Re: The Slopification of the CCG

Not to denigrate anyone's posts, including this one, but one complaint
seems to be the length of posts contributed by AI.  I find it ironic that
many such posts are quite long.  I personally find that content
interesting, but perhaps, in the interest of people with less time to read
than I, any post longer than a paragraph or two should start with a tldr.

--------------
Alan Karp


On Thu, Apr 23, 2026 at 10:20 AM Moses Ma <moses.ma@futurelabconsulting.com>
wrote:

> Nikos,
>
> I wanted to applaud your candor, which was refreshingly straightforward.
> However, I urge everyone to adhere to the unspoken goal of inclusivity
> here. We need to be peacemakers.
>
> Anyway, your post primed the pump of ideas.
>
>  1) A small experiment I ran recently showed ~25% of people in this forum
> can’t reliably distinguish LLM output from human writing. So some of what
> gets labeled “AI slop” is actually just perception. But that’s almost
> beside the point here. The real issue isn’t AI—it’s verbosity as a
> strategy.
>
> 2) Standards groups generally follow a few structural dynamics:
>
>    - inclusion > exclusion
>    - visibility = influence
>    - language = control
>
> What’s changed is that AI has reduced the cost of writing. So individuals
> who were already inclined to, ah, “over-contribute” can now scale that
> behavior, flooding the channel with low-signal, self-promotional, or
> tangential content. These people are like the blowhards at a company who
> believe that talking really loud and constantly mansplaining is a success
> strategy. This is not the kind of leadership we need in the 21st century.
>
> The failure mode isn’t just annoyance—it’s attention capture by volume,
> where a few verbose participants degrade signal-to-noise for everyone else.
> What we really need is an AI with a  *bloviation sensor*. Most of us do
> this internally by simply not reading certain posts, until we’ve had enough
> and lash out. Then the bloviator is justified in feeling attacked.
>
> Therefore, instead of  debating tools or reputation, it may be more
> productive to consider the possibility of placing lightweight guardrails on
> contribution quality:
>
>    - Contribution caps per cycle (forces prioritization)
>    - One idea per message (no multi-topic dumps or longwinded responses
>    like this one)
>    - Editorial compression rights (chairs or AI can edit and summarize
>    without loss of weight)
>    - Track signal-to-noise to reward reputation for increased caps (what
>    actually survives into the draft)
>
> Something like this would get us out of policing individuals or tools, and
> back to protecting the quality of the work.
>
> 3) I think that in a few years, people who refuse to use LLMs—essentially
> “artisanal writers”—will seem as quaint as luddites who refuse to use
> Google.   I added the em dashes to show that a human generated response can
> still use em dashes, they are not the six fingers of LLM writing. I like
> them because they force the mind to “slow the breath” while reading.
>
> To wrap up, the happy ending I’d love to see is something likely
> impossible. My preference is a new kind of process that nurtures growth, by
> encouraging the hesitant to find their voice, novices to get up to speed
> faster, and the emergence of greater self-awareness by the bloviators.
>
> I’m actually working on a Web 89.0 vision of this vision (haha)….
>
> My incubator is building a “stealth-ish mode” startup called EmergentYOU
> with the goal of creating something that could provide a labor transition
> cushion for the AI era—using longitudinal coaching, hyperpersonalized
> learning pathways, and an AI career co-pilot to continuously align people
> with opportunity. It converts disruption into mobility by linking skills,
> employers, and outcomes in a closed-loop system that compounds human
> potential over time. We’ll announce the EmergentYOU concept at Human Tech
> Week in San Francisco next month.
>
> Anyway, I’ve been thinking about we’re extending EmergentYOU into
> EmergentUS, to support teams: an intelligence layer designed to increase group
> cohesion, reduce participation disparity, and enhance group consonance.
> The system nudges quieter participants to contribute, modulates dominant
> voices, and elevates high-signal input—creating balanced, adaptive dialogue
> and measurably stronger collective performance. However, the real issue is
> that the entire team would need to agree to undergo the process. If there
> is interest in piloting EmergentUS in an SDO context… let me know.
>
> – Moses
>
>
> PS, if you’re in the SF Bay Area and would like to attend our event at
> Human Tech Week… let me know too.
