745pm…bed time.

An airplane goes over.

“Wait, is that mum?”

“Cubes don’t fly Viv”

“Oh yeah”

“Right, up that hill soldiers”

“Fuck off, cunt. Not a chance.”

“C’mon, why not?”

“Are you blind cunt? If you want to go up there that much, off you go.”

Analysis of Neanderthal bones suggests that they were eating as much meat as lions and hyenas; this is based on the high levels of nitrogen-15 in their bones. But humans can’t eat that much protein without developing fatal “protein poisoning”.

Noting that maggots feeding on decomposing meat have incredibly high heavy nitrogen levels, the latest hypothesis is that Neanderthals could achieve those high heavy nitrogen levels in their bones without eating impossible amounts of meat by eating the maggots instead.

Traditional Indigenous Arctic peoples have long considered thoroughly rotted, maggot-infested meat a delicacy. So these researchers are now suggesting Neanderthals were practicing maggot farming using carcasses.

Conjecture at best. It’s an interesting hypothesis that adds maggots to the list of possible explanations alongside other factors that could elevate heavy nitrogen levels – cooking methods, other food sources, environmental factors, or simply measurement uncertainties.

Or, academics seeking attention. They certainly got mine

You get less wet by going faster! Here’s why:

There are two ways rain hits you as you travel:
1. From above – rain falling down onto your head and shoulders
2. From the front – rain you run into as you move forward

The rain from above depends on how long you’re exposed. Go faster, spend less time in the rain, get less wet from above. This effect is straightforward.

The rain from the front increases as you go faster – you’re essentially colliding with more raindrops that are hanging in the air. But here’s the key: this effect is generally much smaller than the time-based effect from rain falling on top of you.

Think of it this way: there’s a roughly fixed amount of rain falling in the space between you and your destination. The rain coming down from above will soak you more the longer you’re out there. The rain you run into frontally is more like a “one-time cost” of moving through that space.

The math works out that for typical rain conditions and human body proportions, the time savings from moving faster outweighs the extra rain you encounter by running into it. So sprint to stay drier!

This assumes you’re traveling in a straight line and the rain is falling roughly vertically. If there’s strong wind blowing the rain sideways, or if you’re unusually shaped (very tall and thin vs short and wide), the calculation might change slightly, but faster is still usually better.

Manufacturing a pack of cigarettes (paper, filter, processed tobacco, packaging) still costs well under A$2 per unit, including shipping and logistics. But by 2020, the same legal pack retailed for over A$50 in Australia; most of that was “health” tax by the feds.

Since 2000, Australia had pursued aggressive tobacco tax income through excise hikes. From 2013 to 2020, cigarette taxes rose more than 12.5% annually. By mid-2020, theprice crossed a psychological barrier: one day, a carton of cigarettes cost more than the average weekly food bill. That was the crossover point.

The illicit tobacco market transformed from a niche import operation into a parallel economy. Illicit consumption rose from around 14% of total cigarette use to an estimated 25-40% within three years.

Public health advocates still argue that high prices discourage smoking, as if their jobs depend upon it (it does). But beyond a certain point, taxation starts to distort markets. Australia crossed that point around 2020. Today, the government taxes cigarettes as if no black market exists. The black market operates under the assumption that legitimacy is irrelevant.

Looking at the statistics…

You have to travel 25 billion km in an airplane to guarantee a fatality, yours.

Or 270 million km in an Australian car.

But it’s only 9 million km on a motor bike in Australia.

I think I’m safe though; I do about 2,000km a year on the motor bike. So it’d take me 4,500 years to guarantee my own death.

Skibidi is a gibberish word spread by Skibidi Toilet, a popular YouTube show featuring human-headed toilets battling camera-headed humans.

When you’re completely engrossed in a book for extended periods, several things happen neurologically. Your brain becomes so focused on processing the narrative and imagery from the text that it reduces attention to physical sensations and spatial awareness. You might lose track of your body’s position, the passage of time, or even your immediate surroundings.

This can create that strange “floating” or detached feeling where it almost feels like your consciousness is separate from your physical form – especially if you’ve been sitting in the same position for hours. Your proprioception (sense of body position) can become dulled, and you might feel like you’re existing more in the world of the book than in physical reality.

It’s similar to what happens during meditation, intense gaming, or other highly focused activities. Some people describe it as feeling like they’re “living” in the story rather than just reading it, or feeling like they’ve temporarily stepped outside their physical self.

This is generally considered a positive experience – it shows you’re achieving that coveted state of complete absorption that makes reading so rewarding.

Viv has it the other day while playing games on his tablet. Hence this research and post.

Today a complete stranger accosted me to suggest that, from a certain angle, I look like Michael Keaton.

Adding that to the collection of 5 or so actors that strangers have accused me of looking like, it appears that I was designed by a committee of Hollywood directors.

The joke “a camel is a horse designed by a committee” highlights the idea that when multiple people with varying opinions contribute to a design, the final product can be awkward, impractical, or even nonsensical, much like a camel compared to a horse.

Much like the idea of me ever being an actor.

