I’ve written a “reflections on the year” piece ever since I left the UK government in 2021. In that time I’ve reflected on: going into, and then leaving, freelance work; starting, and then ending, a business; moving to Berlin; and the tech politics of 2024. Previous ones are here: 2024, 2023, 2022, 2021.

This year will be somewhat a continuation of last year’s, partly because I have (for the first time in a long time) held one role all year, as Head of Tech Research for AlgorithmWatch in Berlin, Germany.  But partly because there’s still more reflecting on how to do tech politics/policy given the current state of politics and the world in general.  I’ve clustered the reflections around a series of headings beginning (in sometimes contrived ways) with P: Power, Possible Positive (a)pproaches, Process and its Problems, and Predictions (or not). These are all my personal views, which are broad and mush together analysis of what I’ve been seeing and my emotional responses to all that. 

Power

Last year I had “impunity” as one of my key words, focused mostly on Big Tech CEOs.  This year made me feel that was too narrow, and the question really is the classic one “who has power and how do they use it?” 

Despite (legitimate) concerns, I’m not sure we’ve seen the Trump administration – or anyone, for that matter – successfully orchestrating a total power grab.  And many other governments seem to clearly be losing control.  Rather, than centralised power, it feels like democracies are in constant, visible, internal contest. Everyone is trying to exercise their own power to its limits (and beyond), prompting retaliations from others, etc.  Governments are being replaced by voters every election.  Those currently in power change course based on public opinion, or vocal subsections thereof, or market reactions.  Courts are making consequential decisions (not just in the US – decisions by German Courts have also been important players in the downfall of the previous Coalition government, and in ongoing debates around the powers of the AfD). 

There’s possibly a parallel between democracy and AI discourse here.  AI risk discourse often talks of “off switches”.  I.e. – if something starts going disastrously wrong, can we pull a switch to make it stop?*  Right now it feels like the various off-switches – or reset-switches – of democracy are being pulled a lot.  All these patterns may not be new, but they seem increasingly visible and dramatic.

* = Who “we” are who get to decide something is going “disastrously wrong” is, of course, part of the issue.  I imagine off-switches would be pulled much quicker to protect California than Africa.

This is, of course, a reason for standing up for democracy, precisely when it seems to be at its most chaotic: Democracy is a system with off-switches beyond “violent overthrow of power”.  But existing in this mass of visible power competition is challenging for me, and others like me who (i) want to use our time to “improve the world” in some way, but (ii) don’t have a lot of power to do so, and (iii) are often pushing for changes which go against very powerful actors (tech companies and, sometimes, governments).  What are you supposed to do in this context?

Possible Positive (A)pproaches

So 2025 was a complicated year for anyone trying to do things like develop and enforce political rules, or really any pro-democracy activity.  Various people will have had different experiences of that.  My personal one was a constant sense of “what are we actually supposed to do here?”

I keep thinking of a half-remembered quote from an MP (I think Jess Phillipps but maybe someone else) who came into Parliament from charity and advocacy work around a particular cause.  I have tried and failed to locate the exact quotation, but it’s something along the lines of “I spent ages fighting to change things from within the system, and then realised I needed to change the system”.  I held that view of my work around regulations in 2023/4.  I found that doing research and proposing solutions was getting harder, in the face of increasingly self-interested behaviour by tech companies.  I felt working on systems – of rules, risk assessments, accountability, etc. – was a good way to push back.

But right now it feels that trying to improve any system is a constant challenge. So maybe it’s worth taking a step in the opposite direction to our MP, and thinking of the smaller, more tactical, positive steps that can be taken within the system, unhelpful though the system may be? What might tactical “throwing mud at the wall” approaches be, trying out and improving through doing, rather than up-front strategizing (see the famous art class experiment about that)?  

One of my favourite outlets this year has been 404 Media, who do tech investigative journalism on specific stories, not bigger “system change” work – but, by their own account, still have real impacts.  I was struck by their “behind the scenes” podcast, in which the whole thing really does seem to be quite a tactical and ad-hoc arrangement - a small group, backed up by an organically grown network, pursuing things they find interesting as they arise.

An example from the development of my own work.  In 2024 I (with AlgorithmWatch) produced multiple proposals on how to implement EU platform regulation better (see e.g. our proposal on a “dual track” approach to systemic risks under the Digital Services Act).  I still think that work is useful and was, in that more regulation-friendly time, a good approach.  But this year I focused on more targeted problems rather than underlying systems.  

