About
Activity
1K followers
-
Di Zhu posted thisIf the KiwiSaver Act 2006 had mandated an annual increase of 0.5 percentage point in both employer and employee contribution from day one, New Zealand would have been in a so much better position now when it comes to retirement savings, and it would have made the conversation of reforming universal super a lot easier. Well, nevertheless.
-
Di Zhu shared thisI sometimes wonder what the councils are even doing to warrant such outrageous increasesDi Zhu shared thisI saw a version of this chart for the US recently. Here’s the version for New Zealand showing how consumer prices have tracked since 2002, which is when I started working full time. Over that period, overall consumer prices as measured by the CPI have risen 83%, or roughly 2.5% per year. Looking under the surface, there are some big differences in how prices for different goods have evolved. Many of the areas where we’ve seen larger price increases have been in essentials or non-discretionary areas. That includes petrol, insurance, electricity and council rates. On the downside, the areas where we’ve seen smaller price increases have generally been in discretionary areas or luxuries, like recreational items and electronics. However, a note of caution here: For items like cell phone plans, you’re probably paying more than you used to, but also getting more data. That gets captured in the CPI as a price decline. Those higher costs for essential items mean that households have less to spend on the ‘fun’ stuff that matters for their quality of life. That’s one reason why, even though inflation has slowed from over 7% in recent years to 3.1% now, many households don’t feel like they’re getting ahead. The cost of living is still squeezing their finances. We expect that inflation will be better contained over the next few years, but see the risk of continued large increases in essentials prices for some time yet. #inflation #rbnz Westpac NZ Economics Team Kelly Eckhold
-
Di Zhu shared thisOn a more serious note - NZ economy would have been more stuffed without high global prices for dairy and red meat exports, and as long as NZ is heavily relying on such exports, this is, on net, a positive thing. And no, I don’t believe butter is a household essential, but electricity absolutely is, and we need to talk about electricity a lot more than butter.
-
Di Zhu posted thisNow we can finally see the painful costs of the RBNZ’s extreme hikes. Compare that with the RBA’s patient approach, pausing the hikes at a lower but still restrictive level, and waiting. Australia has seen its workforce expand like never before despite also record migration, unemployment stayed low, and inflation still trended down to target. Its government finances are in much better shape even discounting iron ore prices, and the government thus has a lot more space to spend on long term priorities and pork barrel politics. The lesson: economic scarring should be avoided at all costs, because it takes longer than you expect to heal, and events may well disrupt the healing, like what we are seeing right now.
-
Di Zhu posted thisI think it’s appropriate for the RBNZ to convey a lower OCR trough in their May Monetary Policy Statement. Too much downside to inflation outlook to ignore. Cut below neutral if necessary.
-
Di Zhu liked thisDi Zhu liked thisAI still isn't truly intelligent. It's just patterns in data. Data engineers who understand this will become truly dangerous These patterns go beyond the standard rows / columns format that data engineers are used to which makes them seem much more "magic" but at the end of the day it's still statistics + unstructured data. So how do you move into this new world? Give your unstructured data structure! - For some document types, this is easy because it's already broken up into paragraphs and pages. Think PDF documents, books, etc. - For other document types, it's much harder because it's just a constant stream of words. Think podcast transcripts, video transcripts. You'll notice there's a continuum of structured to unstructured data: rows / columns -> JSON -> books / PDFs -> transcripts The more structured the data is, the easier it is to work with. As you break out of rows / columns, your data modeling techniques need to change as well because the data won't be shoehorned into your tables. - You need more complex data types in your tables like STRUCT, ARRAY, and ARRAY<STRUCT> to truly model the unstructured data with all its complexities - You might need to model it as a graph database using a vertex and edge table as well. Neo4j is a great option for this. GraphRAG is a very powerful technique to make your AI more powerful! My latest newsletter goes in much more detail about how to model your data for the era of AI!
-
Di Zhu liked thisDi Zhu liked thisAI is making things I once enjoyed sort of terrible…. Reddit has become unusable. It’s looking more like Moltbook every day with bots just chatting away farming for karma. LinkedIn is becoming somehow more sycophantic and empty. People are speaking more and saying much less. Even at work, 50% of the emails and Slack messages I receive are suspiciously AI-like. Our reaction to the cognitive overload of being a human brain in a swamp of AI drivel is, ironically enough, to surrender to the AI. This phenomenon is called cognitive surrender. According to researchers at Wharton, participants who were given logic problems with free access to an AI assistant followed it 80% of the time. The kicker is, the AI was secretly programmed to be wrong 50% of the time. Yet, participants’ confidence in their answers increased 12%, regardless of whether the AI was correct. They felt smarter. Yet they were wrong more often. So how do you insulate yourself? 1. Vote with your attention. Don’t use platforms you don’t like until they fix it. 2. Call out (kindly) when someone is sending you AI slop. Don’t take on the cognitive load of reviewing it. Push back. 3. USE YOUR BRAIN. For the love of god, write your own emails and slack messages. You’re not only getting dumber by outsourcing your whole brain, you’re also pissing me (and probably all of your coworkers) off. Stop it.
Experience & Education
-
Zeta Global
View Di’s full experience
See their title, tenure and more.
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Licenses & Certifications
Publications
-
Dynamics of epidemic models with asymptomatic infection and seasonal succession
See publicationIn this paper, we consider a compartmental SIRS epidemic model with asymptomatic infection and seasonal succession, which means the system is periodic and discontinuous. We defined its basic reproduction number (BRN), an important threshold parameter that decides if the disease will spread. We found that when RBN is less than or equal to one, the disease will die out in the end, otherwise it will prevail.
-
Modeling calcium waves in an anatomically accurate three-dimensional parotid acinar cell
Journal of Theoretical Biology
See publicationWe construct a model of calcium waves in a three-dimensional anatomically accurate parotid acinar cell, constructed from experimental data. Gradients of inositol trisphosphate receptor (IPR) density are imposed, with the IPR density being greater closer to the lumen, which has a branched structure, and inositol trisphosphate (IP3) is produced only at the basal membrane.
We show (1) that IP3 equilibrates so quickly across the cell that it can be assumed to be spatially homogeneous; (2)…We construct a model of calcium waves in a three-dimensional anatomically accurate parotid acinar cell, constructed from experimental data. Gradients of inositol trisphosphate receptor (IPR) density are imposed, with the IPR density being greater closer to the lumen, which has a branched structure, and inositol trisphosphate (IP3) is produced only at the basal membrane.
We show (1) that IP3 equilibrates so quickly across the cell that it can be assumed to be spatially homogeneous; (2) spatial separation of the sites of IP3 action and IP3 production does not preclude the formation of stable oscillatory Ca2+ waves. However, these waves are not waves in the mathematical sense of a traveling wave with fixed profile. They result instead from a time delay between the Ca2+ rise in the apical and basal regions; (3) the ryanodine receptors serve to reinforce the Ca2+ wave, but are not necessary for the wave to exist; (4) a spatially independent model is not sufficient to study saliva secretion, although a one-dimensional model might be sufficient. Our results here form the first stages of the construction of a multiscale and multicellular model of saliva secretion in an entire acinus.
Languages
-
English
Full professional proficiency
-
Japanese
Limited working proficiency
-
Chinese
Native or bilingual proficiency
Recommendations received
3 people have recommended Di
Join now to viewExplore more posts
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More