>
>
> <moses@nureon-eda.ai>
>
> On Apr 23, 2026 at 12:55 AM, <NIKOLAOS FOTIOY <fotiou@aueb.gr>> wrote:
>
> Hi all,
> I think we are losing the context here. The problem is that certain
> perfectly identifiable individuals spam the list with mostly meaningless,
> self-promotional content. For example, every couple of messages I receive
> some web 7.0 irrelevant, non sense. AI tools have just made their job
> easier to generate content. Blaming AI tools is just a polite way to say to
> those individuals “please stop you are creating too much noise”
>
> Best,
> Nikos
>
> 23 Απρ 2026, 10:36 πμ, ο χρήστης «Amir Hameed <amsaalegal@gmail.com>»
> έγραψε:
>
> 
> Hi Adrian,
>
> I do not think the concern is about restricting the use of tools. People
> will use whatever tools are available to them—that’s inevitable.
>
> The issue is that reputation alone is not a strong enough primitive for
> systems that aim to operate at scale and across jurisdictions.
>
> In distributed environments, we typically rely on properties like:
>
>    - verifiable provenance
>    - non-repudiation
>    - integrity of authorship
>
> These are not about limiting expression, but about ensuring that
> contributions can be evaluated independent of the individual’s perceived
> credibility.
>
> Saying “my reputation will suffer if I’m wrong” assumes:
>
>    1. reputations are consistently observable across contexts, and
>    2. reputational consequences are sufficient deterrent
>
> In practice, neither assumption holds reliably—especially in global,
> asynchronous systems.
>
> On enforcement: Global enforcement is not realistic. That’s precisely why
> systems tend to push guarantees down to verifiable layers rather than
> relying on behavioral expectations at the application layer.
>
> So perhaps the problem is not tool usage vs. responsibility, but:
>
> how do we make authorship and intent more verifiable without constraining
> participation?
>
> Regards
> Amir Hameed
>
> On Thu, 23 Apr 2026 at 12:15 PM, Adrian Gropper <agropper@healthurl.com>
> wrote:
>
>> It’s fundamentally unfair to restrict my use of technology if I’m willing
>> to take full responsibility for the posting. My reputation should suffer
>> just as much if a post offends regardless of what tools I may have used.
>>
>> The problem seems to be that we have no way to enforce human
>> responsibility.
>>
>> As I see it, this is the only problem. I wish we were discussing
>> solutions.
>>
>> Adrian
>>
>> On Tue, Apr 21, 2026 at 9:04 AM Amir Hameed <amsaalegal@gmail.com> wrote:
>>
>>>
>>> We are discussing decentralised standards on a centralised email mailing
>>> list which is open to receive anything , it worked earlier because there
>>> was a limited capability a user had in terms of what they could research,
>>> type an email, structure it well and then send it to the mailing list, we
>>> had very few people who really were willing to put their work and time and
>>> help develop standards , few years back the same user has been handed over
>>> a tool where he can write a sentence and get multiple paragraphs answer
>>> that too structured in a intelligent way but may not be factual, it’s
>>> obvious that users who ever wished to write an email to the mailing list
>>> but could not do that due to lack of both energy to research , draft and
>>> put it forward for discussion might think of using these tools to overcome
>>> that barrier to entry, it’s similar to industry revolution, there was a
>>> time when only elite could afford a car because there was no assembly line
>>> and it was done with hands manually , once we had assembly lines anyone
>>> could buy a car if they had money.
>>>
>>> Our current technology has reached another assembly line moment, this
>>> time it’s not cars but human skills, reasoning ,  and information systems.
>>> So this points us to something deeper and that is we need to rethink the
>>> entire process now, patching doesn’t always help like Kyle said ,
>>> reputation is not helpful in open ecosystems , we may have to elevate the
>>> criteria of what is valuable once intelligence and skills become a
>>> commodity and we need to think of humans as artists in the industrial
>>> world. Technology is not always the only answer , before we decide anything
>>> , let’s step back and rethink how the whole thing has changed ever since
>>> intelligence became commodity and generative tools became digital
>>> replacement of human skills. We may not have the mailing list itself in
>>> future , transition period is always chaotic and we collectively navigate
>>> it, I strongly believe for a better solution we need to rethink and come up
>>> with some fresh perspectives like verifiable provenance, proof of
>>> expertise, proof human , otherwise human signal will drown in this
>>> asymmetry.
