重新定义 前向传播函数

<span style="color:#ff0000 !important;">返回的是我们归因的对象——最终的预测！</span>

def squad_pos_forward_func(input_ids, attention_mask, position=0): pred = model(input_ids, attention_mask) # 获取预测结果 pred = pred[position] # 当 position 为 0 时，取的是起始位置所在的分布，为 1 时，取的是结束位置所在的分布 return pred.max(1).values # 取分布的最大值，即预测结果

⚠️ 注意这个函数，这个就是自定义的 forwad_func 函数！

predict 函数返回的是一个元组，分别起、止位置的预测logist → pred[0]，pred[1] 就是起始位置的分布，是一个矩阵

关键词模版最好与聚光关键词模版保持一致，减少用户操作成本

• 当搜索-手动出价-优化目标点击量-深度优化目标站外时，竞价策略只有稳定成本
• 当搜索-手动出价-优化目标点击量-深度优化目标互动时，竞价策略只有稳定成本
• 当搜索-手动出价-优化目标点击量-深度优化SOC时，竞价策略只有稳定成本

Known historical users of the Royal KMG:

• Edward Abbey
• John Ashbery
• Saul Below
• Johnny Carson
• Joan Didion
• Bernard Kalb
• Elia Kazan
• Helen Keller (may have been a KMM)
• Grace Metalious
• Arthur Miller
• Carl Reiner
• Fred Rogers
• Rod Sterling
• George Sheehan
• Wallace Stegner

• Production MLOps on GCP rarely relies solely on Vertex AI; teams use core GCP services like Cloud Composer, Cloud Build, and BigQuery for robust workflows.
• Vertex AI excels at managed training, pipelines, model registry, endpoints, and evaluations but lacks full CI/CD, governance, security, and cost control.
• Real stack includes: Cloud Build (CI/CD), Artifact Registry (images), Terraform (IaC), Secret Manager, Cloud Monitoring/Logging, BigQuery (metadata/drift).
• Architecture layers: Source control → CI/CD → Data (BigQuery/GCS) → Vertex AI execution → Deployment (Cloud Run/GKE) → Observability.
• Reasons to avoid heavy Vertex AI: vendor lock-in, cost opacity, limited flexibility for custom auth, traffic control, multi-cloud needs.
• Alternatives: Cloud Composer (orchestration), Compute Engine/Cloud Batch (training), GCS+MLflow (registry), Cloud Run (serving) for portability and efficiency.

Is there an R function which translates local time into universal time ? I suppose this could help. Sorry for errors, English isn't my native language

促進

批判的

• AWS CEO Matt Garman argues against replacing junior developers with AI, calling it "one of the dumbest ideas."
• Juniors excel with AI tools due to recent exposure, using them daily more than seniors (55.5% per Stack Overflow survey).
• They are cheapest to employ, so not ideal for cost-cutting; true savings require broader optimization.
• Cutting juniors disrupts talent pipeline, stifling fresh ideas and future leaders; tech workforce demand grows rapidly.
• AI boosts productivity, enabling more software creation, but jobs will evolve—fundamentals remain key.

Hacker News Discussion

• AI accelerates junior ramp-up by handling boilerplate, APIs, imports, freeing time for system understanding and learning.
• Juniors ask "dumb questions" revealing flaws, useless abstractions; seniors may hesitate due to face-saving or experience.
• Need juniors for talent pipeline; skipping them creates senior shortages in 4-5 years as workloads pile up.
• Team leads foster vulnerability by modeling questions, identifying "superpowers" to build confidence.
• Debates on AI vs. docs struggle: AI speeds answers but may skip broader discovery; friction aids deep learning.

Summary of "How to Lower Cholesterol? Diet, Supplements, or Statins?"

Guest: Magdalena Kaczan (Lipidologist)

The video provides an extensive overview of cholesterol management, the mechanism of atherosclerosis, and the roles of lifestyle, genetics, and medication in cardiovascular health.

1. Understanding Cholesterol and Lipoproteins

• The Nature of Cholesterol: Cholesterol is an essential fatty substance required for building cell membranes and producing hormones [00:02:55].
• The Role of Lipoproteins: Since cholesterol is a fat, it cannot travel alone in the blood. It is carried by "packages" called lipoproteins. The most problematic ones contain Apolipoprotein B (ApoB), which allows them to penetrate arterial walls [00:04:30].
• LDL vs. HDL: * LDL (Low-Density Lipoprotein): Often called "bad" cholesterol. High levels are a primary driver of plaque buildup [00:05:31].
  • HDL (High-Density Lipoprotein): Generally "good" as it transports cholesterol back to the liver, though it can become dysfunctional in some cases [00:06:04].
• The Importance of ApoB: ApoB is increasingly seen as a more accurate marker than LDL alone because it counts the total number of atherogenic (plaque-forming) particles [00:32:09].

2. The Process of Atherosclerosis

• Infiltration: Lipoproteins (like LDL) enter the arterial wall (intima) through a process called transcytosis [00:07:45].
• Oxidation and Inflammation: Once inside the wall, LDL particles oxidize. The immune system views them as intruders; macrophages "eat" them and turn into "foam cells," triggering chronic inflammation [00:08:13].
• Plaque Formation: Over time, a "lipid core" forms, surrounded by a fibrous cap. If this plaque ruptures, a blood clot forms, which can lead to a heart attack or stroke [00:13:07].