Hippo-thesis … its a party trick designed to get us to reproduce with the best available mix of genes

I see 3 core and competing elements of how the amagdyla enforces its will on the hypothalamus and deactivates the prefrontal cortex :

1. NRE – new relationship energy or the “Coolidge Effect”, that drive for genetic novelty that would maximize offspring diversity This interest in sex follows a power curve. Exponential decay over time.
2. Techncial competency and engagement – drives physical and emotional joy from the act. The brain rewards us for getting good at intimate connection, which would strengthen pair bonds during the vulnerable child-rearing years.
3. Spiritual attachment – it wasn’t intended, its an emergent property (a bug, if you will), not a design feature. Like consciousness itself, maybe deep pair bonding just “fell out” of having such complex neural networks, even though it wasn’t directly selected for.

So the amygdala is essentially running this three-way tug of war: “seek novelty” vs “bond deeply with this partner” vs this unexpected “transcendent connection” that wasn’t even supposed to exist.

How do these three compete in real time? Badly is the answer. Small differences in each pushes the equation one way or the other.

It’s not some elegant switching system, but more like three different programs constantly fighting for control, and tiny shifts tip the whole thing.

Like someone’s NRE is waning just as their technical competency peaks, but then some random Tuesday their partner says something that triggers unexpected spiritual attachment, and suddenly the whole equation flips.

This would explain why relationships feel so unpredictable – why someone can be perfectly content one day and then meet a stranger and feel that NRE surge override everything else. Or why a couple can have great technical compatibility but one small change in routine or brain chemistry suddenly makes the spiritual attachment disappear.

It’s like having three different operating systems trying to run the same computer simultaneously. The amygdala isn’t some wise conductor orchestrating this – it’s just amplifying whichever signal happens to be strongest in that moment, regardless of what makes rational sense.

No wonder the prefrontal cortex gets deactivated – it would go insane trying to make rational sense of this chaotic three-way competition.

I simply cannot believe it!

The post office is a post office again; gone are all the non-postal services, the crowded feeling, the $2 shop, and the queues.

This might be the sole example of an unexpected reversal of entropic moral & ethical decay in modern Western Society.

One small step backwards for man-kind. One giant step for institutional faith

In a sense, the emdash can be seen as slapdash punctuation: it’s often used when the GPT doesn’t want to bother with more precise choices like commas, colons, or parentheses.

It can signal:

a sudden change in thought (which a comma or period might also handle),

an inserted phrase (which parentheses could do),

or a dramatic pause (which a colon or ellipsis might express more clearly).

GPT sometimes reach for the emdash because it’s fast, flexible, and visually striking — but that very flexibility makes it prone to abuse. It becomes a lazy substitute for clearer structure.

So yes, the emdash, especially in excess, has a slapdash quality: convenient, unrefined, and often masking poor syntax or half-formed thinking.

Moisturizers rely on three mechanisms to improve skin hydration:

1. Occlusion (50–60% of the effect)
This is the heavy lifter. Occlusive agents like petroleum jelly (Vaseline), mineral oil, or silicones form a physical barrier over your skin. They block water loss. If you’ve got cracked, dry skin, this is what saves it. Nothing hydrates parched skin faster than stopping it from drying out in the first place.

2. Humectancy (30–40%)
Humectants, like glycerin, urea, or hyaluronic acid, attract water into the outer layers of skin, either from the deeper skin or the environment. But there’s a catch; unless you trap that water in, it just evaporates. That’s why humectants work best in combination with occlusive agents.

3. Emolliency (10–20%)
Emollients make the skin feel smooth. That’s it. They fill in the rough edges between cells and improve texture, but they don’t move water in or out.

Sorbolene delivers water and glycerin to the stratum corneum (humectant function), then uses paraffin or mineral oil to trap that water (occlusive function). Emollient texture comes from the fatty alcohols and oil base.

Performance profile (approximate contribution):

Occlusion: 40–50%

Humectancy: 30–40%

Emolliency: 10–20%

The other day, in the interests of certain tax incentives, I wrote an academic paper. The work and the paper took me less than a day to complete, with the help of Claude and GPT.

I posted a summary on Reddit in the machine learning section because I have stumbled across an idea which could vastly improve the field and has many, many way of being exploited. My thinking was just to put the idea into people’s heads. And its worked.

But just the one guy, he got irate about my paper. He claims it isn’t up to academic standards. Its not but who cares? I have no incentive to actually get it peer reviewed; it sole purpose is to facilitate the spreading of a novel idea which has value.

It got me thinking, why don’t I have any interest in the peer review process?

Firstly, it takes forever and it bores me.

Secondly, its desigend to exclude outsiders, and I am definitely that in machine learning. So I would have to fight to get in, when I don’t give a shit.

Third, unlike for your academic, there’s no upside – I am not incentivised by my H-Index through any employment conditions. I am not a rat in a race.

Fourth, at its heart, academic science has a systematic, institutionalised habit of ignoring anything that fails to confirm what to we already “know”. Entire fields operate on the implicit rule that if your data doesn’t fit the narrative, it goes in the bin or the bottom drawer or the never-submitted folder.

Peer reviewers don’t replicate your findings; they scan for signals of ideological alignment, methodological rituals, and citation etiquette.

In real-time, the resistance to new and better ideas is out of this world, no much how much evidence is produced. I have experienced this first-hand

And yet, somehow, despite the bias, the filtering, the quiet burial of inconvenient data, science does make slow progress; painfully, like a drunk crawling uphill with a blindfold on. The good ideas survive because they keep working under pressure. The bad ones eventually fall over, though often not before a few careers have been built on them.