>>>
>>>
>>> PS: it’s written by me no tool was used in this except the mail itself ,
>>> It took me few more minutes but it’s worth it
>>>
>>> Regards
>>>
>>>
>>> On Tue, 21 Apr 2026 at 11:27 AM, Kyle Den Hartog <kyle@pryvit.tech>
>>> wrote:
>>>
>>>> Reputation systems work well as a heuristic metric when you’re
>>>> operating in high re-interaction environments. That’s not really the case
>>>> on the Web because of its openness properties where it's easy to build up
>>>> and spend down identities in an automated fashion. It's made even easier
>>>> with LLMs now too.
>>>>
>>>> For example, on this mailing list spammers could form new emails in
>>>> seconds and form new identities to continue their attacks. If you set up a
>>>> guard to prevent it you've now accepted the tradeoff of reduce openness and
>>>> entered a cat and mouse game at the same time. There are discourse forums
>>>> (polkadot and ZCash are 2 examples where I've encountered this) that have
>>>> these techniques built in where you can only post once you’ve built up a
>>>> reputation. They have specific threads that allow people with low
>>>> reputation to engage and then you earn reputation over time. This comes
>>>> with the tradeoff of reducing the openness of the system in exchange for a
>>>> higher bar of entry. Maybe a poster has something legitimate to add to the
>>>> conversation, but because they didn't build their reputation up enough they
>>>> can't contribute. With automation like LLMs given to attackers these days,
>>>> it's producing an asymmetric attack surface and reverting the solution more
>>>> towards option one (Dark Forest theory - retreat to safe communication
>>>> channels).
>>>>
>>>> Another example where we're dealing with these sorts of low value
>>>> sybils is in Brave's hackerone bug bounty programs. There's evidence[1]
>>>> from BugCrowd this could be security vendors using this to gather training
>>>> data, but it also simply could be someone operating out of a lower wage
>>>> country where one bug bounty report can be worth a month's salary or more.
>>>> So they're incentivized to use an LLM to generate new identities on the
>>>> fly, spam bug bounty programs, and if their signal degrades too much they
>>>> drop and swap them.
>>>>
>>>> Additionally, I’m not sure how much you’ve been following the Web3 and
>>>> public goods funding/DAO spaces, but they’ve actually been relying on these
>>>> identity credential systems as a sybil resistance mechanism for a bit now.
>>>> While there’s been mild success shown, the system over time has had to add
>>>> capabilities to address different attacks that have been conducted. For
>>>> example, Gitcoin Grants 24 saw a 60% reduction in sybil attack influence
>>>> from their GG23 round[2]. They’re the most widely deployed system that I’ve
>>>> seen trying to actively go down the route of identity based protections for
>>>> Sybil attacks and spam. Worth a look for you at least but it's also worth
>>>> pointing out they're producing a system that structurally still faces the
>>>> problem as long as the incentives for conducting the attack are still high
>>>> enough ($1.8 million dollars was given out in GG24). For their system they
>>>> rely on over 20 different potential signals including government IDs,
>>>> biometrics, social signals, and financial signals (Binance accounts which
>>>> require KYC)[3]. Even then, people are still successfully conducting
>>>> attacks against this system and as more systems are built on the same
>>>> identity credential based sybil resistances (aka the reputation system atop
>>>> it) the value of conducting a sybil attack grows because it can be
>>>> repurposed across multiple systems.
>>>>
>>>> There's 2 other deployed identity credential systems that have also
>>>> been working on this problem as well in the Web3 space with some issues.
>>>> Idena[4] and Worldcoin[5] have fallen susceptible to some form of Sybil
>>>> attacks also. From what I've seen, people are conducting "puppeteer
>>>> attacks" where one person "puppets" many people who have digital IDs to
>>>> coordinate in the system and conduct attacks. These typically occur
>>>> through an attacker paying for some action to be taken in order to conduct
>>>> the attack. Again, these attacks are usually successful because they're
>>>> operating out of lower wage countries where the seemingly smaller amount of
>>>> money paid makes the attack worth it.
>>>>
>>>> The point here is that attaching reputation systems onto this means
>>>> you're in for a attack surface that has historically struggled to keep up.