3. Risk Factors and Individual Norms

• Personalized Norms: There is no single "normal" cholesterol level. Targets depend on an individual's 10-year cardiovascular risk (based on age, smoking, blood pressure, etc.) [00:20:01].
• Lipoprotein(a) [Lp(a)]: This is a genetically determined, highly aggressive form of LDL. It acts as an "accelerator" for heart disease and should be tested at least once in a lifetime, as it isn't lowered by traditional diet or exercise [00:36:10].
• Metabolic Factors: High triglycerides, insulin resistance, and obesity significantly worsen the quality of LDL particles, making them smaller, denser, and more dangerous [00:28:22].

4. Dietary Strategies

• Saturated Fats: High intake of animal fats (butter, lard, fatty meats) and certain plant fats (coconut/palm oil) increases LDL levels [00:43:04].
• The Power of Fiber: Soluble fiber (found in oats, legumes, and psyllium) binds bile acids in the gut, preventing the reabsorption of cholesterol [00:45:24].
• Plant-Based Fats: Replacing saturated fats with polyunsaturated and monounsaturated fats (olive oil, nuts, fatty fish) is a primary dietary intervention [00:44:46].
• Carbohydrates and Triglycerides: Excess simple sugars and alcohol are the main drivers of high triglycerides [00:47:48].

5. Pharmacological Treatment (Statins)

• Safety Profile: Statins are described as some of the safest drugs in cardiology [00:01:08].
• Beyond Lowering LDL: Statins do more than lower cholesterol; they have "pleiotropic" effects, meaning they stabilize existing plaques and reduce systemic inflammation [00:56:33].
• Side Effects and the "Nocebo" Effect: * Muscle pain occurs in about 9% of patients in clinical trials, but many subjective complaints are due to the nocebo effect (expecting side effects because of negative publicity) [01:03:06].
  • True statin intolerance is rare; switching to a different type or dose of statin often resolves issues [01:01:15].
• Liver Impact: Serious liver damage is extremely rare (1 in 100,000). Minor elevations in liver enzymes are usually temporary as the liver adapts [01:04:05].

6. Supplements and "Nutraceuticals"

• Supplements vs. Medication: Supplements like berberine or red yeast rice (monacolin K) are not substitutes for medication in high-risk patients (e.g., those who have already had a heart attack) [01:09:46].
• Red Yeast Rice: Contains monacolin K, which is chemically identical to lovastatin. While "natural," it can still cause the same side effects as prescription statins [01:11:14].
• Coenzyme Q10: While statins can lower CoQ10 levels, clinical studies do not definitively show that supplementing it reduces muscle pain [01:06:19].

7. Key Takeaways for Longevity

• Start Early: Prevention is more effective than treating advanced disease.
• Test Extensively: Go beyond a basic lipid panel; request ApoB and Lp(a) tests [01:13:05].
• Continuity: Lifestyle changes and medications are long-term commitments. If you stop the intervention, the risk levels typically return to their baseline [01:14:11].

5 Surprising Longevity Drugs – Comprehensive Summary

1. Study Background & Methodology * The Cohort: The study analyzed data from the UK Biobank, involving 501,169 participants aged 37 to 73, followed over a period of approximately 14 years [00:03:42]. * Prescription Data: Researchers examined nearly 56 million prescriptions issued to roughly 222,000 patients [00:03:58]. * Control Pairing: To determine the effect of a drug, patients taking a specific medication were paired with "control" subjects of similar age, sex, and health status (e.g., matching two diabetic males) who did not take the drug [00:06:46]. * Endpoint: The study used mortality (death) as the primary hard endpoint, as it is the most objective and difficult to manipulate in medical research [00:01:27].

2. Key Risk Factors for Mortality * Smoking: The highest risk factor, with a Hazard Ratio (HR) of 2.0 (doubling the risk of death) [00:04:42]. * Cancer: HR of 1.88 [00:05:00]. * Age: HR of 1.72 [00:06:05]. * Diabetes: HR of 1.65 [00:05:22]. * Sex: Being male carried an HR of 1.64 [00:05:56].

3. The Most Correlated Drugs with Longevity (The "Winners") * SGLT2 Inhibitors (Flozins): The top performer with a 36% reduction in mortality risk (HR 0.64). These drugs cause the body to excrete glucose through urine independently of insulin. They also act as a "weak ketosis," increasing ketones and LDL cholesterol while protecting blood vessels [00:15:50], [00:23:03]. * PDE5 Inhibitors (e.g., Viagra/Sildenafil, Cialis/Tadalafil): * Tadalafil (Cialis): Showed up to a 28% reduction in mortality risk at a 10mg dose (HR 0.72) [00:19:51]. * Sildenafil (Viagra): Showed a 15% reduction at a 50mg dose (HR 0.85) [00:20:19]. * Mechanism: These drugs stabilize Nitric Oxide (NO) levels, maintaining healthy arteries and preventing cardiovascular incidents [00:18:21]. * Estrogens (Hormone Replacement Therapy): Women taking estrogens saw a 24% reduction in mortality risk (HR 0.76). Positive results were seen across various forms, including oral, transdermal, and vaginal [00:13:50]. * Naproxen: A non-steroidal anti-inflammatory drug (NSAID) that showed a 10-11% reduction in mortality risk. Unlike Ibuprofen (2-hour half-life), Naproxen stays in the body for 17 hours, effectively blocking COX enzymes and reducing blood clotting (thromboxane) [00:17:36], [00:25:26]. * Atorvastatin (Statins): While statins as a group had a minimal effect (3% reduction), Atorvastatin specifically showed a 13% reduction at 20mg. However, higher doses (80mg) actually increased the risk of death [00:16:31].