A topic I was particularly proud to work on (and am still working on) was Non-Consensual Sexualisation Tools (NSTs), often known as “nudify” apps.  As the name suggests, these tools can create sexualised images and videos of real people, without their consent.  My approach, with AlgorithmWatch, to this topic has been:  EU regulations should make it easier to research how big platforms may contribute to the problem, and push for mitigations. We test that, and if it doesn’t work, make that clear to the EU.  I do not have answers on how to improve this system if it isn’t working.  But if the system of online regulations can’t help against non-consensual nudity (e.g. if X can just refuse to give us relevant data, despite there being rules to show they should), that is something to point out.

Some things I’ve been thinking about for next year is what positive tech-and-democracy examples would look like.  Given my personal interests, this has largely focused on how to design a more pro-social and pro-democracy internet.  But I’m also intrigued by the possibilities of innovations to unblock internal governmental and regulatory problems (see next section).  Next year I plan to highlight more of these positive examples and best practices of innovation in politics, society, democracy – and the role of technology in all that.  Ideally I’d be able to find examples from sources that my field doesn’t normally notice.  I’m sure sometimes this will sound like irritating “toxic positivity”, and sometimes it’ll just be “this seems interesting” without a wider strategy. But I think it’s a good approach.

This isn’t an argument against being strategic and coordinated some of the time.  That is probably necessary for really big, systemic-level improvements which are (hopefully) in our future.  And, of course, a binary divide between “change system” or “act within system” is too simplistic.  It is possible to push on multiple fronts at once, joining up little bits of power that different groups have, and trying to be greater than the sum of our parts.  The nudify work I mentioned before, for example, is drawing heavily on prior research from, and very helpful conversations with, organisations like Indicator Media, Bellingcat, and the Institute for Strategic Dialogue; the specific gap I wanted us to fill, based on AlgorithmWatch’s expertise, was how EU regulation might be part of the picture.  I stand by a point I made last year, one of the joys of my field is many of the people are a joy to work with.

But it can also get quite complicated.  And being a part of that as an individual – particularly one who is sitting between these various nodes – can be disorientating.  Should I be doing XYZ, or is someone else already doing that? What happens when one node (e.g. regulation) is becoming slow and complicated, with knock-on effects on the rest of the system?  This is not to argue against coordination, but also to point to the risks when everything becomes extremely joined-up and complicated, and every new change produces a flurry of reports and thinkpieces and workshops [OM1] which one simply must engage with.  This question of everything being complicated and interdependent is a bigger one, which leads us to our next topic.

Process and its Problems

Last year I wrote:

There’s now a broader coalition who are very sceptical of regulation, running from the unlikely bedfellows of the MAGA movement in the USA to Mario Draghi and supporters in the EU. There is an increasingly pervasive narrative that “regulation” (in general) in the EU is part of its economic stagnation. ... It’s a running joke that everyone (including CEOs) says “I’m in favour of regulation, but good regulations”, before defining “good” as “whatever supports me”. But it’s a question I want to treat as more than a joke this year.

I think this topic is increasingly going to define progressive politics over the next few years.  In fact, this section got so out of hand that I’m going to do a separate post on it.  But some of what I was going to say is below.

In the USA, a deregulatory approach for Democrats is gaining prominence as part of the “Abundance” movement, spearheaded by the highly influential columnist Ezra Klein.  This argues, borrowing from the blurb of Klein’s book Abundance (with Derek Thompson) that “laws meant to ensure that government considers the consequences of its actions have made it too difficult for government to act consequentially.”  There is now an Abundance Summit, an Abundance network, and I’ve even seen it referred to as “the A-word” in left-wing American circles.

As another example, the damaging government-stripping of DOGE prompted interesting responses from the Re:coding America movement, led by Jennifer Pahlka who co-founded the US Digital Service in the Obama administration (the Service was transformed into DOGE, which she has roundly criticised).  The Re:coding America movement calls for government to work better, largely via technology, with some deregulatory flavour in mottos like “Paperwork Favors The Powerful” and “Reinvent Oversight”.