>>>> I'm not convinced that an email list is ready to deal with this let alone
>>>> technology built through a standardization process that takes years to
>>>> iterate on. Especially when the human(s) who are participating is actively
>>>> coordinating with agents to conduct the spam or sybil attacks. So yeah,
>>>> that's why I'm not really convinced identity credentials are going to be
>>>> that useful. I'd be happy to be wrong, but what I'm seeing both in terms of
>>>> real world adoption as well as attacks I've had to deal with (we've seen
>>>> these sybil attacks against other systems in Brave too) identity
>>>> credentials only go so far in solving the problem and they come with
>>>> tradeoffs that normally aren't worth it.
>>>>
>>>> Here's some links for the citations made above as well.
>>>> [1] Bugcrowd:
>>>> https://www.bugcrowd.com/blog/bugcrowd-policy-changes-to-address-ai-slop-submissions/
>>>> [2] Gitcoin reduces attacks:
>>>> https://gitcoin.co/research/quadratic-funding-sybil-resistance
>>>> [3] Gitcoin Signals:
>>>> https://support.passport.xyz/passport-knowledge-base/stamps/how-do-i-add-passport-stamps/the-government-id-stamp
>>>> [4] Idena:
>>>> https://stanford-jblp.pubpub.org/pub/compressed-to-0-proof-personhood/release/5
>>>> [5] Worldcoin:
>>>> https://www.dlnews.com/articles/regulation/singapore-officials-warns-against-worldcoin-account-trading/
>>>>
>>>> -Kyle
>>>> -------- Original Message --------
>>>> On Tuesday, 04/21/26 at 05:16 Casanova, Juan <J.Casanova@hw.ac.uk>
>>>> wrote:
>>>>
>>>> Kyle,
>>>>
>>>> You say
>>>>
>>>> Identity credentials are highly unlikely to stop this either which I
>>>> suspect is where many in this community would want to turn. Identity
>>>> credentials just turn the issue back into a key management problem and we
>>>> don’t really have a great way to prevent a user from sharing their keys
>>>> with their agent. That problem persists whether the system has a delegation
>>>> solution or not too.
>>>>
>>>> I think there may be an important "but" to this. I think some of the
>>>> things you suggest later may relate to it, or some of the ideas that Will
>>>> discussed later. I'm definitely sure that there has been much more
>>>> discussion about things like this and more attempted approaches to similar
>>>> things that I am aware, as I still consider myself a newbie here. However,
>>>> let me state my view...
>>>>
>>>> While you can't prevent a user from sharing their keys with their
>>>> agent, you can have, like you said "pseudo-reputation" systems attached to
>>>> keys, that take time and good contributions to build, and are deteriorated
>>>> when providing lower quality contributions. I believe this can be achieved
>>>> without systematically breaking sovereignty. These hypothetical system(s)
>>>> could span across multiple mediums, they don't need to be constrained to
>>>> single contexts, and be optional and complementary rather than strictly
>>>> enforced, but they could help both as a deterrent for people haphazardly
>>>> sharing unfiltered AI contents (I refuse to use the word slop because I
>>>> feel it has connotations that challenge civil conversations and is pretty
>>>> much a slur, even if I understand what people mean by it), and as a way for
>>>> people to identify and neutralize persistent sources of it.
>>>>
>>>> In my view, this is no different to what we already do in our physical
>>>> embodied life. We have face recognition embedded into us (most of us), and
>>>> we learn to create an internal opinion of other people based on their
>>>> interactions with us. When somebody consistently steals our time with
>>>> pointless drivel and unfiltered contributions, we don't need to put them in
>>>> jail, put a sign over their heads that says they are unworthy, or
>>>> (generally speaking) prohibit them from participating in public life. We
>>>> simply don't pay as much attention to them, because we know who they are
>>>> and what their usual approach to contributions is. Identity online simply
>>>> can replace the face recognition in a way that is more flexible and
>>>> preserves sovereignty better, as well as being better equipped to deal with
>>>> the volume.
>>>>
>>>> As I said, I'm sure I am unaware of the extent to which similar ideas
>>>> have been proposed and explored. I am also very aware that in the same way
>>>> that some people here are using questionable predictions of what AI *
>>>> will become *that, whether grounded or not, remain just a prediction
>>>> and not a current reality that can be wielded as a definitive argument for
>>>> what to do right now; what I am discussing here is also a prediction or a
>>>> hope, rather than a current reality. But in the same way that I think it's
>>>> valid to work towards better AI tools, I think it's valid to work towards
>>>> systems that enable us to better * filter through the ocean of
>>>> information* in ways that respect sovereignty for all sides involved,
>>>> can be personalized, and respect our own intelligence. I think it's a dream
>>>> worth pursuing, and I believe it relates directly to the current matter.