4. Surprising "Losers" or Neutral Drugs * Metformin: Long considered a longevity staple, it showed no significant effect on lifespan in this specific cohort (HR 1.01) [00:11:22]. * ACE Inhibitors: Despite being common for blood pressure, they correlated with an 11% increase in mortality risk [00:10:36]. * Morfine & Opioids: Correlated with a 400%+ increase in mortality risk (HR ~5.5), likely due to the terminal conditions (cancer, post-surgery) for which they are prescribed [00:08:16]. * Paracetamol: Correlated with a 48% increase in mortality risk (HR 1.48) [00:08:50].

5. Critical Insights * Correlation vs. Causation: Most drugs (92% of the 169 significant ones) showed a negative correlation with lifespan, largely because people who need medication are generally in poorer health [00:07:42]. * Flozin Paradox: SGLT2 inhibitors protect the heart and extend life significantly even though they increase LDL cholesterol, challenging the traditional view that lowering cholesterol is the only path to heart health [00:23:13]. * The Role of Nitric Oxide: PDE5 inhibitors are highlighted as "longevity drugs" of the future because they restore physiological arterial regulation [00:19:35].

Now that's some solipsism.

I wish people's works were more online—more accessible (preferably, with stable, citable URLs). But to say that the people themselves should be more public in the sense that the author means here is just a continuation-without-lessons-learned of the modern (Facebook and onwards) social networking era.

It's great if all of a person's contributions to, say, mailing lists are preserved and available—and aren't just holes, missing from the record.* That's different from whether or not it's great to be able to click on that person's name, find a profile page for them, and then encounter an exhaustive, reverse chronological feed of all of their activity across all mailing lists. Mailing lists rarely enable this, but virtually every modern social network does, and they're actually built around it.

The former is topic-based indexing, and the latter is actor- (person-) based indexing. Actor-based indexing is bad, and we know that it's bad.

Actor-based indexing is like running into someone you know (or a stranger, even) at the post office and then, through some mechanism where their physical presence wired to some metaverse data source, being able to perform some tap/gesture at the virtualized floating bubble over their heads that lets you see all the public places where you would have seen them earlier that day if you had been at any one of those, and then having a log of every interaction for the day prior, and the day before that, and so on, stretching back over their entire life, including the grocery store, the restaurant, the houseparty they were at, their work, etc. This would be bad. That means it's not good. And it's not good "online", either, for the exact same reasons.

* as unfortunately, many Mastodon (and other ActivityPub-powered) interactions turn out; Mastodon happens to achieve the worst of both worlds!

How to Learn 10x Faster? – Summary of Bartosz Czekała’s Insights

1. The Failures of Traditional Learning * The "Sieve" Effect: Traditional learning methods (reading textbooks, filling in blanks) are highly inefficient, resembling an attempt to carry water in a sieve [00:03:48]. * The Forgetting Curve: Based on Ebbinghaus’s research, without deliberate reviews, we lose about 80% of new information within a month [00:05:10]. * Passive vs. Active: Reading and highlighting are "passive encoding" methods that rarely result in long-term retention [00:03:52].

2. The Foundation: Spaced Repetition Systems (SRS) * Algorithms over Intuition: Manual planning of reviews is impossible for large amounts of data. Using software like Anki is essential [00:19:12]. * How it Works: The program calculates the optimal interval for the next review (e.g., 1 day, 3 days, 1 week, 1 month) based on your self-assessment of how well you remembered it [00:13:44]. * Reducing Decision Fatigue: The system makes learning "binary"—you simply open the app and complete whatever tasks are scheduled for that day [00:14:54].

3. Techniques for Creating Effective Flashcards * Atomization: Each flashcard should contain exactly one question and one specific piece of information in the answer [00:26:09]. * Deep Encoding: Creating your own flashcards (rather than using pre-made decks) forces the brain to manipulate information, building stronger neural pathways [00:35:47]. * Contextualization: For language learning, the deepest encoding comes from creating sentences using the new word rather than just memorizing a definition [00:30:13].

4. Language Learning Strategy (Case Study: Czech in One Month) * Pareto Principle: Start with frequency lists—memorize the words used most often in daily communication [00:46:36]. * Reference Points: Use analogies from languages you already know (e.g., using Polish or Russian roots to learn Czech) to drastically speed up the process [00:52:38]. * Self-Talk: Actively producing speech out loud, even to yourself, is the deepest form of active encoding [00:50:27].

5. Diet and Lifestyle for Brain Optimization * The Danger of Sugar: Glucose spikes and high glycemic index meals hinder memory. Chronic high blood sugar can even lead to brain atrophy [00:02:52]. * Intermittent Fasting (16/8): Fasting increases blood flow and oxygen to the prefrontal cortex, enhancing logical thinking and concentration [00:14:10]. * Ketones: Low-carb diets and ketosis stabilize neuronal networks and provide "mental clarity" often missing in high-carb diets [01:13:14].

6. Critique of Supplements and "Nootropics" * False Hopes: Most "smart drugs" provide negligible benefits (around 1%) compared to the massive gains from a proper learning system and diet [01:16:47]. * The Real Nootropic: The best way to learn faster is to accumulate knowledge. The more you know, the easier it is to "attach" new information to your existing mental framework [01:17:34].

