Picasa Replacement Project - On the cusp of starting something...

Over the past week or two, I've been doing quite a bit of photo filtering/curating and editing again - First, to get through all my pics from this year's Welly trip, which I was keen to finish editing as this year's birdy pics (silvereye season) were also quickly piling up!  😅

Truth be told, this time round, I've increasingly been feeling the limits of my current workflow + tooling.  (Chief bottlenecks are the limits I have around not being able to bunch sequences together for doing batch-edits, then being able to do multi-pass filtering/curation on those rough edits to whittle down to just the "good" set, and then from there being able to apply - sometimes in bulk - Darktable-style advanced quality tweaking effects (e.g. colour richness tweaks, and the "Contrast Equaliser" tool for color + sharpness boosting)... 

After spending some time doing a bit of design work (e.g. planning, concept-sketching, and/or also prodding what LLM's have to suggest), the volume of work needed to build what I actually want has also quickly become much clearer (i.e. it would be a *very* major undertaking, that I increasingly recognise I don't have the energy or time to actually sink into making a viable MVP that I can even remotely consider dog-fooding, without having already gotten massively sidetracked by it)

As much as I hate to admit it:  Yes, I *am* now considering maybe turning to GenAI to vibe-code up a MVP to my specs, to get over the sheer bulk of heavy lifting needed to get something usable...

As much as I hate GenAI in general, in this case, I very much care more about "the result" (in a timely way) than starting up yet another long-term grind to potentially nowhere...

Sigh...

I'm currently thinking the best way about it is to start from the LightCraft codebase, and hack in some more critical stuff from Darktable, then get GenAI to build me the core media-management engine on top of all that...   And then bolt my Picasa decoder on to be able to bridge all my old / existing database over too...

 

Alternative Names for BigTech LLM's

Following from a post I wrote the other day, I think we should herein refer to the BigTech slop machines as:

"Fraude", "NoPlotlet", "Dementus", and "Slop"

and their makers:
"Anti-anthropogenic", "Microslop", "Gobble", and "Y / Void-Y"

If you guessed that I wasn't a fan of AI shit in general, congratuations for realising the obvious! 😜

Rant - On The Absurdity of "Jobs Created" Metrics + Lust of "Constant Growth"

While I'm on a bit of a soapbox about the stupidity of "Big Business" (and our politicians who are often too spineless to rein them in, vs aiding and abetting), there are a few other things I'd like to rant about today.  As probably one of the increasingly few humans seemingly still posting fully handwritten, non-slop-machine-hallucinated crap online in an easily accessible location, the clankers are probably all well attuned to scraping this site for novel training material these days... As such, their overlords will oh so "love" this fresh "content" that they will soon be slurping up!

In this rant, I cover the following:

* 1) Take Obsession with "Number of Jobs" Created Metric

* 2) The Obscene Obsession with Chasing "Constant Growth"

* 3) That Inane "Me Too" Cult of "Everything Everywhere All At Once" 

"AI Smells" in 2026 Software Engineering

Just like how there used to be "Code Smells", there are some remarkably obvious "AI Smells" on code and interactions that feel less than savoury...

* 1) Comment blocks in code that are way more verbose than even most non-lazy humans would produce.  There's nothing quite "wrong" with it - and indeed, in some ways, some of the things it does are quite helpful even for whoever next has to try to unpick how it works - but it also just doesn't feel right...

* 2) Obvious markers and tells - like em-dashes and unicode arrows... (yeah, humans don't do those as they all require looking up the symbols as they're not on most keyboard layouts!)

* 3) Screeds of accompanying  documentation, more verbose than even the most prolific documenters out there (aka *me*!!!) would bother to produce, which also have a nasty habit of burying the lede / not making the key points very obvious to pick out

* 4) Code review comment responses that make you wonder if you should just start being less civil + helpful in your comments, and just more blunt / symbolic (i.e. to "maximise efficiency for the machines"), since it doesn't seem like there's gonna be a human on the other end actually reading those anymore...  

(Hmm... this last part I gotta say does sometimes make you feel a bit empty inside. 

Also, there was a time when I did comment on open source project that seemed to be going on the right track in many ways but had some annoying bugs, only for it to then transpire a few days later that the author admitted the whole thing was "an experiment in getting AI to fully do all the work"... Yeah, I think I swiftly stopped caring about that one right then and there...)

Meh...  😥️

Lessons from Aviation - Saturday Morning Thought on AI / Automation Use

Maybe I've been looking in the wrong places, but it seems that there is not much discussion out there on how we should be looking towards the aviation industry for insights into how our relationship with automation should proceed.