>>>>
>>>> But in the meantime, I feel that discussing like we are doing seems to
>>>> already be shaping a lot of moderate people's views into compromises that
>>>> may make this mailing list more comfortable for everybody involved. One way
>>>> or another, we will find out.
>>>>
>>>> *Juan Casanova Jaquete*
>>>>
>>>> Assistant Professor – School of Engineering and Physical Sciences –
>>>> Data Science GA Programme
>>>>
>>>> *j.casanova@hw.ac.uk* <j.casanova@hw.ac.uk> – Earl Mountbatten
>>>> Building 1.31 (Heriot Watt Edinburgh campus)
>>>>
>>>>
>>>>
>>>> Email is an asynchronous communication method. I do not expect and
>>>> others should not expect immediate replies. Reply at your earliest
>>>> convenience and working hours.
>>>>
>>>>
>>>>
>>>> I am affected by Delayed Sleep Phase Disorder. This means that I am an
>>>> extreme night owl. My work day usually begins at 14:00 Edinburgh time, and
>>>> I often work late into the evening and on weekends. Please try to take this
>>>> into account where possible.
>>>>
>>>>
>>>>
>>>>
>>>> ------------------------------
>>>> *From:* Kyle Den Hartog <kyle@pryvit.tech>
>>>> *Sent:* Sunday, April 19, 2026 06:28
>>>> *To:* Steve Capell <steve.capell@gmail.com>
>>>> *Cc:* Melvin Carvalho <melvincarvalho@gmail.com>; Marcus Engvall <
>>>> marcus@engvall.email>; Manu Sporny <msporny@digitalbazaar.com>;
>>>> public-credentials@w3.org <public-credentials@w3.org>
>>>> *Subject:* Re: The Slopification of the CCG
>>>>
>>>> You don't often get email from kyle@pryvit.tech. Learn why this is
>>>> important <https://aka.ms/LearnAboutSenderIdentification>
>>>> ****************************************************************
>>>> Caution: This email originated from a sender outside Heriot-Watt
>>>> University.
>>>> Do not follow links or open attachments if you doubt the authenticity
>>>> of the sender or the content.
>>>> ****************************************************************
>>>>
>>>>
>>>> In case it helps, here’s how things are going in terms of AIPREFs WG
>>>> and the impact on search crawlers:
>>>>
>>>> https://x.com/grittygrease/status/2044152662673752454?s=20
>>>>
>>>> In other words, we don’t really have any enforcement mechanisms here to
>>>> stop this. In fact I highly suspect some people are using them in this
>>>> conversation right now unless their writing styles dramatically changed in
>>>> the past few years. My email client started noticing it via machine
>>>> learning I suspect and filtering threads to my spam inbox like this most of
>>>> the time given I engage a lot less these days. Personally that’s been a
>>>> good enough solution for me.
>>>>
>>>> Identity credentials are highly unlikely to stop this either which I
>>>> suspect is where many in this community would want to turn. Identity
>>>> credentials just turn the issue back into a key management problem and we
>>>> don’t really have a great way to prevent a user from sharing their keys
>>>> with their agent. That problem persists whether the system has a delegation
>>>> solution or not too.
>>>>
>>>> So where do we go? I’m not exactly sure. Here’s the leading theories
>>>> and their tradeoffs that stand out to me for the generalized solution of AI
>>>> generated content:
>>>>
>>>> 1. https://www.ystrickler.com/the-dark-forest-theory-of-the-internet/
>>>> - users just stop engaging in these spaces and retreat to closed door
>>>> forums. Then we lose the open collaboration that made the Web great.
>>>>
>>>> 2. Re-hash DRM debate by making it so users can’t actually access their
>>>> keys used to sign their identity credentials. This seems to be the current
>>>> path governments like. It optimizes enforcement but also entrenches access
>>>> to the Web around a select number of OSes and reduces who’s allowed to
>>>> access and contribute to conversations on the Web. I also see that as a bit
>>>> short sighted.