• Ineffectiveness of the Education System: Traditional education often focuses on delivering vast amounts of material without teaching the actual tools or techniques for effective memorization and information retention.
• Passive vs. Active Learning: Scientific research (notably by Craik and Watkins in 1973) demonstrates that passive repetition has almost zero impact on long-term memory and knowledge building.
• Definition of Passive Reproduction: This refers to "mindless" repetition where information is maintained in short-term memory without any cognitive processing or mental engagement (e.g., repeating a phone number just long enough to dial it).
• Common Mistakes: Rote memorization of facts/dates, highlighting text without deep thought, and copying notes verbatim are largely ineffective and represent a significant waste of time.
• The Importance of Active Engagement: To build lasting knowledge, one must engage with the material through "active learning"—this involves speculating, questioning, drawing connections, and integrating new facts into existing mental models.
• Efficiency and Time Management: Using active methods can allow a learner to process in 15 minutes what might otherwise take 5 to 10 hours using passive, repetitive methods.
• Building Mental Frameworks: True specialization in any field requires effective memory tools to connect isolated bits of information into useful, functional models of knowledge.

EXTENDED SUMMARY: How to Feel Good and Be Healthy on a Budget

In this deep-dive conversation, Bartosz Czekała explores the intersection of biology, psychology, and lifestyle, providing practical advice on how to optimize health without spending a fortune.

1. The Biological Root of Mental Health

• Inflammation and the Brain: Czekała argues that mental health issues like depression and anxiety are often driven by systemic inflammation. Chronic inflammation increases the permeability of the blood-brain barrier, allowing pro-inflammatory molecules to affect the brain [00:00:47].
• Serotonin Inhibition: Inflammation doesn't just make you feel physically ill; it actively blocks the uptake of serotonin and lowers its overall levels, mimicking or causing clinical depression [00:00:36].
• Therapy vs. Medication: He notes that while millions rely on antidepressants, psychotherapy often shows better long-term results. He emphasizes BDNF (Brain-Derived Neurotrophic Factor) as a critical marker for brain health and recovery [00:01:07].

2. Hormonal Health and Body Composition

• Fat as a Hormonal Organ: Adipose tissue (body fat) is not just stored energy; it is an active endocrine organ. The more body fat a person has, the higher the activity of an enzyme called aromatase [02:37:30].
• The Testosterone-Estradiol Balance: In men, aromatase converts testosterone into estradiol (estrogen). High levels of body fat can lead to low testosterone and physical symptoms like gynecomastia ("man boobs") [02:37:48].
• Risks of Steroid Use: Czekała warns against the misuse of exogenous testosterone (steroids), noting that supra-physiological doses are hepatotoxic (liver-damaging) and can damage the heart, often leading the body to convert excess testosterone into estrogen as a defense mechanism [02:38:09].

3. Low-Cost "Biohacking" and Lifestyle

• Ergonomics for Longevity: One of the cheapest health interventions is changing how you work. He suggests working from the floor or a mat rather than a traditional chair to maintain mobility and cardiovascular health during home office hours [00:00:26].
• Nutrition as a Foundation: He advocates for a diet rich in high-quality animal products and nutrient-dense meats as a way to prevent deficiencies and maintain hormonal balance [00:03:36].
• Nature and Circadian Rhythms: Simple, free practices like spending time outdoors, grounding, and aligning with natural light cycles are cited as powerful tools for reducing systemic inflammation.

4. Diagnostics and Critical Thinking

• Recommended Testing: To truly understand one's health, Czekała recommends testing not just Total Testosterone, but also Estradiol, DHEA-S, Androstenedione, and markers of systemic inflammation [02:38:54].
• Evaluating Science: He draws a distinction between "hard" sciences (physics/math) and "soft" sciences (psychology/sociology). In human biology, results are rarely black-and-white; the answer is almost always "it depends" on the individual context [00:22:30].

5. Conclusion

The central takeaway is that health is a result of low inflammation, balanced hormones, and intentional movement. By focusing on biological fundamentals—diagnostics, diet, and environment—one can achieve significant health improvements without relying on expensive supplements or "magic pill" solutions.

fisher

slide 37

ненормативные характеристики

ошеломляющий

Обозначена главная проблема, которая будет развита далее в статье. То как люди используют искусственный интеллект предопределяет положительное или отрицательное воздействие он оказывает.

притуплять

• Main Thesis: While coffee is often marketed as "neuroprotective," there is significant scientific evidence suggesting it may have negative effects on brain health, including a reduction in gray matter.
• Antioxidants: Coffee is a major source of dietary antioxidants, but the video argues that exogenous (external) antioxidants can interfere with the body's more effective endogenous (internal) antioxidant systems [00:04:54].
• Cerebral Blood Flow: Caffeine acts as a vasoconstrictor. Studies using PET scans show that consuming 200-250 mg of caffeine (about 2-3 cups) can reduce blood flow throughout the brain by approximately 30% [00:22:24].
• Gray Matter Impact: Research indicates that even short-term regular caffeine consumption (e.g., 10 days) can lead to a detectable decrease in gray matter volume in the medial temporal lobe [00:26:50].
• Adenosine Blocking: Caffeine works by blocking adenosine receptors, which normally signal the brain to rest. This leads to an artificial increase in stimulating neurotransmitters like adrenaline and glutamate [00:17:53].
• Genetic Variability: The speed of caffeine metabolism is largely determined by the CYP1A2 enzyme. "Slow metabolizers" can experience up to ten times higher concentrations of caffeine in their system compared to "fast metabolizers" [00:13:49].
• Toxins and Quality: Many commercial coffees, especially instant varieties, contain detectable levels of mycotoxins (like ochratoxin A). The cumulative effect of these toxins across different food sources is a potential health concern [00:10:32].
• Neuroprotection Claims: Most evidence for coffee's benefits is based on epidemiological correlations rather than clinical trials. While there may be a link to reduced Parkinson's risk, large meta-analyses have found no significant link between coffee and Alzheimer's prevention [00:33:15].
• The "U-Shaped" Rule: Any potential benefits from coffee appear to follow a U-shaped curve; consuming more than four cups a day generally eliminates any statistical health advantages and may increase risks [00:35:06].