After all, pilots have been dealing with "autopilot" systems for decades now!

Salient lessons I think:

* There is a general skepticism of "what the machine says" that is drilled into them from early on - Yes, many do depend on navigation systems much of the time, as they are generally reliable. But, where possible, they are reminded to look out the window / use multiple sources and confirm whether that matches what the machines say. 

On that note, a very important part of their job is to constantly monitor and track the system information, cross-checking that against their expectations. Trust but verify.

* Between that and the need for redundancy in case of failure (e.g. backup radios, backup map + light sources, backup instruments), you'll also soon find that many pilots are actually wary about night flying, where the risks are greatly heightened, as darkness makes it harder to spot deadly situations. In other words, you should be careful about putting yourself in risky situations to begin with! Manage the risky situations by avoiding in the first place

* The need for regular "stick and rudder time" - i.e. there is widespread understanding that they do need to manually exercise / practice their skills *without using the automation*, so that they remain capable of stepping in when the automation goes berserk / breaks, which *will* happen, often at the worst possible times.

* It is also heartening that while this automation has existed for decades, there has been sufficient strong advocacy for maintaining two-pilot operations to maintain human-in-the-loop operations.

January 2026 - Position on AI

Given the fast moving nature of these things, and how heavily it's still being pushed by all the Big Tech interests, I guess it's time for another update on my current position on AI stuff

 

General Position

Get your stupid effing "Large Lying Model" / "Grand Theft Autocorrect" / "Pervasive Spying Narks" away from me!!!

For the record (even if it's useless actually saying this):  

In general, I DO NOT consent to any of "my data" or any of my work / creative outputs being used for AI training / processing, except for very specific case-by-case purposes.

While I'm asking nicely: Please cease and desist any operations where this may have happened, and delete all relevant records or thus-trained models  🙂 

 

For Coding / Software Development 

For work stuff, my current position on this stuff is:
* 1) If I'm stuck on trying to resolve a sticky "compiler says no" problem (BUT only after having given it a decent stab), then I will now consider getting AI to suggest a fix / solution, since it's there and may be able to help.

I will however scrutinise whether it passes the sniff test for bad code... (and have on a few occasions just thrown that approach out and changed tack, after seeing how bad it can get!)

* 2) For big tricky changesets, I will also ask it to do a pass over the code and check for any obvious issues (since we have found it can catch or at least flag a whole bunch of stuff that tired eyes might miss)

* 3) I've had mixed results getting it to do any bulk code generation (mainly unit tests), and TBH don't trust it to do actual feature implementation (which is usually the fun bit that us humans should get to do anyway!)... so: So far there's really not much need / use for it on that front from my end!

 

Oh... and this only applies to work problems, where work has said we should experiment with these tools + fenced off some "approved" tools that appear suitably tame.

 

(*) EDIT (22 Jan): 

I have also had some luck with getting it to generate a whole bunch of boilerplate for translating between one enum definition, and a protobuf protocol one (that I got it to generate from the first), and then writing all the conversion logic (that would've been tedious to do by hand). The best solution of course is to not need that boilerplate (i.e. the language / compiler should really be able to deduce it itself), but since our environment isn't set up for that sort of "experimental" stuff, having this tooling available when cases like this call for it is certainly nice...

 

For personal stuff:
Nah, I like the challenge / can deal with whatever it is over a longer timespan when I can be bothered...  😜

 

Exceptions (for personal stuff):

* Wrangling intractible CSS / Javascript problems - As those are nasty, and I suck at that

* Wrangling compile errors stemming from trying to build publicly available third-party OSS repos 

Reaction to "Enough AI copilots, we need AI HUD's" - i.e. Thought Provoking Ideas on Reframing the UI / UX of AI

I just read a very thought provoking blog post from Geoffrey Litt, based on some ideas pitched some 30-ish years ago by Mark Weiser

Blog Title: "Enough AI Copilots, We Need AI HUD's"

https://www.geoffreylitt.com/2025/07/27/enough-ai-copilots-we-need-ai-huds


My (Non-AI Powered) Summary, with additional spin on top: 
1) The industry at large is going about this all wrong!

2) The "me too" industry frenzy currently of just piling in on top of the "tasteless" (to paraphrase Steve Jobs) zombie train of building digital versions of Tom Riddle's horcrux-diary is stupid.  