>>>>
>>>> 3. Re-introduce fingerprinting (and pseudo reputation to that
>>>> fingerprint) based identity like what CAPTCHAs do. That works well for
>>>> service side enforcement but in mailing lists like these not so much. So
>>>> likely will need user controlled filtering like what spam filters in email
>>>> clients do as well.
>>>>
>>>> 4. Is the most interesting but most unproven. We shift how people are
>>>> reachable and build out Horton Protocol like what Mark Miller proposed
>>>> years ago at ActivityPub conf. They may have already tried this and had
>>>> issues. I’m not exactly sure:
>>>> https://www.youtube.com/watch?v=NAfjEnu6R2g
>>>>
>>>> In any case though, we don’t have much of a solution right now in our
>>>> particular forum and outside things like 3, I don’t expect much to change
>>>> in a coordinated manner right now. Looking forward to seeing what we come
>>>> up with though over the next decade and hopefully the trade offs we make
>>>> don’t take away too much of what originally made the Web great.
>>>>
>>>> -Kyle
>>>>
>>>>
>>>> -------- Original Message --------
>>>> On Sunday, 04/19/26 at 13:10 Steve Capell <steve.capell@gmail.com>
>>>> wrote:
>>>>
>>>> Challenge : there’s an increasing amount of AI generated content that,
>>>> whilst possibly containing useful insights, takes more time to read than to
>>>> generate and, given the size of this mailing list, is likely to lead most
>>>> of us to unsubscribe, rendering the list worthless
>>>>
>>>> Constraint : AI used well is a genuinely useful tool and can
>>>> dramatically improve quality of output.  “Used well” is key and,
>>>> unfortunately, many do not use it so well.  Nevertheless, this group can’t
>>>> become anti-LLM luddites or this list may equally become worthless for the
>>>> opposite reason
>>>>
>>>> Goal : to continue to enjoy intelligent discussions between real humans
>>>> that feel empowered to use AI to improve the value of their human
>>>> contributions.  So the goal, it seems to me is not to block AI content but
>>>> rather to block content that has little evidence of human analysis and
>>>> interpretation.  Perhaps counterintuitively, LLMs themselves might be the
>>>> best tool to detect such content
>>>>
>>>> Proposal : rather than continuing to discuss whether AI content on this
>>>> list is good or bad, let’s collectively agree a rubric in the form of an AI
>>>> prompt that can act as an automated list moderator.  The rubric should
>>>> focus on requiring evidence of human assessment rather than blocking AI
>>>> content
>>>>
>>>> I had a go at this myself with several of the messages in this thread
>>>> and earlier ones and it seemed quite effective at blocking the ones that I
>>>> would have blocked myself.  I know that there is a token cost associated
>>>> with such a moderator but I for one would delighted to contribute.
>>>>
>>>> Disclaimer : this message was written with blurry eyes and fat thumbs
>>>> on my iPhone - with no AI assistance whatsoever
>>>>
>>>> Kind regards
>>>>
>>>> Steven Capell
>>>> UN/CEFACT Vice-Chair
>>>> Mob: +61 410 437854
>>>>
>>>> On 19 Apr 2026, at 10:03 am, Melvin Carvalho <melvincarvalho@gmail.com>
>>>> wrote:
>>>>
>>>> 
>>>>
>>>>
>>>> ne 19. 4. 2026 v 1:49 odesílatel Marcus Engvall <marcus@engvall.email>
>>>> napsal:
>>>>
>>>> Hi all,
>>>>
>>>> I’m glad to see that we have some healthy discourse in this thread with
>>>> a variety of views. I would like to address some of the points made.
>>>>
>>>> On 18 Apr 2026, at 01:50, Melvin Carvalho <melvincarvalho@gmail.com>
>>>> wrote:
>>>>
>>>> LLMs have the advantage that they know most or all of the specs
>>>> inside-out, due to their training. Most humans (with notable exceptions),
>>>> including on this list, have partial understanding of the complete works of
>>>> web standards.