• Non-Binary Process: Vocabulary acquisition is not a simple "know it or don't" situation; it is a gradual progression through multiple levels of familiarity [00:00:10].
• Initial Exposure: The first stage involves hearing or seeing a word for the first time. You recognize you've seen it before but don't yet know the meaning [00:00:34].
• Emerging Recognition: After looking up a word a few times, you may occasionally recall the definition, but it is not yet consistent [00:01:07].
• Conscious Knowledge: You reach a stage where you can provide the correct definition, often aided by tools like Spaced Repetition Systems (SRS) or flashcards [00:01:25].
• The "Wall of Sound" Challenge: Even if you "know" a word, there is often a delay in processing it during live audio or reading, which can cause you to fall behind in a conversation [00:02:22].
• Subconscious Processing: With more exposure, the mental translation time decreases from seconds to being processed instantly without conscious effort [00:03:31].
• Passive vs. Active Vocabulary: It takes additional time and practice for words to move from passive recognition to active use in speaking [00:04:11].
• Role of Compelling Content: While flashcards help reach initial recognition, "compelling content" like reading, podcasts, and movies is what builds the subconscious strength needed for fluency [00:05:05].
• The Power of Reading and Listening: Repeated exposure to common words in natural contexts (like books or games) reinforces knowledge until translation is no longer necessary [00:05:30].

VoiceInk , installed it.

you can have various enhancement prompts (wrt styling, type of output etc), and you can switch through keyboard shortcuts.

Has context (both window and clipboard)

connecting to ollama let's me choose the model for text enhancement

Ah, enhancement can be done locally too, by connecting to ollama.

Enhancement models (to clean up transcripts), mentions things I have locally, but suggests it uses third party external services for it.

VoiceInk recommends Parakeet (Nvidia) and Whisper large v3 Turbo as local models. I wonder what if I connect it to Euro context models (like Swiss Aperture)

Prakash Joshi Pax is Nepali (says their X account). Not sure where their business is based.

VoiceInk is open source, and on Github

Require Apple silicon, and its 'Neural Engine' what's that? You can connect to the cloud though as alternative.

VoiceInk does have a cloud based component, opt-in, for processing already transcribed text by third parties.

one time 25USD for single device usage.

allows own dictionary

Voice Ink, says it works w local models. Not sure yet, if I can set the model.

VoiceInk is said to be fully local for dictation

Only some Dutch is dealt with locally, English, French and German are all done in the cloud. So Apple Dictation not suited for my work

MacWhisper works locally, only on Mac. Uses OpenAI's model though. Really fully local?

SuperWhisper a mac only fully local tool

Can work offline / on-device, not in all languages though.

Wispr Flow uses both open (Llama) and closed LLMS, ao OpenAI . Server side though

One can opt-out of dictation data being stored on their servers, default is storing it though. I suppose this means any dictation is dealt with online, server side and not locally? Otherwise they'd said that, no?

wispr flow is California, USA based, and VC funded, ao by individuals from OpenAI, Dropbox, Coinbase

yes, agency does increase, where people realise this works locally.

History suggests this is the likely path. Most people will accept the phone, ipads and computers as they are. Like IndieWeb there is a population of people going against that current, but not a tide. Unless e.g. interoperability reqs from DSA, DMA force the issue. Sideloading is actually installing.

Two moats, app store for legitimacy and the hurdle for creation. Second one has become more shallow and narrow now.

Three elements for making his iphone apps Xcode (which I use) Apple Developer account (99USD / yr) AI support in coding (he uses Claude Code, vgl [[Mijn vibe coding set-up 20251220143401]]

Making personal tools makes generic ones obsolete. Yet, the generic ones do serve as starting point for inspiration and design choices. Personal iterations on top of what went before.

[[Martijn Aslander]] has made several personal tools for his iphone.

[[Martijn Aslander]] uses Wispr Flow to talk to his laptop.

Yup, also in ref to DMA, DSA

Pricing free tier of 2k words/week on laptop individual / team tier, 144E/yr

wispr flow available on Mac

but cooking meat reduces this by 70-100%?

US admin sanctioned people involved in shaping the DSA. The wilfull misreading of DMA, DSA, AIR, GDPR in the US and bigtech is a clear confirmation of its need for the European market.

Article seems too narrow in looking at the dynamics. Tech platforms are not the context, single market and market rules and access are, including outside digital. Meaning every other party dealing with platforms has a very different set of considerations when choosing platforms of any size. Loss of market access is not about the tech, but about whether there are others willing to do business with you.

The US admin escalation is likely a spasm, and if not a cause of bifurcation rather than response. The base case is shoulder shrugs anywhere outside the USA. The endpoint is no market access for non-compliant platforms of any size. Which does not mean a ban or tech blockade, but the absence of possibility to interact with the EU market as corporates, including the ad market e.g. The law of two feet is the largest fine here (not just the platforms need to be compliant, their businesspartners too, and for them to walk is the cheaper compliance path.

Not a wild card (The wild card is the zero sum behaviour of US admin), but an aimed for outcome. Standardisation, transparency and interoperability are key digital policy aims. Note that it is exactly what big tech is clamoring against at the moment.

This is not second order, but a primary policy aim for the EU digital single market.

Regulatory differences are of all time and splinternet it is not, which implies hard (tech) breaks. Additionally the DSA is unifying for Europe, part of the digital single market. It only looks like divergence to any incumbents outside the EU.