We can look beyond the US. The Abundance agenda is also reportedly of interest to centre-left governments in Australia and the UK (the book’s blurb features Andrew Marr calling it “Downing Street's current hot read”).  In the EU, following Mario Draghi’s report last year there have been a string of proposals – not clearly “progressive” coded in the same way but still quite “centrist” or “mainstream”- for “simplification” or “deregulation” of EU rules (the label you get depends on the person you ask), including the controversial Digital Omnibus Package. Here are some other examples of progressive deregulatory approaches from a range of countries.  (The UK, as always, is sitting somewhere in the middle of the US and the EU.  This could be an interesting opportunity, with the application of some strategic thinking…).

There have, of course, been past progressive deregulatory moves (a classic example being Al Gore’s National Partnership for Reinventing Government). Many of the macro regulation vs. deregulation arguments are not new, whether about the economy and innovation, level playing fields, or whether vulnerable groups are helped or harmed.  But the current sense of existential despair may encourage more openness amongst progressives to different approaches than in the past. 

Also, something I’m noticing more in the discourse is a sense of bipartisan frustration at regulation, the “good people are trying to do good things but can’t because…” arguments.  The concern (which I share): cumulatively, such rules impact many different people in many different ways.  The resultant frustration can have political valence, as people feel penned in and progress held back by a mesh of well-intentioned Chesterton’s Fences.  That leaves room for more macho, far-right-coded, “this is too complicated smash everything” approaches.

Responding to such concerns with “progressive deregulatory” approaches may require some uncomfortable discussions (the Abundance agenda is already quite polarising, probably not helped by Klein’s mixed reputation in left-wing circles).  Even phrasing the issue is challenging.  What does it mean to have “more” or “less” regulation, in the abstract?  What do phrases like “light-touch regulation” or “simplification” mean – when are they euphemisms for “don’t actually regulate”, and when are they “know goal, try to reach in relatively straightforward way”?  What trade-offs are truly fair, particularly when comparing severe harms to minority groups against more diffuse benefits to wider numbers? 

The obvious response is those questions cannot be answered in general.  We have to look case-by-case at regulations – whether new ones, or old ones – aim to maximise the good bits and minimise the bad bits, including any complicated overlaps between regulations, loopholes etc.  The problem is, that in itself requires a lot of evaluation and discussion and huge amounts of time and effort to do properly.  And you end up again with that flurry of reports and consultations and debates, and attempting to simplify becomes very complicated – even before you start implementing. Again, I’m going to return to these thoughts more fully in a later piece. 

But to briefly give my view – and here I am borrowing heavily from Re:coding America, specifically Pahlka’s book (which I recommend): Implementing regulation better is key.  In particular, one issue may be too much up-front thinking and not enough doing, testing, and responding.  Pahlka quotes the recollection of Tom Loosemore, one of the founding members of the UK Government Digital Service, on his experience when when policymakers worked more closely with implementers:

… "You know," [a policymaker colleague] said, "I've come to the realization that what I'm holding is really six hundred pages of untested assumptions. I'm just going to set it aside for a while."… [Tom] has now seen that realization occur many times. "Why," he asks, "is so much policy educated guesswork with a feedback loop measured in years?"

This is something I feel I see a lot with EU regulation – changes end up being effected through lengthy reviews, highly controversial “big bangs” such as the Digital Omnibus package, and mediated by the interests of the EU Commission (usually calling for more centralisation, as noted by Mark Scott). 

Pahlka’s suggestion is that we need more use of technology and focus on giving officials more autonomy to deliver.  Again, there are difficult discussions to be had about such steps, particularly when governments outsource decision-making to technology (how politicians and officials use chatbots is another of my ongoing topics with AlgorithmWatch). Also, continually changing regulations comes with its own issues, both in terms of compliance and also which groups have most resources to engage with continual proposals and reworkings (hint, probably companies more than NGOs).  But again, these complexities are all things I’ll discuss in a future post. 

In the meantime, in the spirit of trying to look for possible positive case studies, cases of Implementing Regulation Better is somewhere where I’m likely to look.  Suggestions welcome.

Predictions (Or Not)

It’s increasingly popular in turn-of-year blogs to do predictions and hold oneself accountable for them a year later. It’s a good method for increasing one’s analytical skills in general (I talked a bit about Superforecasting a while ago, here).  I haven’t done that, i.e. a particularly systematic review of what I’d predict for the year ahead, partly because of the general world chaos and partly festive lethargy.  Maybe I’ll sit down and make some more precise predictions another time.  For now my main, if vague, prediction is the one above, that the regulation / deregulation debate is going to become a big topic of contention amongst progressives in the US and Europe. 