>>>>
>>>>
>>>> This is a real advantage that these tools have and it should not be
>>>> understated. I use them professionally for referential lookups and for
>>>> confirming hypotheses, and I have no doubt that they have the ability to
>>>> accelerate otherwise excellent standards work. But I am also careful to not
>>>> fall into the trap of assuming that their lexical consistency can fully
>>>> substitute  for human judgement. LLMs are probabilistic models with
>>>> encyclopaedic knowledge, they are not deterministic oracles with the
>>>> capacity to rigorously derive that same knowledge. In the context of the
>>>> kind of work done in this group I think it is important to not confuse the
>>>> two. I trust an LLM to give me a comprehensive overview of a standards
>>>> framework - I do not, however, trust it to prescribe the framework itself
>>>> without and human review and editorial judgement.
>>>>
>>>> I do however concede on your point on testing methodology, and I think
>>>> you raise a good point that Manu eloquently touched on.
>>>>
>>>>
>>>> Good points. However LLMs outperform humans on medical exams,
>>>> olympiad questions and many other tests, often by wide margins. They are
>>>> much more than prediction machines or probabilistic guessers. What I'm
>>>> saying is that I predict LLMs would exceed humans in the standards setting
>>>> on any quantitative evaluation. We just have not the tools to evaluate yet.
>>>> However, I believe the picture will be much clearer one year from now.
>>>>
>>>>
>>>>
>>>> On 18 Apr 2026, at 02:24, Manu Sporny <msporny@digitalbazaar.com>
>>>> wrote:
>>>>
>>>> Technology transitions, especially ones around human communication can
>>>> be rough to navigate. This one is no different, and sometimes it takes
>>>> decades to figure out the norms around a new medium (the printed page,
>>>> radio, television, BBSes, mailing lists, AOL, ICQ, Napster, Twitter,
>>>> Digg/Reddit/Discord, and so on).
>>>>
>>>>
>>>> You are completely right that this is a transition, and I think we are
>>>> all trying to map this new technology onto our existing mental models of
>>>> what discourse should and could be. Friction and contention is bound to
>>>> arise. It is clearly counterproductive, as you and later Amir rightly
>>>> stated, to enforce neo-Luddism and reject the technology wholesale.
>>>>
>>>> My point however is that the ability to passively follow and
>>>> occasionally contribute to developments and discussions in this group is
>>>> immensely valuable, both commercially and technically. Compressing the
>>>> signal-to-noise ratio raises the bar for both comprehension and
>>>> participation, and my fear is that the inevitable intractability will, as
>>>> you pointed out in the other thread, overwhelm people and alienate them,
>>>> especially those of us with many other commitments and who do not have the
>>>> time or ability to participate in every group call. That said, it is,
>>>> as you suggested, our responsibility to moderate our own information
>>>> ingestion, as has been the case for time immemorial in any rhetorical forum.
>>>>
>>>> Perhaps LLMs will simply change the structure of how discourse is
>>>> conducted in forums like these rather than drown it out, as some other
>>>> writers have suggested in the thread. If the cost to contribute text tends
>>>> to zero, naturally the valuable discussions will shift elsewhere to forums
>>>> that still have a cost, such as the group calls. I just hope the work
>>>> doesn’t lose the diversity of opinions that is crucial to develop a refined
>>>> and well-considered standard.
>>>>
>>>> --
>>>> Marcus Engvall
>>>>
>>>> Principal—M. Engvall & Co.
>>>> mengvall.com
>>>>
>>>> ------------------------------
>>>>
>>>> Founded in 1821, Heriot-Watt is a leader in ideas and solutions. With
>>>> campuses and students across the entire globe we span the world, delivering
>>>> innovation and educational excellence in business, engineering, design and
>>>> the physical, social and life sciences. This email is generated from the
>>>> Heriot-Watt University Group, which includes:
>>>>
>>>>    1. Heriot-Watt University, a Scottish charity registered under
>>>>    number SC000278
>>>>    2. Heriot- Watt Services Limited (Oriam), Scotland's national
>>>>    performance centre for sport. Heriot-Watt Services Limited is a private
>>>>    limited company registered is Scotland with registered number SC271030 and
>>>>    registered office at Research & Enterprise Services Heriot-Watt University,
>>>>    Riccarton, Edinburgh, EH14 4AS.
>>>>
>>>> The contents (including any attachments) are confidential. If you are
>>>> not the intended recipient of this e-mail, any disclosure, copying,
>>>> distribution or use of its contents is strictly prohibited, and you should
>>>> please notify the sender immediately and then delete it (including any
>>>> attachments) from your system.
>>>>
>>>>