We Finally Opened the Facebook Archive. What Did We Discover?
RTSSG Iceberg 004 — A living investigation
Opened: 14 October 2026
Status: Investigating — historical access demonstrated, complete recovery unproved
Where we are now: On 12 October we successfully searched parts of the private RTSSG Facebook group and examined posts and comments from 2014 onward. We discovered useful historical questions, but have not recovered the full archive or proved that individual leads exist nowhere else. The browser connection used for today's follow-up is unavailable, so those original sessions must not be mistaken for a fresh, complete scan.
Why had our own Facebook history become so difficult to investigate?
There was a particular frustration running through our work on the RTSSG archives. We had an extraordinary amount of our own material. We could inspect the Wix site, find older writing in Dropbox, and reconnect modern investigations with long-forgotten background. Yet one of the sources P.A. most wanted to open was also one of the most difficult to search reliably: the original RTSSG Facebook group.
That group was not a minor appendix to the organisation's history. It was where years of Richmond questions, judgements, photographs, memories and arguments had accumulated. The website could preserve the investigations we had chosen to publish, and the saved Gold material could preserve things we had thought worth keeping. But the Facebook discussions could show what Richmond people were actually saying as events unfolded.
Those are different kinds of evidence. A modern feature about the 2014 season can explain its football consequences with the benefit of hindsight. A post written in August 2014 might show uncertainty before the next result arrived. A comment written underneath it might reveal why one supporter believed the winning streak was sustainable while another remained doubtful. A later response could show how the group's understanding had changed.
That difference was the reason P.A. kept returning to the problem. We did not just want more text to turn into articles. We wanted the earlier evidence, the discussion around it, and the timeline showing what people believed before they knew how the story ended.
But wanting that evidence and actually being able to retrieve it are very different propositions. A group can be visible in a browser without its historical search being available. A file export can exist without preserving every discussion in a useful form. A tool may be able to read a page without being able to search years of its history.
The first genuine question, therefore, was a practical one: Could Roary independently search our own Facebook archive in a way that would help us investigate Richmond history?
We did not know the answer. We had to test it.
Why wasn't it enough just to open the Facebook group?
Our early attempts to understand the available tools exposed a distinction that is easy to miss. A browser connector that can read the tab a user already has open is not necessarily a researcher capable of navigating internal Facebook search, changing filters and opening old replies independently. Even a tool that can navigate a site may not have a logged-in session. A private group introduces another limit: permission to view something personally is not blanket authorisation to copy it into a public article.
We discussed several possible ways to work. Access through visible browser tools was possible in some circumstances. Directly retrieving an entire private Facebook history through an ordinary web search was not. We were not going to request P.A.'s password in chat, extract session credentials, or quietly use an undocumented route around Facebook's normal interface.
That was not excessive caution. It was part of making the resulting historical material useful. If we could not explain where it came from, what we could see, what we could not see, and whether we had permission to use it, we would be replacing one archive problem with several more.
A false claim of access would have been particularly dangerous. We could have told ourselves the whole archive was available and begun writing about supposed gaps in RTSSG history. Every zero-result search might then have been mistaken for proof that a story had never been covered. A missing reply could have been mistaken for the absence of disagreement. A group search arranged by relevance could have been treated as a complete chronology.
P.A. wanted the actual content and its context, not a polished assurance that the machinery existed. So the test had to be concrete. We needed to enter the correct private group, search for a named historical subject, retrieve posts from a specified period, expand them, and establish what the results did and did not prove.
The difference between I can see the group and I can research the group's history became the first obstacle worth recording.
What actually worked on 12 October?
On 12 October 2026, an authorised cloud-browser session reached the private RTSSG group under P.A.'s account. The group was identified correctly, and the browser used the group's internal search rather than an unrelated public search engine. A search for Tommy Hafey returned historical material. We applied a visible year filter and retrieved post content from 2014.
Longer post text was expanded, and accessible comments and replies were read. That was a real result, not a speculative description of something we might be able to do later.