Not really, it does not favor scale, as it's progressive compliance. The next sentence says as much. One platform's bifurcation is the same as having two smaller ones, who have less compliance costs and thus won't be crushed. Which is already the case even, global platforms already cater to diff regulatory regimes (and morally questionable at that). The underlying faulty assumption is that of global platforms for everyone and everything being the desired outcome at all. SV thinking and funding is the root cause. Other paths exist, just not in their world. Zebra's not unicorns. [[Zebra bedrijven zijn beter 20190907063530]] Vgl physical e.g. the German industrial base actually is one of many medium sized orgs being market leaders in some niche, not the car manufacturers usually mentioned as such.

yes, likely in the short term. Thing is: once people globally see the outcome of those diff modes, which will they prefer? Vgl GDPR

yes, it means the USA cannot be treated as an equal or rational counterpart. n:: EU moving to adversarial interoperability in a sense? n:: Brusselscountereffect?

in contradiction to the entrenching above. Of these adtech is the key thing, and algorithms aimed for engagement (ie rage)

Not following. By def the strictest stuff applies to the largest platforms, so no moat. n:: The compliance burden is progressive, like taxes are /should be.

Yes, such paragraphs need to be up front.

yes, sort of. 'distribution layer of democracy' interesting phrase. amplification, throttling, monetisation (Freudian misspelling there?), discoverability all important. But the key thing: the platforms in question are not platforms in the strict sense, they actively shape the information there. So liability protections for platforms should not apply. Or gov also can set the boundaries of such shaping.

There are many other fines levied (all without making any dents in the behaviour fined though), the 120ME one for Twitter was the first under DSA illegal content rules (which don't specify what illegal content, but mechanisms for moderation against them)

The choice of words is wrong on many diff levels. Global / censorship / industrial complex, three diff long explanations. One would need more populist labels for the actual character of platforms (not just bigtech) afoul of the DSA

Sanctioned are Thierry Breton (EC in the previous period), and people from the Global Disinformation Index, and HateAid. Such orgs have a role in research into the inner workings of platforms.

Uh, dangerous!

It isn't, and that's a problem I think... because you've normalised the consumerist service slavery to people that can ask for whatever and get it whenever.

There is a whole bundle of rationalisations there, provided by the male enterprise.

Dated, but surely the stats haven't changed that much.

The cost of lack of skills is not just in the individual knowledge worker, it gets externalised to others to fix it, or multiplied in groups waiting on you to get something working. The incompetence spreads out.

I'm just not a scalpel person! Ha!

Basic skills often still lacking. Keyboard shortcuts first, or even knowing how to interact w interfaces through the keyboard, not the mouse

knowledge workers are the largest group of people who don't know their own tools. Vgl [[Kenniswerk is ambacht 20040924200250]]

baseline digital fluency not reached

All the hype on the potential of AI, likely the potential of just knowing how to use what we have is bigger

Knowing your tools is key to knowledge work, know your computer. It's not cute to play tech-incompetent

When silos gonna silo (and Europeans get sanctioned by the USA for making laws to hold silos to account), hypelinks are subversive and the open web is resistance.

Universal Resource Locator

Is turned into Universal Resource name with permanence and resolvability cryptographic verifiability

大幅度特

测试

春寒日暖,来煎人寿

El RGPD,es un documento marco de referencia de gran valor para El derecho continental en lo referente al derecho de intimidad y protección de datos

DOI: 10.1016/j.xpro.2025.104268

Resource: RRID:IMSR_GPT:T054816

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_GPT:T054816

What is this?

DOI: 10.1016/j.xpro.2025.104268

Resource: FlowJo (RRID:SCR_008520)

Curator: @scibot

SciCrunch record: RRID:SCR_008520

What is this?

DOI: 10.1016/j.phymed.2025.157730

Resource: (BCRJ Cat# 0278, RRID:CVCL_0132)

Curator: @areedewitt04

SciCrunch record: RRID:CVCL_0132

What is this?

DOI: 10.1016/j.phymed.2025.157730

Resource: (KCLB Cat# 80008, RRID:CVCL_0159)

Curator: @scibot

SciCrunch record: RRID:CVCL_0159

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: RRID:IMSR_JAX:000664

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_JAX:000664

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Leica Biosystems Cat# NCL-L-SYNAP-299, RRID:AB_564017)

Curator: @scibot

SciCrunch record: RRID:AB_564017

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Thermo Fisher Scientific Cat# A-21242, RRID:AB_2535811)

Curator: @scibot

SciCrunch record: RRID:AB_2535811

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Wako Cat# 019-19741, RRID:AB_839504)

Curator: @scibot

SciCrunch record: RRID:AB_839504

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Cell Signaling Technology Cat# 9167, RRID:AB_561284)

Curator: @scibot

SciCrunch record: RRID:AB_561284

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Millipore Cat# MAB377X, RRID:AB_2149209)

Curator: @scibot

SciCrunch record: RRID:AB_2149209

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: None

Curator: @scibot

SciCrunch record: RRID:AB_2338242

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Thermo Fisher Scientific Cat# A-21141, RRID:AB_2535778)

Curator: @scibot

SciCrunch record: RRID:AB_2535778

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (MP Bio Cat# 0855463, RRID:AB_2334481)

Curator: @scibot

SciCrunch record: RRID:AB_2334481

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Abcam Cat# ab124754, RRID:AB_10971665)

Curator: @scibot

SciCrunch record: RRID:AB_10971665

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Abcam Cat# ab190565, RRID:AB_2732785)

Curator: @scibot

SciCrunch record: RRID:AB_2732785

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Thermo Fisher Scientific Cat# 14-0195-82, RRID:AB_2637159)