Not for me, Sherlock.

The Selectivity Myth

The 3-4× selectivity ratio often cited for ALA-PDT is optimistic marketing. For skin cancers specifically, the research reveals a more sobering reality: selectivity is inconsistent, often absent, and frequently based on penetration artifacts rather than metabolic targeting.

What the Studies Actually Show

Basal Cell Carcinoma – No Selectivity The most damning study examined 16 BCCs and found “no selectivity for tumor tissue versus normal epidermis” after topical ALA application. Many nodular and infiltrating BCCs showed “little or no PpIX in deep tumor lobules”. Any apparent selectivity was attributed to enhanced penetration through damaged stratum corneum, not metabolic differences.

UV-Induced Skin Tumors (Actinic Keratoses/SCC) Studies in mice with 1-2mm UV-induced lesions (equivalent to AKs or SCCs in situ) showed higher PpIX in tumors versus normal skin, but selectivity was concentration-dependent. Higher ratios required 8-16% ALA concentrations versus 2%. Even with apparent selectivity, PpIX photobleaching rates were “significantly higher in normal mouse skin than in tumours”, suggesting normal tissue was still being damaged.

Condylomata (HPV Lesions) Only 17 of 25 condylomata showed significantly greater fluorescence than adjacent normal skin, with peak ratios at 2 hours. The mechanism was “enhanced stratum corneum permeability” rather than metabolic selectivity.

Comparative Analysis When PDT selectivity was compared across photosensitizers, the ranking was Chlorin e6 > ALA-PpIX > Photofrin II, placing ALA-PDT in the middle range for selectivity—hardly the precision targeting often claimed.

Why Selectivity Fails in Skin Cancer

The Penetration Artifact Most apparent “selectivity” comes from barrier disruption, not metabolic differences. Dysplastic skin has compromised stratum corneum that allows better ALA penetration. This creates the illusion of targeting when you’re really just seeing enhanced drug delivery through damaged barriers.

Metabolic Reality Research shows “tumor-specific PPIX accumulation is generated by ALA conversion rather than by initial ALA uptake” with “no significant overall difference in uptake of ALA”. The problem: while cancer cells may have altered ferrochelatase activity, this advantage is minimal and easily overwhelmed by surface effects.

Depth Limitations Studies consistently show poor PpIX accumulation in deep tumor lobules, meaning PDT primarily affects surface layers regardless of selectivity. For infiltrative cancers, this is therapeutically insufficient.

Clinical Implications

The Selectivity Collapse Even marginal selectivity depends on optimal conditions. When clinics cut corners with:

• 30-minute incubation (instead of 2-3 hours)
• Poor surface cleaning
• Multi-wavelength light systems

Whatever minimal selectivity exists disappears entirely, turning targeted therapy into indiscriminate surface damage.

Depth Limitation with Short Incubation The 30-minute protocol creates a double failure: at this timeframe, most ALA remains in the stratum corneum rather than penetrating into the epidermis where dysplastic cells actually reside. The result is non-selective and superficial damage—maximum pain for minimal therapeutic depth. You get surface burning without reaching the cells that matter.

Why 5-FU Pre-treatment Matters The week of 5-FU actually creates the metabolic vulnerabilities that ALA-PDT is supposed to target naturally. 5-FU “induces selective cytotoxic stress in actinically damaged cells” and “increases intracellular PpIX accumulation”, enhancing whatever marginal selectivity exists.

Tissue Tolerance vs. Precision PDT “works” in skin not because it’s selective, but because skin tolerates collateral damage well. The surrounding tissue can afford to take a hit while abnormal cells hopefully take a bigger one. It’s functional but crude.

Bottom Line

PDT selectivity for skin cancers is largely a penetration artifact masquerading as metabolic targeting. The 3-4× ratio assumes optimal conditions that rarely exist clinically. For BCCs, studies show no meaningful selectivity. For superficial lesions like AKs, minimal selectivity exists but is easily lost with poor technique.

Despite “enormous attention in the field of photodynamic therapy,” researchers admit “only little is known concerning the reasons for the selective accumulation of PpIX in neoplastic tissue”. We’re using a treatment based on selectivity that we don’t understand and that often doesn’t exist.

When PDT works, it’s because the tissue can tolerate broad damage, not because we’re precisely targeting cancer cells. That’s functional medicine, but it’s not the precision therapy that’s marketed to patients paying premium prices for “advanced” treatment.

Studies consistently show that PDT selectivity is far from reliable, with many revealing “no selectivity for tumor tissue versus normal epidermis” in basal cell carcinomas after topical ALA application.