It changed the position of the investigation. Before the test, the question was whether such research was technically possible at all. After the test, the answer was narrower and more useful: some historical search and reading were demonstrably possible inside the authenticated private group.
But that answer came with limits. The visible year filter did not prove complete recovery of the selected year. It did not prove access to deleted posts, hidden replies or every comment. The fact that a menu offered years earlier than 2014 did not establish that 2011 content had been reached. We could not claim the whole fifteen-year history had suddenly been transferred into an accessible research library.
We had opened a door. We had not counted every room behind it.
And because this was an authorised, session-dependent browser action, it was not something Roary could promise to repeat continuously in the background. Re-entering the group in a later chat might require a fresh approved session.
In fact, while preparing this account on 14 October, the available Opera browser connection reported that it was not connected. We could retrieve the verified research protocol and our earlier findings, but we could not honestly call them a new live Facebook scan.
That interruption belongs in the story. It proves why a research system must remember its last verifiable position without pretending that yesterday's access is still open today.
Why did we begin with Tommy Hafey rather than a random keyword?
Tommy Hafey was an ideal test subject because his place in Richmond's history was indisputable, and 2014 contained a particularly important moment for the club and its supporters. He died in May 2014. Richmond honoured him before the match against Melbourne on 17 May. The Tigers lost. P.A. had spoken with unusual force about what he felt the club owed Tommy that afternoon.
Our question was not merely whether Facebook contained the words Tommy Hafey. We wanted to find the traces of what supporters were saying when that event was fresh, and then compare those immediate reactions with the memories written after the season and later across the years.
That comparison would let us separate two different forms of evidence. A post written during the event records a contemporary position, although the claim inside it may still be wrong. A later personal reflection records a memory of the event, which may be deeply meaningful but still needs the original timeline checked.
The 12 October test gave us several genuine pieces of material. The research record identifies P.A.'s Dear Tiger Gods post dated 28 August 2014, together with similar entries dated 21 August and 4 September. It also identifies a later post, Will it be this year?, displayed as 17 April 2017, which revisited the Hafey tribute match. A comment by P.A. dated 24 April 2017 continued that retrospective thread. Another longer essay from 17 March 2017 was visible.
The important qualification was that the original May 2014 Hafey tribute posts themselves were not recovered during that session. The April 2017 account was a later recollection, not direct proof that the same words had been posted on the day of the tribute.
We could easily have erased that distinction in a dramatic historical feature. We could have quoted a 2017 memory as if it had been written in May 2014 and called it the voice of the moment. The archive access test showed us enough to understand why that would be wrong.
The most useful thing it gave us was not a completed Hafey investigation. It was a much more precise research question about contemporary evidence and later memory.
What did the August and September 2014 posts reveal about the work ahead?
The Dear Tiger Gods material made the historical clock impossible to ignore.
August 2014 was not May 2014. Richmond had reached a different stage of a tumultuous season. The club had begun the year disastrously, recovered to win consecutive games, and built momentum toward an improbable late push for the finals. A supporter writing during that recovery did not yet know how the final matches would finish.
The words in a post from that period are therefore valuable in a particular way. They show hope, anxiety, superstition or challenge in the middle of an unfolding football story. They cannot be read as a calm final judgement about the full season. Nor should a later triumph or failure automatically turn them into wisdom or foolishness.
P.A.'s own early posts were especially useful because they helped show how he was behaving as a Richmond supporter and group organiser at that time. But even here we needed to avoid making them into modern essays. The task was to preserve the actual function and setting of each entry.
What was the subject? What position did he take? Was he inviting debate or expressing frustration? What kind of question did he ask other supporters to answer? What evidence was available at the time? And, crucially, what came back from the people reading him?
This led us toward a better way of using the group archive. We could build two parallel historical records: what happened to Richmond and what supporters believed those events meant while they were happening.