But – I am thinking about the future a lot currently, in a different way.  I’m interested in where we should be “looking for the future”.  Half-baked thoughts on that coming up. 

I’ve been increasingly interested in some of the origins of Trumpism, from the 1990s up the 2010s – particularly following a trail left by the excellent Sam Freedman in this piece on “The origins and ideology of Muskworld” and his interview with John Ganz (author of When the Clock Broke about Trump-esque figures around the 1992 US election).  In this period there were people noting and paying attention to this bizarre conglomeration of nationalism, conspiracism, media savvy figures, and (increasingly) links to technology companies, even if it was not being taken seriously.   

A more specific example is the rising importance of internet-facilitated toxic masculinity AKA “the Manosphere”.  This can be seen at the political level – witness the prominence in the current US administration, as well as sympathies for the far right from young men (interesting piece on that by Cas Mudde here), as well as the prominence of the UK government’s new violence against women and girls strategy.  It’s also prevalent in other contexts: witness the success of the show Adolescence, and reported concern amongst teachers about misogyny in pupils. 

The Manosphere is something which many prominent voices in my field have been talking about for years (e.g. Julia Ebner), and I think its current importance could have been addressed earlier in circles beyond these expert groups [OM3] (but what would have been different if we had? Don’t know. See my earlier points about power – would the issues of the online Manosphere really have been effectively mitigated by tech platforms, even if there was more of a push to do so?).

Who or what might be the trends that are recognised, known about now but their significance not yet fully understood? Rather than going all-in on particular movements or ideas as “the future”, it may be worth continually scanning around.  Jamie Bartlett’s excellent book Radicals attempted to do this in 2017.  Some of the groups he studied – such as the new far right – are now, indeed, shaping the mainstream.  Others – such as transhumanists and psychedelic enthusiasts – have not (yet).

Maybe it’s time for an update; just as the 1990s may have held the seeds of Trumpism, might our current times hold surprising seeds of something (hopefully) better? And what, if we found them, could we do with that knowledge?

I’ll finish with one particular thing I think worth watching, which could have a range of possible impacts (positive and negative).  This is the emergence, into voting and political activity, of people who get their news and politics almost entirely from “digitally native” content.  While social media has been a huge determinant of what people see for a while now, it was often pushing material and narratives shaped largely by mainstream media.  

That is different to a substantial proportion of the population receiving most or all their narratives – on politics, on news, on health, whatever – from a disparate collection of influencers, podcasters, etc. etc.  Joe Rogan, the “fitness and general masculine-coded culture” podcaster turned Trump backer, is the key example of this**. But there will be others.  And that may mean many of the concerns that I, and others, say are currently overblown – online hyperfragmention, living in separate realities etc. – become more real.*** 

** From the Reuters Digital News Report: “One-fifth (22%) of our United States sample says they came across news or commentary from popular podcaster Joe Rogan in the week after the inauguration”.  That is a serious rival to many mainstream media offerings.

*** Even then, we should still avoid ascribing our political situation too much to online platforms, and downplaying the influence of things like economic reality/perceptions.

I think this topic could and should become a key focus of investigations, with some urgency.  These should look at YouTube, TikTok, and (maybe) some Meta platforms as well as podcasts and actual data on emerging platforms (and not on X, which is small and overemphasised in popular discourse).  They should ideally have components of ethnographic data, seeing how people use these platforms in their daily lives, as per danah boyd’s classic 2014 book It’s Complicated on teenagers and the internet (though if this research is to have a wider life beyond academic circles, the conclusions can’t be “it’s complicated”).  And, finally, they should learn from studies of how people use social media in Global Majority countries, where the media (social and mainstream) landscapes can be quite different and – maybe – an insight into where other countries may head.  With that, I’ll leave this thought there.

Personal Recommendations

These ones are “from across 2025” so a broader mix than usual.

In case anyone means to do this but, like me, may need a reminder: Donate to some charities you care about.  ‘Tis the season. Some additional thoughts on doing that in times of crisis, such as situations like the Gaza conflict, here.

For anyone starting 2026 either looking for or offering new roles in European (inc UK etc.) tech policy, here is a googledoc I try to maintain.