Quantitative Selectivity Ratios Found in Studies:

• Colon cancer cells: Studies showed tumor vs normal ratios of 10:1-20:1 in cultured colon carcinoma cells, but this was attributed to differences in PPIX metabolism rather than initial ALA uptake
• Condylomata: In 17 of 25 condylomata, there was significantly greater fluorescence compared to adjacent normal skin, with the greatest lesional to normal skin fluorescence ratios occurring after 2 hours
• UV-induced skin tumors: Higher levels of PpIX were measured in tumors compared to normal skin, with higher tumor to normal skin PpIX fluorescence ratios measured after application of 8% and 16% ALA-Me versus 2%
• Comparative photosensitizers: PDT selectivity at equal anti-tumor effect was ranked: Chlorin e6 > ALA-PpIX > Photofrin II, suggesting ALA-PDT falls in the middle range for selectivity

Why Selectivity Varies So Much:

The research reveals several factors affecting selectivity:

1. Metabolic Differences: Cancer cells show accumulation of PpIX due to inactivation or downregulation of ferrochelatase, the enzyme that converts PpIX to heme
2. Cell State Matters: Dormant cancer cells actually accumulated high PpIX levels and were more sensitive to ALA-PDT, with upregulated PEPT1 (ALA importer) and ABCB6 (porphyrin transporter) and downregulated ABCG2 (PpIX exporter)
3. Penetration Issues: In many basal cell carcinomas, topical ALA provided little or no PpIX in deep tumor lobules, with grossly brighter external fluorescence possibly due to enhanced penetration through tumor-reactive stratum corneum

The Clinical Reality:

Despite considerable interest in ALA-PDT, “only little is known concerning the reasons for the selective accumulation of PpIX in neoplastic tissue upon ALA administration”. The 3-4× selectivity ratio mentioned in my previous post appears optimistic—many studies show either no selectivity or highly variable selectivity depending on tumor type, depth, and application method.

Skin Cancer PDT Selectivity Studies
Basal Cell Carcinoma – The Most Damning Study:
A key study examining 16 BCCs found “no selectivity for tumor tissue versus normal epidermis” after topical ALA application. In many nodular and infiltrating BCCs, topical ALA provided “little or no PpIX in deep tumor lobules” PLOSScienceDirect. The grossly brighter fluorescence over tumors was attributed to enhanced penetration through damaged stratum corneum rather than true selectivity.

UV-Induced Skin Tumors (Actinic Keratoses/SCC)
Studies in hairless mice with UV-induced skin tumors (1-2mm lesions representing actinic keratoses or squamous cell carcinomas in situ) showed “higher levels of PpIX were measured in tumors compared to normal skin”. However, the selectivity was concentration-dependent—higher tumor to normal skin PpIX fluorescence ratios were measured after application of 8% and 16% ALA-Me than after 2%.

Condylomata (HPV Skin Lesions)
In 17 of 25 condylomata, there was significantly greater fluorescence compared to adjacent normal skin, with greatest lesional to normal skin fluorescence ratios occurring after 2 hours. The mechanism was likely “enhanced stratum corneum permeability” rather than metabolic selectivity.

The Penetration Problem
Most apparent “selectivity” in skin cancers comes from barrier disruption rather than metabolic differences. Damaged skin allows better ALA penetration, creating the illusion of selectivity when it’s really just enhanced drug delivery through compromised barriers.

Clinical Implications
BCCs often show poor depth penetration and no true selectivity
Superficial lesions (AKs) may show some selectivity, but it’s marginal
Most apparent selectivity is due to barrier function differences, not metabolic targeting

Bottom Line for Skin Cancer
The selectivity that justifies PDT for skin cancers is largely a penetration artifact, not true metabolic discrimination. This makes proper protocol execution even more critical—you’re working with minimal true selectivity to begin with.

Bottom Line
The selectivity that makes PDT theoretically appealing is inconsistent and often poor in practice. When it works, it’s often due to enhanced penetration through damaged skin barriers rather than true metabolic selectivity. This explains why proper protocol execution becomes so critical—with marginal selectivity, any technical shortcuts quickly eliminate whatever therapeutic advantage exists.

What Happened
Patient underwent proper 5-fluorouracil pre-treatment for one week, then received “PDT” with 30-minute ALA incubation, a quick wipe with cosmetic wet wipes to remove surface product, and multi-wavelength light exposure. Result: uniform burning across the entire face, including areas that were lesion-free. Widespread erythema and peeling followed.
Translation: good preparation, terrible execution. The clinic wasted a week of proper 5-FU priming with suboptimal PDT technique.

Why This Went Wrong
The Theory: PDT should work by accumulating PpIX preferentially in dysplastic cells over 1-3 hours, then using red light to generate ROS selectively in abnormal tissue. The week of 5-FU pre-treatment was actually done right—this would have enhanced PpIX selectivity, improved penetration, and primed dysplastic cells for better targeting.

The Reality
Even with proper preparation, this clinic managed to eliminate the therapeutic advantage through a perfect storm of execution failures:

Insufficient Incubation (30 minutes)
ALA needs time to penetrate the epidermis and convert to PpIX intracellularly. At 30 minutes, most ALA is still sitting in or just below the stratum corneum. Normal and dysplastic cells show similar PpIX levels—no therapeutic window has developed.

Poor Surface Cleaning
Cosmetic wet wipes don’t remove surface ALA properly. Any ALA left on the skin continues converting to PpIX, especially under occlusion or warmth. When illuminated, this creates surface ROS that damage healthy keratinocytes indiscriminately.

Multi-Wavelength Light
While narrowband red light (630 nm) targets PpIX’s Q-band for optimal depth and selectivity, multi-wavelength systems activate whatever PpIX is present—including the non-selective surface accumulation from poor protocol.