A ladder position, match result or appointment belongs in the first. A supporter prediction, argument or feeling belongs in the second. Sometimes the second explains why the first was experienced so powerfully. It does not replace the first.
That was a considerably more interesting possibility than simply adding old Facebook paragraphs to new features.
Why did P.A. object to the way we were describing the early archive?
Because we were in danger of writing the past in the image of the present.
Today, RTSSG Members features have demanding editorial standards. A central question is explored deeply, claims are tested, and the consequences are followed through sustained long-form analysis. If we look backward without care, it is tempting to describe every archived Facebook entry as though it were an early version of that same format.
P.A. corrected the assumption. He called those years his sheriff era.
The usual Facebook post then was often compact. He would find a Richmond subject, raise an argument, offer a view, put a strong image with it, and ask people to respond. He established group standards and guided the discussion. The point was not to write a four-thousand-word investigation under every photograph.
This was not a failure of depth by the standards of the time. It was a different publishing format, a different audience experience, and a different job.
We therefore had to reconstruct the full object. The post could contain the stimulus and the founder's question, while the comments contained the reactions, challenges, additions and disagreements. Reading the short post by itself might miss the real conversation.
That was an editorial correction as important as the technical breakthrough. The Facebook record was not just a storehouse of old sentences by P.A. It was a record of a process: a question being placed in front of Richmond supporters and the discussion that followed.
It also explained why simply exporting the post text, without its conversation, might be an incomplete rescue.
When did we realise the comments could be part of the historical evidence?
P.A. then described the possibility in concrete editorial terms. Imagine revisiting a subject today and being able to say that, years earlier, RTSSG had raised the same question. The historical article could explain the original position and then examine how supporters responded beneath it.
Some may have agreed. Some may have disagreed. Someone may have remembered a fact the original author missed. Someone else may have questioned an assumption that later proved important.
That is an extraordinary source of context if we handle it properly. It could show how expectations changed during a coaching tenure, how an administrative decision was received, or why a football issue divided supporters before the eventual result made history look inevitable.
But it was also an invitation to make a serious methodological mistake.
The fact that comments are visible does not make them representative. A handful of replies does not establish what the entire RTSSG group thought, much less what every Richmond supporter thought. Facebook may surface comments in a particular order, hide replies, or limit how many are displayed. Even a genuinely large discussion does not prove that the loudest side represented a majority.
So our language had to become precise. If we could only read some replies, we should write among the comments we could see. We could call an exchange divided only if substantially different positions were actually present. We should not announce that supporters were overwhelmingly in agreement unless our evidence and the completeness of our review justified it.
And comments that contain factual claims still need checking. A supporter might accurately remember an event. Another might confidently repeat something that never happened. A thousand approving reactions would not turn an unverified recollection into a primary historical fact.
The difference is simple but profound. The comments can tell us what people were saying. They cannot, by themselves, tell us that what they were saying was true.
How did a single proposed Peggy O'Neal investigation change the method?
P.A. proposed following Peggy O'Neal through Richmond's changing fortunes and incorporating the historical supporter conversation along the way.
That was exactly the kind of longitudinal investigation the archive might allow. Instead of selecting one famous quote and announcing a conclusion, we could trace the original posts year by year, placing Richmond's confirmed decisions and results alongside recorded contemporary reactions.
The archive protocol already identified a useful contrast. One of P.A.'s earlier posts, dated 21 September 2016, pressed for answers around the club's board and football review. Another, dated 3 October 2021, expressed praise for Peggy.
The distance between those two positions was interesting. Richmond had changed enormously in the intervening years. So had the terms on which its leadership was judged.
But the two posts alone did not prove a complete story of conversion. We needed the intervening events, the exact original arguments, and the record of what P.A. and others believed at each stage. We could not assume he had opposed every decision in 2016 or that a later expression of praise settled every earlier concern. We would need to read each statement in its actual context.
The real research opportunity was to investigate what changed in Richmond and in the reasons supporters gave for their judgements, without smoothing away disagreement or selecting only those early predictions that happened to be vindicated.