Some learnings for me this year in terms of productivity, self-organisation, etc:

• Put to-do lists, things-to-read lists, etc. in places I habitually check, not in some different app or document which I have to consciously open. My to-do lists are now in my calendar app; I send random thoughts and hyperlinks to myself on Signal; my ebooks are in my mobile phone not just a separate e-reader; German language things-to-remember are my desktop background. 
• Personal work coaching has been really good at giving me a chance to discuss and reflect on how I work, what I need, etc.  (Though a meta-learning, it definitely benefits when you put in your own work and preparations to think about what you need from sessions).  I have been with Sequoya Coaching, specifically Svenja, and it and she have been extremely helpful.
• Some people use the Pomodoro method (25 min focused work, 5 min break) for writing.  I now use it for basically all desk work.  It helps set boundaries around tasks and forces me to realise e.g. that I have misjudged how long a particular task is taking rather than sticking at it for too long.  Windows laptops have a “focus mode” in the clock app built for that.
• I have learned to recognise little signs that I’m falling ill (usually, irritability) and then become very strict about working from home and cancelling plans for a day (or two) if necessary, as a preventative approach.  For a year with many infections flying around, I’ve avoided feeling too ill at any point.  (I’ll accept that I may be lucky in being able to spot such things and respond quite flexibly).
• My usual work tip, learned from the Civil Service – when dealing with busy, senior people and decisionmakers, they want proposals and recommendations (even if that’s multiple possible options), not just problems.  I said more about that here.

I didn’t record books that I read and struggling to remember many, which suggests there weren’t many stand-outs.  The aforementioned Re:coding America is good (I’m only partway through Abundance so no comment yet).  Alan Rusbridger’s book Breaking News was also good on journalism in the 21^st Century, and impacts of the internet – pointing out how many issues were bubbling up before the explosion of social media into politics.  Wolfgang Münchau’s Kaput was interesting and very brutal on postwar German economic strategy (also honourable mention to The Shortest History of Germany which was good, if not actually that short and I still preferred Faust's Metropolis: A History of Berlin by Alexandra Richie). Fiction-wise, This Is How You Lose The Time War by Amal El-Mohtar and Max Gladstone is clever and creative and fun, even if (like me) you aren’t normally a science fiction fan.

Newsletters, blogs, and other regular written media:  404 Media and The Indicator on tech journalism; Digital Politics on, well, digital politics (mostly regulatory/governance issues); Better Conflict Bulletin on polarisation, with tech angle; Comment is Freed on UK and Geopolitics, also strategy; Young Vulgarian for funny/intriguing cultural commentary.

Podcasts:  For tech-focused stuff 404 Media, again, but also Techdirt (various tech topics), Decoder (interviews with various leading Tech people), Ctrl-Alt-Speech (online issues), and Tech Policy Press podcasts. Capitalisn’t, Lawfare, Not Another Politics Podcast and Ones & Tooze are not tech-focused but good detailed dives into law / political sciences / economics and related matters from expert perspectives, and increasingly tech features in these (especially Lawfare, with their new-ish series about AI).  I keep up with UK politics/government from Berlin largely via Institute for Government, plus Political Currency for more insider perspectives.  German listeners may also enjoy the very funny Too Many Tabs.

Fun Fact About… Squirrels

In German they call Squirrels “Eichhörnchen”.  (The British Embassy in Germany, as an April Fool, claimed that tourists may need to say this before entering, along with “Quietscheentchen” and  “Gewürzgurke”).

Breaking down the work, “Eiche” is acorn, “Horn” is horn, and “chen” plus changing the o to ö implies something is small and cute (one of my favourite German words is “Hallöchen”, a sort of cute hello).  So it could roughly translate as “Little cute acorn-horn” which… kind of makes sense?  But – sadly that etymology is far too neat and it’s actually from Middle High German eichurne.

BUT – it turns out the etymology of the English “Squirrel” is actually more interesting!  It comes from the Ancient Greek word σκίουρος, or skiouros – or “shade tale” - from σκιᾱ́ ('shade') and οὐρᾱ́ ('tail').  So maybe related to “obscure” in root? (My in-house ancient languages advisor can neither confirm nor deny at this stage).

The other thing I learned when looking this up is that “Hörnchen” is used in German to refer to various squirrelly animals.  So please enjoy, both as animals and as excellently long German words, the Bünthörnchen (“colourful Squirrel”), the Königsriesenhörnchen (roughly “King Sized Squirrel”), and Bhutan-Riesengleithörnchen (“Bhutan Giant Gliding Squirrel”).  Happy new year.