    I'm with Mark on this - as those who've followed me for long enough know, I'm very much against all this "chat" and "agent" style crap that all the AI-TechBros are currently all still hyping up.

3) HCI + Interaction Design specialists/experts such as myself should really be stepping in and stepping up to meet this challenge head on - to get back in there to steer the ship as it were, instead of being backseat passengers to an almost certain trainwreck - by doing what we do best:

    a) Taking a step back, and asking the essential basic questions about what exactly we're trying to really achieve, and **WHY**...

    b) Crucially, not letting the existing framing cloud our judgement, and obscuring our own personal ethical + philosophical principles on the direction that we wish to steer technology in

4) What is it that we have to offer then?

   A) Show Don't Tell - If your tech is really that fancy, it doesn't need to be in our faces all the time. We shouldn't *need* to be constantly "conversing" with it, like micro-managing a third-rate annoying minion

   B) An Assistive Superpower - (Now where have we heard *that* before?  ;)    Yep, the emphasis here should be on the tech taking care of the mundane stuff, while using its strengths to highlight stuff we can't figure out as humans...

*NOW* we're talking.
  i) Harnessing tech to help people do what they couldn't otherwise do
  ii) Augmenting + Building Up, NOT Replacing + Subjugating!

These are indeed very much things that are right in my wheelhouse, and problems that I can get behind!  (Whereas the "AI" discourse to date has very much been a very alienating, and unpalatable soup of world-destroying slop that make me and countless others sick to our very core)

Thanks for the reminder Geoffrey!

 

As for the rest of us:   It's high time we started getting cracking, and righting this ship! There's lots of work to do!

Thoughts About "AI" (Winter 2024 Edition) - AKA: No, I do NOT want to have to "talk" to your "chatbot"

I briefly interrupt coverage of my Music Visualisation Project to cover a brief rant about the topical "AI" issues that are all the rage right now.

 

My current position on all this "AI" hype is:

1) TBH, I bloody HATE all this "me too" bandwagon jumping crap that's going around at the moment, and hope it all blows over sooner rather than later - just like "Crypto" and "NFT's" and "Metaverse" fads before it did. The sooner the better!

See also this "supremely on the point" blog post ;) -  https://ludic.mataroa.blog/blog/i-will-fucking-piledrive-you-if-you-mention-ai-again/

 

2) The UX of all these "AI" tools is fundamentally flawed:  i.e.  

     "I do NOT want to have to fucking 'talk' to your bloody 'chatbot' to do stuff!"

 

3) The majority of all this "AI" hype is all being poured into all the wrong directions: 

    "We should be focussing our efforts on helping people do what they cannot do otherwise (i.e. augmenting human abilities),  NOT trying to replace them  (i.e. destructive misery causing)"

    That there is perhaps the best way to sum up the ethical line / standard I use to decide what I spend time working on. I'm only interested in working on stuff that betters humanity's ability to do stuff they otherwise wouldn't be able to do without the technology. Other stuff (e.g. ad networks, DRM, fintech, killer robots, facial recognition, tracking + surveillance tech, making people/industries/etc. "redundant", etc.) I refuse to work on  (and really, anything I am not interested in, I do a categorically *awful* job at...)

 

4)  In that light, will I work on or play with AI stuff at some point?

     Short Answer:  If AI is the right tool for the job, I will consider it.

     Operative word: "right tool"

     So far, none of the problems I have been working on have required reaching into that toolset, so I haven't bothered to really delve too deeply into it. But if the opportunity arises where AI presents a better solution than we can achieve otherwise, I will consider it then.

     Prime Example:  With some of the image generation + editing tech out there now, we finally have the a set of powerful tools for fixing a whole bunch of previously prohibitively difficult-to-fix problems, giving us the ability to do spot fixes for defects that would've previously ruined many images / videos. In that sense, these user-guided "repair" tools are precisely the "powerful magic fix-it tools"  that we've all dreamed of having all these years, and so, by my previously stated principles, they may well be the right tool for the job in those cases. But using these tools to construct entire fabrications from scratch, trained off everyone's data (however ill-gotten)? Nope - that's pretty much something that should not be done!

Useful AI Wishlist

 List of AI things would be actually worthwhile to pursue (instead of some of the junk that goes on):

* Earthquake Prediction. Not "forecasting" (i.e. vague regions + probabilities, over long periods). Prediction.

* Motion blur removal from single-shot, low resolution JPG's.

* Using content recognition to identify similar features across multiple content streams, for precise timestamp synchronisation.

* Removing rap vocals from music while keeping the general idea