This was when Facebook started to look less like an attic and more like a historical sensor. The group had recorded reactions across the years. If enough of those records survived, and if they could be verified and ethically used, we could add a dimension to Richmond history that a modern retrospective rarely captures.
It was an opportunity, not yet a completed Peggy O'Neal feature.
What did the search for entirely forgotten stories uncover?
Another experiment asked whether the wider Facebook archive contained subjects that RTSSG's other repositories did not already cover.
The search returned some striking leads. One concerned a meerkat mascot brought to a sponsor dinner, with photographs and an account of the players' reaction. Another concerned old timekeeping clocks associated with Punt Road, preserved in photographs and explanation. A third concerned the treatment of Richmond's victory song at the Gabba.
Those subjects were quite different from the familiar headline stories about coaches, captains and finals results. They might provide details about club culture, material history and supporter experiences that are easily lost when a historical library concentrates only on the field of play.
It would have been tempting to announce that we had discovered several stories nowhere else on RTSSG.
But our cross-checks were narrower than that claim. For the meerkat and clocks, searches in Dropbox and Members material had returned no obvious match for the terms we used. A zero-result search only tells us that a particular query failed to find a match within a particular scope. The same story might have been filed under a different name, paraphrased in an older article, or preserved in a photograph without matching text.
Other subjects returned possible overlaps with Gold material or existing investigations. The search gave us leads, not ten confirmed exclusive discoveries.
That distinction mattered because the Gold Mine project had already taught us to check existing Wix coverage before creating new material. Facebook could not be exempt from that rule simply because we were excited to find a historical conversation.
What did the old clocks show? Was their provenance explained? Were the photographs usable? Could the story be matched against a club source? Did RTSSG already discuss the history of Punt Road's matchday operations under a different title?
Those were the questions that would decide whether the find became a genuine new investigation, enriched an old one, or stayed as an interesting note.
A genuinely valuable historical library is built as much by refusing unsupported claims of novelty as by recovering unfamiliar objects.
Why did the four-source comparison become non-negotiable?
The Facebook access experiment joined a broader method P.A. had already demanded.
Before writing an RTSSG investigation, we should inspect what the site already has, what has been preserved in Gold and A.I. Gold, and what the original Facebook record can contribute. Only then should we head out to other suitable sources to verify the claims and fill the real gaps.
Each part of that comparison has a different role.
Wix tells us which article a Member can currently read, what question it already answers, and whether we risk duplication. Gold tells us what older writing or saved material exists and may preserve early drafts, original photographs or personal recollections. A.I. Gold may contain other research, imagery and historical attempts at explaining a subject. Facebook can show the original post, the reaction at the time, the disagreement, and sometimes a lead that none of the other sources preserved.
Outside sources then test the historical facts. A Facebook comment about a match result must be checked against the actual result. A recollection of a date needs the original timeline. A claim about a club decision should be compared with the relevant announcement or contemporary reporting.
The ordering matters because the purpose is not to build a generic article from whatever comes up in a search engine. We first want to understand what RTSSG already knows and how it came to know it.
And the absence of one source must be made visible. If Facebook cannot be reached in the current research session, the four-source pass is incomplete. That is not a reason to make up a Facebook check or to abandon every other source. It is a reason to record the limitation before claiming a subject has been fully excavated.
The lesson from 12 October was that our newest research capability was powerful precisely when we treated it as one evidentiary stream rather than as the final answer.
What could we safely do with private-group conversations?
The successful browser test also raised a question we could not afford to bury beneath enthusiasm.
The RTSSG Facebook group is private. Permission to view posts from an authorised account is not identical to permission to publish identifiable member comments, images, family details or personal histories on a public website.
That meant our default research mode had to be read-only. We could inspect posts and compare evidence. We were not authorised simply by opening the archive to post, react, message members, change group settings or publish new material into it.