Curator: @scibot

SciCrunch record: RRID:AB_2637159

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 135208, RRID:AB_2159184)

Curator: @scibot

SciCrunch record: RRID:AB_2159184

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (PBL Assay Science Cat# 32100-1, RRID:AB_387871)

Curator: @scibot

SciCrunch record: RRID:AB_387871

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Thermo Fisher Scientific Cat# 12-6502-82, RRID:AB_10855034)

Curator: @scibot

SciCrunch record: RRID:AB_10855034

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Abcam Cat# ab190195, RRID:AB_2716282)

Curator: @scibot

SciCrunch record: RRID:AB_2716282

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 100412, RRID:AB_312697)

Curator: @scibot

SciCrunch record: RRID:AB_312697

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BD Biosciences Cat# 745730, RRID:AB_2743205)

Curator: @scibot

SciCrunch record: RRID:AB_2743205

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BD Biosciences Cat# 563786, RRID:AB_2732919)

Curator: @scibot

SciCrunch record: RRID:AB_2732919

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 123108, RRID:AB_893502)

Curator: @scibot

SciCrunch record: RRID:AB_893502

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 101242, RRID:AB_2563310)

Curator: @scibot

SciCrunch record: RRID:AB_2563310

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 107620, RRID:AB_493527)

Curator: @scibot

SciCrunch record: RRID:AB_493527

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 138410, RRID:AB_10643582)

Curator: @scibot

SciCrunch record: RRID:AB_10643582

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (BioLegend Cat# 103114, RRID:AB_312979)

Curator: @scibot

SciCrunch record: RRID:AB_312979

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: R Project for Statistical Computing (RRID:SCR_001905)

Curator: @scibot

SciCrunch record: RRID:SCR_001905

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (IMSR Cat# JAX_029899,RRID:IMSR_JAX:029899)

Curator: @scibot

SciCrunch record: RRID:IMSR_JAX:029899

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Jackson ImmunoResearch Labs Cat# 011-000-120, RRID:AB_2337123)

Curator: @scibot

SciCrunch record: RRID:AB_2337123

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Jackson ImmunoResearch Labs Cat# 712-605-153, RRID:AB_2340694)

Curator: @scibot

SciCrunch record: RRID:AB_2340694

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Vector Laboratories Cat# S-1000, RRID:AB_2336615)

Curator: @scibot

SciCrunch record: RRID:AB_2336615

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Agilent Cat# K4001, RRID:AB_2827819)

Curator: @scibot

SciCrunch record: RRID:AB_2827819

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Jackson ImmunoResearch Labs Cat# 015-000-120, RRID:AB_2337194)

Curator: @scibot

SciCrunch record: RRID:AB_2337194

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: EthoVision XT (RRID:SCR_000441)

Curator: @scibot

SciCrunch record: RRID:SCR_000441

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: FlowJo (RRID:SCR_008520)

Curator: @scibot

SciCrunch record: RRID:SCR_008520

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Jackson ImmunoResearch Labs Cat# 711-545-152, RRID:AB_2313584)

Curator: @scibot

SciCrunch record: RRID:AB_2313584

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Agilent Cat# K4003, RRID:AB_2630375)

Curator: @scibot

SciCrunch record: RRID:AB_2630375

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Molecular Probes Cat# A-21271, RRID:AB_221448)

Curator: @scibot

SciCrunch record: RRID:AB_221448

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: None

Curator: @scibot

SciCrunch record: RRID:AB_3698328

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Thermo Fisher Scientific Cat# A-21127, RRID:AB_2535769)

Curator: @scibot

SciCrunch record: RRID:AB_2535769

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Abcam Cat# ab4729, RRID:AB_2118291)

Curator: @scibot

SciCrunch record: RRID:AB_2118291

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Proteintech Cat# 18205-1-AP, RRID:AB_2093197)

Curator: @scibot

SciCrunch record: RRID:AB_2093197

What is this?

DOI: 10.1016/j.neuron.2025.11.006

Resource: (Thermo Fisher Scientific Cat# A-21240, RRID:AB_2535809)

Curator: @scibot

SciCrunch record: RRID:AB_2535809

What is this?

DOI: 10.1016/j.neurobiolaging.2025.12.009

Resource: (IMSR Cat# JAX_012708,RRID:IMSR_JAX:012708)

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_JAX:012708

What is this?

DOI: 10.1016/j.neurobiolaging.2025.12.009

Resource: (Millipore Cat# ABN91, RRID:AB_11205760)

Curator: @scibot

SciCrunch record: RRID:AB_11205760

What is this?

DOI: 10.1016/j.isci.2025.114533

Resource: RRID:Addgene_8453

Curator: @areedewitt04

SciCrunch record: RRID:Addgene_8453

What is this?

DOI: 10.1016/j.exger.2025.113011

Resource: (IMSR Cat# GPT_N000013,RRID:IMSR_GPT:N000013)

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_GPT:N000013

What is this?

DOI: 10.1016/j.celrep.2025.116797

Resource: None

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_GPT:T055135

What is this?

DOI: 10.1016/j.celrep.2025.116781

Resource: RRID:Addgene_12253

Curator: @areedewitt04

SciCrunch record: RRID:Addgene_12253

What is this?

DOI: 10.1016/j.celrep.2025.116753

Resource: None

Curator: @areedewitt04

SciCrunch record: RRID:MGI:5450373

What is this?

DOI: 10.1016/j.celrep.2025.116717

Resource: (IMSR Cat# JAX_008069,RRID:IMSR_JAX:008069)

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_JAX:008069

What is this?