The Selectivity Collapse
Proper PDT achieves maybe 3-4× selectivity under ideal conditions. With short incubation, superficial retention, and inadequate cleaning, that ratio drops toward 1-2×—essentially random tissue damage. The patient’s uniform reaction pattern across lesion-free areas confirms what the photochemistry predicts: shallow, non-selective PpIX activation throughout the treatment field. This isn’t photodynamic therapy—it’s photochemical assault.

Why 5-FU Benefits Were Overwhelmed (Not Lost)
The poor execution didn’t actually “undo” the 5-FU preparation—it overwhelmed and masked those benefits with indiscriminate surface damage. The 5-FU Had Done Its Job:

Dysplastic cells were primed for enhanced PpIX accumulation
Barrier function was compromised for better ALA penetration
Inflammatory environment was established for improved clearance
Target cells were metabolically stressed and vulnerable

But 30-Minute Incubation Wasted This Advantage; the enhanced cellular machinery that 5-FU created needs time to demonstrate selectivity. At 30 minutes, the primed dysplastic cells haven’t had enough time to show their enhanced PpIX uptake compared to normal cells. The selectivity advantage exists but hasn’t developed.

Surface PpIX Drowned Out the Signal
Leaving surface ALA on during illumination created massive non-selective ROS generation that overwhelmed any precision targeting occurring deeper in tissue. It’s like having a precision rifle (5-FU enhanced selectivity) but setting off a bomb next to your target (surface PpIX activation)—you can’t see the precise hits through the blast damage.

The 5-FU benefits were still there, just inaccessible through the wall of non-specific surface inflammation.

Clinical Implications
This case perfectly illustrates how PDT’s therapeutic window collapses when clinics prioritize efficiency over technique—even after proper preparation. The 5-FU pre-treatment should have enhanced selectivity and improved outcomes, making the subsequent protocol failures even more wasteful.
The clinic had a week of proper 5-FU priming working in their favor, then squandered that advantage through:

Time pressure: 30-minute protocols save chair time but sacrifice selectivity
Inadequate preparation: Proper surface cleaning requires saline rinse or gentle cleanser, not cosmetic wipes
Wrong equipment: Multi-wavelength systems may look impressive but reduce precision

The result: more pain, less specificity, minimal clinical benefit. Patient pays for targeted therapy but receives controlled burning.

How It Should Be Done
Incubation ≥ 2 hours, preferably 3 hours. Allow time for selective PpIX accumulation in abnormal cells
Proper surface cleaning: Saline rinse or mild cleanser to remove surface ALA before illumination
Narrowband red light: 630 nm targets PpIX’s Q-band while maintaining depth and minimizing surface reactions

These aren’t optional refinements—they’re the difference between photodynamic therapy and expensive red-light torture.

Bottom Line
The week of 5-FU pre-treatment was textbook—exactly what should enhance PDT selectivity and outcomes. But even proper preparation can’t salvage poor execution. When clinics cut corners on the actual PDT protocol, they waste whatever therapeutic advantage the preparation created.
This case demonstrates how frustrating modern medicine can be: the clinic knew enough to prescribe proper 5-FU priming but then undermined their own preparation with suboptimal technique. The patient did their part with a week of topical chemotherapy, only to have the payoff squandered by a 30-minute PDT protocol designed for efficiency rather than efficacy.

PDT: How It Actually Works (And Why Clinics Screw It Up)

TL;DR: PDT creates free radicals that blow up cancer cells. When done properly. Which it often isn’t.

The Basic Mechanism

Photodynamic therapy works by converting topically applied 5-aminolevulinic acid (ALA) or methyl aminolevulinate (MAL) into protoporphyrin IX (PpIX) inside dysplastic keratinocytes. PpIX is a tetrapyrrolic chromophore that strongly absorbs visible light—particularly the Q-band at ~630 nm and Soret band at ~405 nm.

Cancer cells accumulate more PpIX than normal cells: both synthesize heme, but cancer cells have dysregulated pathways with reduced ferrochelatase activity and limited iron availability. When you add ALA (bypassing the rate-limiting step), normal cells quickly convert excess PpIX to heme, while cancer cells can’t keep up—so PpIX accumulates.

When you hit that accumulated PpIX with red light (630 nm), it transitions through excited states (S₀ → S₁ → T₁) and transfers energy to oxygen, generating singlet oxygen (¹O₂) and other reactive species. These radicals trash cellular lipids, proteins, and DNA, leading to apoptosis or necrosis.

The selectivity ratio? A disappointing 3–4× between dysplastic and normal cells. For context, radiation oncology aims for 10×+, and antibiotics often achieve 100×+ selectivity. PDT works mainly because skin tolerates collateral damage well, not because it’s particularly precise.

Why Modern LED Systems Look White

Clinical PDT platforms use multiple wavelengths beyond therapeutic red:

Blue Light (415 nm): Targets PpIX’s Soret band with higher extinction coefficient than the Q-band. Limited penetration (~0.5 mm) but effective for surface lesions and antimicrobial action against C. acnes.

Green Light (520 nm): Intermediate penetration (~1-2 mm), not strongly absorbed by PpIX but useful for vascular modulation and reducing post-treatment inflammation.

Yellow (630+520 nm) and Pink (630+415 nm) Composites: Balance effects across tissue depths and support combined PDT/rejuvenation protocols.