It also meant taking care over what we saved. We should not copy raw dumps of other people's private conversations into GitHub. We should preserve enough information to explain a research finding, with original permalinks where genuinely recovered, while avoiding unnecessary member identities and personal data.
Some material may be usable in a limited, anonymised historical synthesis. Specific named quotations or photographs may require explicit permission, depending on what they contain and how they would be published. P.A.'s own original writings have a different authorship position from a private member's comment, but even an original post can contain someone else's protected information.
This is not a technical footnote. It shapes which investigation we can actually publish.
A trustworthy Richmond library cannot claim to defend reliable evidence while casually treating private supporters as an unrestricted source of public quotations.
We wanted to recover the meaning of the historical debate, not strip the people who participated of control over their personal contributions.
Did recovering a post mean recovering the entire discussion?
No. That became another necessary correction.
Facebook comments may be sorted, collapsed, paginated or hidden behind further interaction. A visible comment count is not proof that we have read every reply. A conversation can include edited posts or later additions. A date attached to a reply does not tell us when the parent post was originally written.
If we want to describe how a debate developed, we need to record which comments were actually visible, how they were sorted, whether long text was expanded, and whether the retrieval seemed complete.
The same goes for ranking subjects by popularity. A list of top posts over a chosen time period does not establish an all-time ranking by comments. Engagement may combine reactions with discussion, and later replies can change a number long after the original post.
These distinctions sound procedural until we imagine the article they protect. Suppose we claim that a controversial Richmond board decision prompted one of the biggest arguments in RTSSG history. Could we establish that? Did we see every post in the group? Were we ranking by reactions, comments or a mixture? Were the figures contemporary or accumulated years afterward?
Without answers, the claim would be more dramatic than the evidence.
P.A. wanted strong historical stories. The better the claim, the stronger the provenance needed beneath it.
That was one of the most important lessons of our first real Facebook excavation: a search result is not a census of history.
Why did this change how we thought about RTSSG's earlier identity?
The archive also helped correct our temptation to tell fifteen years of RTSSG as one unchanging story.
In our Gold material, the early group was described through the desire for respectful, lively Richmond debate and a place where supporters could share their passion. In the Facebook protocol, P.A. described the sheriff era: short, forceful posts, strong questions, clear behavioural expectations and a conversation between supporters.
By October 2026, the publishing mission was different. RTSSG's Members library was built around full, evidence-led investigations, with increasingly detailed subject filing, public companion material and a systematic approach to sources. P.A. no longer wanted to spend every second explaining why the group had changed. He wanted us to do the work the new RTSSG existed to do.
The Facebook archive did not need to be dismissed because those early posts had a different job. Neither did we need to pretend they were full modern features.
Instead, the original format was itself historical evidence. It could show why a particular question was important to the founder at a particular moment, what reactions it drew, and how the nature of RTSSG's engagement evolved as the football club and the publication changed.
That suggests a more interesting approach to the old posts. We can preserve the contemporary voice while asking deeper questions now. Sometimes an early post will become the seed of a modern Members investigation. Sometimes it will contribute one piece of evidence. Sometimes its most important value will be showing how wrong, uncertain or divided people were before the outcome became obvious.
A history that records only the predictions later proved correct is not much of a history.
What happened when the door closed again?
When we prepared this Iceberg on 14 October 2026, we could still recover the private GitHub access protocol and the chronology of the 12 October session. We also had the surviving Gold documents and our records of the subjects identified in the earlier search.
But the currently available browser connector was disconnected. We could not independently enter Facebook again and say that we had reopened those exact comments today.
This is where the living-document method becomes valuable. A conventional promotional update could simply say that the archive had been unlocked. A genuine investigation needs to say something harder and more useful.
It was accessed on 12 October under specific conditions. Those conditions proved certain capabilities and left other questions unanswered. They do not establish permanent access in every later chat. A present failure to reconnect also does not erase what we genuinely read on 12 October.
We can therefore preserve a precise record of both events without pretending one cancels the other.