DOI: 10.1016/j.celrep.2025.116717

Resource: (IMSR Cat# JAX_016962,RRID:IMSR_JAX:016962)

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_JAX:016962

What is this?

DOI: 10.1016/j.celrep.2025.116717

Resource: (IMSR Cat# JAX_016963,RRID:IMSR_JAX:016963)

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_JAX:016963

What is this?

DOI: 10.1016/j.celrep.2025.116717

Resource: RRID:IMSR_JAX:036761

Curator: @areedewitt04

SciCrunch record: RRID:IMSR_JAX:036761

What is this?

DOI: 10.1016/j.ymthe.2025.12.037

Resource: None

Curator: @scibot

SciCrunch record: RRID:SCR_023051

What is this?

DOI: 10.1016/j.ymthe.2025.12.004

Resource: University of California San Francisco Parnassus Flow Cytometry Core Facility (RRID:SCR_018206)

Curator: @scibot

SciCrunch record: RRID:SCR_018206

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Thermo Fisher Scientific Cat# 459200, RRID:AB_10838019)

Curator: @scibot

SciCrunch record: RRID:AB_10838019

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: None

Curator: @scibot

SciCrunch record: RRID:AB_3711262

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Agilent Cat# P0217, RRID:AB_2728719)

Curator: @scibot

SciCrunch record: RRID:AB_2728719

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Jackson ImmunoResearch Labs Cat# 711-607-003, RRID:AB_2340626)

Curator: @scibot

SciCrunch record: RRID:AB_2340626

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Agilent Cat# P0447, RRID:AB_2617137)

Curator: @scibot

SciCrunch record: RRID:AB_2617137

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Cell Signaling Technology Cat# 4903, RRID:AB_10559205)

Curator: @scibot

SciCrunch record: RRID:AB_10559205

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (BD Biosciences Cat# 562205, RRID:AB_10893402)

Curator: @scibot

SciCrunch record: RRID:AB_10893402

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Thermo Fisher Scientific Cat# 12-5841-80, RRID:AB_914364)

Curator: @scibot

SciCrunch record: RRID:AB_914364

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (BioLegend Cat# 306513, RRID:AB_2089652)

Curator: @scibot

SciCrunch record: RRID:AB_2089652

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (BioLegend Cat# 330706, RRID:AB_1089242)

Curator: @scibot

SciCrunch record: RRID:AB_1089242

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (Thermo Fisher Scientific Cat# MA5-23650, RRID:AB_2608686)

Curator: @scibot

SciCrunch record: RRID:AB_2608686

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (R and D Systems Cat# IC2085A, RRID:AB_2891298)

Curator: @scibot

SciCrunch record: RRID:AB_2891298

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (BD Biosciences Cat# 561552, RRID:AB_10714781)

Curator: @scibot

SciCrunch record: RRID:AB_10714781

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: RRID:Addgene_48138

Curator: @scibot

SciCrunch record: RRID:Addgene_48138

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (BD Biosciences Cat# 562711, RRID:AB_2737738)

Curator: @scibot

SciCrunch record: RRID:AB_2737738

What is this?

DOI: 10.1016/j.scr.2025.103895

Resource: (BioLegend Cat# 330412, RRID:AB_1089198)

Curator: @scibot

SciCrunch record: RRID:AB_1089198

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (SouthernBiotech Cat# 1030-30, RRID:AB_2794300)

Curator: @scibot

SciCrunch record: RRID:AB_2794300

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (Cell Signaling Technology Cat# 5568, RRID:AB_10694505)

Curator: @scibot

SciCrunch record: RRID:AB_10694505

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (Cell Signaling Technology Cat# 19245, RRID:AB_2734735)

Curator: @scibot

SciCrunch record: RRID:AB_2734735

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (SouthernBiotech Cat# 4050-32, RRID:AB_2795963)

Curator: @scibot

SciCrunch record: RRID:AB_2795963

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (Thermo Fisher Scientific Cat# 12-8863-82, RRID:AB_891602)

Curator: @scibot

SciCrunch record: RRID:AB_891602

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (Nolan lab - Stanford Cat# 3579, RRID:AB_2864411)

Curator: @scibot

SciCrunch record: RRID:AB_2864411

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (Cell Signaling Technology Cat# 4755, RRID:AB_1264259)

Curator: @scibot

SciCrunch record: RRID:AB_1264259

What is this?

DOI: 10.1016/j.scr.2025.103890

Resource: (Thermo Fisher Scientific Cat# MA1-021-D488, RRID:AB_2536688)

Curator: @scibot

SciCrunch record: RRID:AB_2536688

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (BioLegend Cat# 801201, RRID:AB_2313773)

Curator: @scibot

SciCrunch record: RRID:AB_2313773

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (Molecular Probes Cat# A-21207, RRID:AB_141637)

Curator: @scibot

SciCrunch record: RRID:AB_141637

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (Thermo Fisher Scientific Cat# A25618, RRID:AB_2885001)

Curator: @scibot

SciCrunch record: RRID:AB_2885001

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (Agilent Cat# A0008, RRID:AB_2650473)

Curator: @scibot

SciCrunch record: RRID:AB_2650473

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (Millipore Cat# MAB4381, RRID:AB_177638)

Curator: @scibot

SciCrunch record: RRID:AB_177638

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (Sigma-Aldrich Cat# A1978, RRID:AB_476692)

Curator: @scibot

SciCrunch record: RRID:AB_476692

What is this?

DOI: 10.1016/j.scr.2025.103889

Resource: (Thermo Fisher Scientific Cat# A-21202, RRID:AB_141607)

Curator: @scibot

SciCrunch record: RRID:AB_141607

What is this?