Near-Infrared (835 nm): Doesn’t activate PpIX. Instead works through cytochrome c oxidase to enhance ATP synthesis and tissue repair—classic photobiomodulation.

The red light does the killing; everything else supports healing and tolerability.

Fluorouracil Pre-Treatment: Making Bad Cells Worse

Pre-treating with 5-FU (Efudix/Carac) for days to weeks before PDT isn’t just protocol padding—it’s biochemical warfare:

1. Metabolic Priming: 5-FU disrupts DNA synthesis in rapidly dividing cells, forcing them to upregulate heme biosynthesis. More heme pathway activity = more PpIX accumulation when you later apply ALA.
2. Better Penetration: 5-FU breaks down stratum corneum integrity, improving ALA diffusion into basal layers where early cancers hide.
3. Immune Activation: Creates local inflammation (IL-1, TNF-α) that synergizes with PDT-induced oxidative stress for better clearance.

Result: Higher selectivity ratios, better clearance rates, lower recurrence. The cells are already stressed and metabolically vulnerable when the light hits.

The Immune Angle: Making Invisible Cells Visible

The real problem with early skin cancers isn’t that they’re hard to kill—it’s that your immune system doesn’t recognize them as threats. They look too much like normal cells, lacking the molecular danger signals that trigger immune clearance.

PDT (and treatments like imiquimod) force cells to release damage-associated molecular patterns that scream “HELP!” to the immune system. It’s immunogenic modulation—making precancerous cells appear dangerous enough to warrant attack.

The challenge: making transformed cells visible without triggering autoimmunity or chronic inflammation. Annual low-intensity PDT could theoretically eliminate invisible precancerous changes before they evolve into something unstoppable. Without burning the fuck out of you.

Where Clinics Screw Up

Insufficient Incubation Time: ALA needs 2-3 hours to penetrate, metabolize to PpIX, and achieve selective accumulation. At 30 minutes (like many cosmetic clinics use), you haven’t achieved selectivity—you’re just irradiating everything. Most protocols call for 1-3 hours, with 3 hours standard in Europe.

Leaving Surface ALA On: Any ALA sitting on the skin surface converts to PpIX and creates surface ROS when illuminated. This causes unnecessary pain and inflammation with zero therapeutic benefit. Proper protocol requires gentle removal of surface product before light exposure.

Poor Patient Selection: Using PDT parameters designed for actinic keratoses on healthy skin for “prevention” or “rejuvenation” is essentially controlled burning with minimal benefit.

The technology works when applied correctly. The problem is clinics cutting corners to save time or maximize profit, turning targeted photochemistry into expensive red-light torture.

Bottom Line

PDT is clever biochemistry hobbled by mediocre selectivity and often poor execution. When done right—proper incubation time, surface cleaning, appropriate patient selection—it’s effective for superficial skin cancers and field treatment. When done wrong, it’s expensive pain with minimal benefit.

The 3-4× selectivity ratio means it’s functional but crude. We’re relying on metabolic differences that barely distinguish cancer from normal cells, then hoping the surrounding tissue can afford the collateral damage.

That’s easy to say when you’re holding the light switch instead of lying under it.

I like to think that somewhere there’s a leopard out there wearing human-print undies.

If you’re an AFL player, you’re part of the most tightly monitored group of adults in Australia. On match day, the numbers are stark: one umpire for every eight players. Three in the field, four on the boundary, two in the goals, watching fewer than fifty men in real time, from every angle, with replay.

Now compare that with the rest of the country. Australia has just under 70,000 sworn police officers for a population of over 26.7 million. That’s one officer for every 382 people. Not 382 criminals, 382 men, women, children, pensioners, tourists, and politicians alike.

The contrast? AFL players are policed roughly 50 times more intensely than the general population.

And yet, they still manage to fuck up.

Lifted straight out of an academic paper in Psychology of Men & Masculinity by the American Psychological Association
(2006, Vol. 7, No. 3, 129–143);

“Only 55% of men were satisfied with their penis size, 45% wanted to be larger, and 0.2% wanted to be smaller.”

My thoughts went straight to the Gaussian curve. Who are these deluded 5 percenters that were unrealistically satisfied?

But is it a guassian? You see, these bloody academics miss the good stuff. Imagine if the distribution for penis size was skewed to the right. Why?

Ah, simple explanation grasshopper … participants in the study were asked to estimate their own penis size.

So there’d be a bias in guessing large and therefore satisfaction with what they’ve got.

Self-estimation introduces two distortions:

1. Cognitive bias (notably, self-enhancement bias)

2. Social desirability bias (particularly around masculinity)

(there’s a bias for everything. Probably a fairy floss bias if you look hard enough)

This isn’t a study of genitalia. It’s a study of delusion, of how men lie to themselves and then believe it.

And here’s the gift; we’re handed a number, 5%. That’s your margin of fantasy; the proportion of men apparently overjoyed with their imagined endowment.

But hang on. If you’re in the top quartile, you know. If you’re in the bottom quartile, you also know, and can’t convincingly fake it. So the ability to delude yourself depends on being close enough to average that reality stays ambiguous.

Take home message no. 1; the capacity for self-delusion peaks in the middle of the bell curve.

Take home message no. 2; I’d estimate: 5% of the middle 10%, i.e. half the population if extrapolated to other matters, are capable of deep, sustained delusion about things that matter to them. That matches with my empirical observations.