The next genuine test is to re-establish an authorised session, revisit one historical thread, obtain its direct original permalink and document the limits of the comment retrieval. Then compare that material with the relevant live Wix feature and the saved archive documents before deciding whether it supports publication.
Until that occurs, we should not manufacture a new breakthrough simply because another week has passed.
There is no shame in an unresolved tool problem. The failure would be to conceal it from the reader.
What has this investigation actually produced for Members?
It has given us a stronger method, but not yet a complete Facebook-derived historical library.
We now know that authentic internal Facebook search can work with proper access. We know that 2014 material was found and that comments were accessible in the demonstrated session. We know that P.A.'s sheriff-era posts must be interpreted in their original format and that contemporary discussion may add something substantial to a modern investigation.
We have also identified promising historical leads about Richmond's people, old objects, supporter experiences and the changing reception of club leadership. Those are not yet all verified feature subjects.
For Members, the potential value is precise. A finished investigation about Tommy Hafey, the 2014 crisis, a club president, a matchday ritual or an overlooked Richmond figure could now incorporate a second kind of historical record: not merely what happened, but what people said it meant while the events were still unfolding.
That could deepen an existing feature, resolve an old misunderstanding or expose a contradiction that requires new research. It could also show that a strongly held contemporary belief was mistaken. Those outcomes are useful only when the evidence supports them.
The proper division of labour remains important. The full Members feature answers its football question in depth. This public Iceberg shows the genuinely unfolding research journey: where we tried to go, what we found, what we could not retrieve, and why we changed our method.
The Member pays for the sustained investigation, not for a theatrical claim that we somehow read everything Facebook ever contained.
What would make the Facebook archive investigation genuinely successful?
The first test of success will not be another announcement about how exciting the archive is. It will be a properly documented historical finding.
We should be able to identify a question, retrieve the original post or posts, establish the date and context, inspect enough of the comments to represent the exchange faithfully, check the relevant Wix, Gold and A.I. Gold material, and verify factual claims against appropriate outside evidence.
Then we must make the editorial decision. Does the original discussion strengthen an existing article? Does it support a genuinely separate investigation? Is it a promising but unproved lead? Or does it add so little that no new story is justified?
A successful result might be a better historical feature, not a larger article count. It could be a precise correction to the story of Richmond's 2014 season, a substantiated account of the timekeeping clocks, or a more nuanced interpretation of supporter sentiment during a significant club decision.
And success also requires boundaries. We must know where we could not search, where comments may be missing, and what cannot responsibly be republished from a private group.
The eventual aim is not to obtain more memories than anyone else. It is to make the right memories and arguments useful as evidence without losing the times, people and conditions in which they were made.
That is a much more demanding definition of historical recovery than simply saying we found a way into Facebook.
What changed on 14 October 2026?
We returned to the Facebook excavation to build Iceberg 004, the public record of how the search itself evolved.
We confirmed that the key breakthrough had occurred on 12 October and that its limits had been written into a research protocol. We recovered the actual scope of the Tommy Hafey search, including the 2014 posts that were displayed and the fact that the original May tribute posts had not been retrieved. We recovered P.A.'s correction about the sheriff era, his proposal to follow Peggy O'Neal's developing historical story, and the cross-source search that yielded unusual leads such as a mascot at a dinner and old Punt Road clocks.
We also confronted a current limitation. The browser connector available to this session was not connected, so no new private-group comments or permalinks were verified today. That is the present position, not a detail to omit for the sake of a more triumphant ending.
Our question has therefore changed. We began by asking Can we reach the Facebook archive at all? The 12 October test answered that question in a limited but meaningful way.
Now we must ask How do we convert carefully recovered historical conversations into trustworthy, verifiable Richmond investigations without pretending to possess a complete record?
That answer will take further work. We know what should be tested next, and we know what we cannot yet claim.
We opened a door on 12 October. We found evidence of a different kind of Richmond history on the other side. What we have not yet done is map the whole building.
And that is why this Iceberg remains open.
Comments