental cutting" tests, such as with a backhoe, showed that such events posed no greater danger than cutting copper or alumium cabling. Reliability results in operation were mixed — while few reliability problems were reported with the cables themselves, the low-voltage variety of Nacon cables appeared to have unreliable end connectors, causing some of the cabling to need to be repaired during 13 years of utility-scale testing.
Ultimately, it was economics, not technical factors, that doomed sodium wiring. Lifecycle costs, at 1970s pricing, showed that using sodium wiring was similar to or slightly more expensive for utilities than using alumium. Without an unambiguous and significant economic case to justify taking on the risks of going larger scale, there was a lack of utility interest, and Nacon ceased production.
107862702
submission
Rei writes:
Today, the results of a study by the MIT Center for Transport and Logistics on autonomous system driver attentiveness were released, and the results were conclusive: "drivers do not appear to over-trust the system to a degree that results in significant functional vigilance degradation in their supervisory role of system operation". The study, involving 323,384 miles driven (34,8% on autopilot) and 8682 "tricky situations" identified. Of the "tricky situations", 0% of incidents involved slow driver responses or missed detections; 4,5% rapid/timely responses; 90,6% anticipatory reaction (preventing the situation from occurring); and 4,9% "other". The study suggests that this is the result of two effects: 1) drivers effectively learn the limits of the system through usage; and 2) "tricky situations" are common enough so as to prevent excess trust by the driver in the system — creating the counterintuitive result that the better the systems become, the worse the driver may become. While the study is limited by the age of the vehicles (under a quarter were even running HW2, vs HW3 which is being released now — and due to the length of the study, most of the miles were accumulated on older software versions), it offers positive conclusions — but also a precaution — about the integration of humans and driver assist systems.
In other news, Tesla has announced an April 22 Autonomy Investor Day to showcase the capability of its development versions of the software in city driving, and has started rolling out stoplight detection, no-confirmation automated lane changes and exits, and a limited rollout of advanced summon (navigates through parking lots without a driver).
107039656
submission
Rei writes:
For anyone who thought that there was too much Tesla news this weekend, I've got some bad news for you — Elon Musk just announced the unveiling of the Model Y SUV on March 14th at Tesla's LA Design Studio. A surprising number of details were revealed, including non-falcon doors, a similar appearance to the Model 3, and pricing about 10% more than a Model 3, with slightly lower range, due to the increased mass and cross section. The unveiling will not include the Tesla pickup truck; that will be later this year. Model Y is to share 75% of its hardware with Model 3 to simplify the development process, with volume production targeted for late 2020, and initial production in early 2020.
Musk also stated: "First public Tesla V3.0 Supercharger Station goes live Wed 8pm" (V3 is the much awaited new generation of higher power, cheaper to operate Superchargers).
106185964
submission
Rei writes:
Yesterday, Tesla reported their 4th quarter earnings, representing their second consecutive profit. While earnings per share missed analyst expectations ($1,93 vs. $2,20), revenue beat expectations by around $100M and free cash flow ($910M) was more than double the First Call consensus of $395M. Model 3 margins were maintained at an impressive >20% level despite significant reductions in the average sale price in Q4; labour hours fell by 20% in Q4 and 65% in the second half of 2018 alone. With $3,7B in the bank, Tesla is now well positioned to repay its $920M March convertible bond obligations in cash. Severance costs and an increase in inventory in transit due to shipments to Europe and China are expected to hurt Tesla's profits in Q1, but guidance for Q2 onward in 2019 is strong. Highlights planned for 2019 include introduction of faster V3 Supercharging early in the year, Model Y and pickup unveiling in the middle of the year, base Model 3 unveiling in the middle of the year, and full-vehicle production in the under-construction Shanghai Gigafactory by the end of the year — the first wholly foreign-owned auto plant in China, which has seen extensive governmental support.
Despite a generally positive earnings report and conference call, the atmosphere was soured by the news that Tesla's 11-year Tesla veteran CFO Deepak Ahuja was re-retiring. Having previously retired in 2015, Deepak returned to Tesla in 2017 to replace outgoing CFO Jason Wheeler. Ahuja will remain with the company for several months as CFO and then become a senior advisor, while his protege Zach Kirkhorn fills his role. The market reacted negatively to the news, with Tesla trading down 4,5% premarket.
103779183
submission
Rei writes:
When Tesla announced late last year that it was targeting sustained profitability in 2H 2018, reaffirming this target throughout the year, the markets reacted with skepticism. Indeed, despite repeated insistence that the company had no need for a capital round, news analysts have treated the concept of Tesla dilution to raise more capital as inevitable and urgent to pay off convertible bonds next spring, even suggesting insidious theories that the reason it hadn't was that it "couldn't". Well, today Tesla put the doubts to rest with a blockbuster Q3 report — not simply eking out a profit and small free cash flow growth, but $2,92 per share profit and $881M FCF — almost raising the entire value of their convertible bond debt in a single quarter. While many were skeptical about Tesla's claims that it would go from near zero profit margin on Model 3s to their claimed target of 15%, Tesla instead hit a >20% margin on the Model 3 (now the highest-revenue car in the US), with a 25,8% overall automotive gross margin. This was all achieved with only $52M of ZEV credits claimed this quarter.
While Tesla bears will likely claim that this quarter was a one-off that won't be repeated, Tesla reiterates guidance for sustained profitability from herein, barring a force majeure event.