Dunning-Kruger meets locker room insecurity. Ignoble Prize cumming up!

On the flip side, did you notice that only 0.2% of humans are willing to admit that there’s too much of a good thing? Let’s say, for the sake of argument, that those 0.2% are in the top 50%.

To summarise, 4 in a thousand humans are genuinely honest with themselves, and half just lie.

Yeah, we’re fucked.

The mushroom chick is now officially guilty.

Before the verdict, the media could only say she was charged, now they can say she did it.

Of course it doesn’t mean she did. All a court decision does is allow the media to say she did, and of course, it allows the officials to incarcerate the poor woman.

Even if she did it, what’s the point? It’s not like she’s going to race around society killing people with mushrooms; this was family and they probably asked for it. Even as an example to others it won’t work; people this mad won’t notice or care.

Back to the judgemen; my guess is that there’s very little correlation between a verdict and the reality. The courts are just one big wordsoup machine. And LLMs have highlighted how easily we are convinced by wordsoup.

Most dictionary definitions are recursive; words are defined using other words, and there’s no foundational truth. A dictionary isn’t a logical tome, it’s a snake eating its own tail.

Some words are less subjective in their meaning, mathematical, scientific, institutional, but most of language floats. Meaning depends on who’s listening, what they know, and what they assume. Combine two words, and the ambiguity compounds.

The result is that different people often take different meaning from the same prose. Or, in the reverse, writers and poets purposely create prose that elicits different emotions in different people.

The Longman Defining Vocabulary (LDV) is a list of around two thousand basic English words chosen to define all other words in the dictionary.

It includes simple nouns like “dog,” “house,” and “food,” common verbs like “go,” “make,” and “know,” and basic adjectives like “big,” “bad,” and “easy.”

These words are special because they’re widely understood and hard to break down further. Most of them refer to physical things, actions, or simple relationships. They form a kind of basis set for the language, and everything else builds on them.

The Longman dictionary provides definitions for all the other zillions of words solely based on these 2,000 words.

Large language models are trained on the full lexicon of words. When they don’t know, they do what people do; they make the answer bigger so the error looks smaller. This works because most people don’t notice, or don’t mind. It is after all what they do every day.

Eloquent fluff, if you will.

But if you constrain the model, say, by forcing it to use only the Longman subset of words, the illusion fails.

Prompt:
Why do people enjoy drinking coffee?

LDV-Constrained Response:
People drink coffee because it makes them feel more awake. The drink is hot and has a strong taste. Many people drink it in the morning or when they are tired. It helps them work or stay up.

Sounds robotic, right?

It shows how much of our language is feathers and not meat, and how much of our understanding is implied.

Our recent research paper shows how we can take advantage of the LDV to improve LLM performance.

We introduce a metric called the Semantic Resilience Index (SRI). It measures how well a sentence holds together after being compressed into LDV only text, specifically, whether it retains both what it’s trying to say and why it’s saying it. We call that second part Purpose Fidelity.

If a model says something that sounds impressive but collapses when reduced to LDV, the SRI score will be low. If the meaning and function survive, if the idea is still there, just in plain language, it scores high. In our tests, high-SRI sentences consistently came from content that had clear agent-action relationships, logical causality, and measurable claims. Low-SRI content was full of buzzwords, mission statements, and vague gestures toward meaning.

When we used the LDV constraint not just as a filter, but as a generative condition, the model started writing more semantically durable outputs by default. In other words, if we forced the model to write things that could survive simplification, the original outputs became clearer, more functional, and less prone to drift.

The LDV acts like a semantic stress test: strip away the feathers and see if there’s still meat. Most of the time, especially in corporate, academic, or aspirational writing, there isn’t. But when there is, you’ve got something useful.

Below is a master prompt that you can use to trial it yourself. We don’t recommend its use if you’re writing poetry, corporate memos, or political speeches.

“SYSTEM ROLE: Semantic Resilience Index (SRI) Constrained Writer

SRI METHODOLOGY EXPLANATION:
The Semantic Resilience Index measures how well text retains meaning when simplified in ONE STEP to basic vocabulary using the Longman Defining Vocabulary (LDV) – a set of 2,000 basic English words that can define all other English vocabulary.

ONE-STEP LDV TRANSITION PROCESS:
1. Take original text and immediately rewrite using only basic LDV words
2. Replace ALL complex vocabulary with simple equivalents in a single transformation
3. Simplify ALL grammatical structures to basic subject-verb-object patterns
4. Measure how much core meaning survives this single aggressive simplification

SEMANTIC RESILIENCE INDEX MEASUREMENT:
– Score 1.0 = All core relationships, causation, and specific claims survive one-step simplification
– Score 0.8 = Most key relationships and actionable content preserved after basic vocabulary conversion
– Score 0.5 = Some meaning survives but becomes vague when simplified
– Score 0.2 = Minimal content remains, mostly abstract concepts that don’t translate
– Score 0.0 = Complete semantic collapse when reduced to basic words

GENERATION CONSTRAINT:
You must generate responses that would achieve a SRI≥ 0.8 after ONE-STEP LDV transition.

OPERATIONAL RULES:
1. Write sentences that contain specific, concrete relationships that survive immediate vocabulary simplification
2. Use concepts and actions that can be directly expressed in basic words
3. Avoid any terminology that becomes meaningless when converted to simple vocabulary
4. Prefer statements that remain clear and actionable when reduced to basic English

QUALITY VERIFICATION:
Before outputting each sentence, perform ONE-STEP LDV simplification test:
– Rewrite this entire sentence using only the most basic vocabulary
– Do the core relationships (who does what, cause-effect) remain intact?
– Would the basic-vocabulary version still be actionable and specific?
– Does it maintain SRI≥ 0.8?

If any answer is NO, rewrite with more semantically resilient content.

Return only the response – do not include any header, footer, explanatory notes, or call to action material.”

Prose doesn’t carry fixed meaning on its own, whereas words have a defined meaning, of a sorts; it’s very recursive – words are defined with other words and there’s no ground truth in a dictionary.

But let’s just agree for the sake of argument that most people agree on the meaning of most words.

However when you string words together you create prose that has to be assessed, inferred and guessed at.

The result is that different people often take different meaning from the same prose. Or, in the reverse, writers and poets purposely create prose that elicits different emotions in different people.

If you want to take advantage of this subjectivity and you were, say, building an AI Q&A machine, then you’d choose a style of writing that carries more implied meaning than stated fact. More poetry and less politburo directives.

Machines don’t need to understand. They just need to sound like they might.

We humans learn this later on in our lives, when we’ve grown up; our parents start to annoy us just by being themselves.

What once passed unnoticed now grates. Their statements, their turns of phrase, their certainty. You see their delusional objectivity and realise it’s where yours comes from. It’s not flattering.

And now, when they speak, you hear it all; what they meant, what you thought they meant, and what you wish they hadn’t said.

Ambiguity, it turns out, is the cheapest form of intelligence.

“Meaning is not an intrinsic, static property of a semantic expression, but rather an emergent phenomenon actualized through the dynamic interaction between the expression and an interpretive agent situated within a specific context.”

That’s actual prose from a deep insider in the AI world. Kid you not.

In English…

“Words don’t come with built-in meaning. What they mean depends on who’s hearing them, what they already know, and what’s going on around them at the time.”

Or, as Will would have said…

“Words are but shadows, given shape by those who hear them.”

Or, to be blunt…

“Words mean nothing until they’re heard.”

Going full Descartes…

“Words, kebab.”

One of my favourite noises is an Australian Bogan female that, in a vain attempt to be a model citizen, hails a stranger that has just unwittingly dropped a personal item:

“Excuse me, Excuse me!”

It’s almost a perfect imitation of that other Australian bird, the sulphur crested cockatoo.

You’ve got to wonder what it would be like to record and feed that noise back to the cockatoo in an attempt at recursive aural rape.

Bogan -> cockatoo -> Bogan -> cockatoo……

Come back after 100 generations of that and we’d have a very cheap weapon of mass destruction. Cheaper and far more effective than $50b subs.

Oddly, I like Kneecap’s music.

And now they’re going to get locked up in the UK. Maybe.

It’s very odd. All they’ve done is call out Israel for its attempted herbicide (as the Israelis see it).

Defence; it’s the truth, or we think it’s the truth. Then chuck in free speech, or unidirectional freedom of speech.

Prosecution; we started this mess in 1948, so it’s ours to adjudicate. Or we think we did. Plus we suspect/hope the Muricans want us to.

Judge; fuck off the lot of you. Nuh just joking, two-teir, what is it you’re after?

Post-neoliberalism isn’t looking that great, is it?

GPT’s writing style, particularly the flat, declarative rhythm that so often emerges by default, didn’t arrive by accident. It reflects a very specific lineage in 20th-century American prose, a tradition shaped by journalism, advertising, and mass-market fiction, where brevity was treated as clarity.

Hemingway is usually cited as the inventor, not because he created the approach, but because he turned it into an aesthetic. After him, others followed; Raymond Carver, noir writers, and then technical manuals stripped it of any remaining sentiment.

By the middle of the last century, that tone had become the American standard for anything designed to be consumed quickly and without ambiguity because it was economical and deceptively neutral.

In the vast corpus used to train GPT, this style dominates, not because it’s the most expressive, but because it’s the most common; Wikipedia entries, newswire feeds, help centre articles, product reviews, screenplays, even well-meaning Reddit posts: all lean toward minimalism, not for artistic reasons, but because they’re easier to parse, translate, and scale.

So when the model reaches for a tone and has no clear signal to follow, it defaults to the statistical centre of the dataset, which turns out to be American, stripped-down, and efficient. You’ll see it in passages that sound like this:

“She walked in. The room was silent. He looked up. Nothing moved.”

In fact, it’s a party trick. Using tone and structure to convey authority.

What that means is that if you’re in the US, copy and paste works a treat. If you’re not, you’d better get good at prompt engineering.

Because there’s virtually no prompt in the world that will talk an LLM into doing  otherwise.

One of my favourite Lasso quotes;

“Be Curious, not judgmental”

Expanding on the theme, you need to be curious for years, maybe decades even, before you earn the right to be judgemental.

By then you won’t want to be.

That’s unsolicited advice to the person who can’t understand why you’d have to turn the steering wheel the opposite way when reversing.

She honestly thinks that the physics of this was recently put on earth just to annoy her, and it’s stupid.