A Big Thing Or A Small

I. PANDEMIC

The People’s Pop Polls end this weekend after four years, three months, and something like 47 tournaments of almost 12,000 pop songs. They started in March 2020, a way of solving a very acute problem which turned into a way of solving something a lot broader.

The very acute problem, which it felt like everyone in the world was trying to solve at the same time, was “how do we get through this?”. This being lockdown, COVID-19, the pandemic. I got the idea of doing a Number 1s poll on Twitter when I realised I was spending all my online time looking at the site, and what I was looking at was driving me gradually insane. The first pop poll was, in fact, a poll about whether I should do a poll? 97% said yes. I started right away.

I got COVID early. A couple of weeks before the polls began, I came home from a conference feeling hot and weird and didn’t move for four days of fever and the worst full-body aches I’ve ever had. By the time I put together the list of tracks we were going to poll, I was bone-deep exhausted but recovering. I assumed it wasn’t COVID, it seemed so unlikely that I’d be one of the first to have it. When my wife, who caught it from me, shifted from coughing to gasping for breath, I knew I’d been naive.

By that time I’d started doing the polls, 4 matches a day, spaced out. They became an anchor as I nursed Isabel through what became a frighteningly serious case and looked after the kids, a bit of cheery normality in a surreal situation. We hit the worst of it on March 23, the day Johnson finally declared lockdown, and, as it happens, my birthday. A sharp decline; a 999 call; the promise of a callback and an ambulance (neither happened; the system was in pieces); and then, thankfully, signs of recovery. Twitter records show that on that day, one of the worst of my life, I also put up two pop polls. Anything to keep going.

You might conclude from this I was a little deranged. A lot of us were. Once it was clear Isabel was past the worst, our experience of lockdown was boredom more than fear. We’d had COVID – I think I probably did naively assume it might be a one-off, or at least for a while. I would occasionally mention it to people on my allotted trips out – “oh no, I’ve had it” – but stopped after I noticed how they flinched a little.

So I was able to think about the polls. They were a distraction, but I was also feeling like they were a kind of public service, especially on a site so racked with dread and speculation and, increasingly, lies. I was doing my bit to help a few hundred people have less of a shit time than they might otherwise. The Number Ones poll ended as a procession for Donna Summer. I started a Number Twos tournament the next day. This required a little more selection, but was even better received. Still, all good things must come to an end: I decided to do a third and final poll, and turn the vague social good I felt I was doing into something more tangible before everyone got tired of it.

The third poll would be a charity one, raising money for domestic violence charity Refuge, in which anyone could nominate any song, but they’d have to pay to put it in there. I assumed it would be one big sack of well-loved hits, and some of it was (David Bowie’s “Heroes” won it). But something happened that I wasn’t expecting. People nominated songs they loved, tracks which plainly had no hope of winning but that they just wanted people to hear. A gloriously 80s French hit by Gold. Irish comedy troupe The Rubberbandits. Folkie Grace Petrie and her blistering anti-transphobia song “Black Tie”.

The poll was chaos, but glorious chaos. Far fewer people voted – my own Twitter followers would only follow so far – but I was finding wonderful songs I’d never heard. I didn’t want the fun to stop, and besides, things weren’t going “back to normal” any time soon. Why not give the polls their own account, and run new themed tournaments on a regular basis?

I never imagined how well it would work.

II. POLLS

I think from the outside, the Peoples Pop Polls must have looked like the geekiest, most pointless thing I’ve been involved in. Endless cycles of Twitter polls – Twitter polls! – in which famous tracks inevitably beat obscure ones, and then fought it out amongst themselves. I took it terribly seriously, and allowed it to eat a ferocious amount of my time, working out how to tweak the formats and organise the tournaments in order to get the maximum amount of heartbreak, joy and surprise out of every month’s event. I think I became extremely good at this. My tournaments were the Rolls-Royces, the Faberge eggs, of song-based popularity contests on social media.

As skills go this is not an especially valuable one. There is no need to go to this level of detail and craft to run a poll on Twitter, then or now. You can very easily get a list of 64 cakes and call it The World Cup Of Cakes and get lots of engagement – more votes, given a decent follower base, than any of the pop polls. But raw engagement, that debased coin, isn’t the point. Well, OK, it is a bit. I was excited when we hit the sweet spot of 500 or so voters which seemed to give us intriguing matches with not-that-conservative results. The crucial number, though, wasn’t the maximum voters we ever got (about 1500 in the back half of the 1977 poll), or even the average, it’s the minimum. How many people can you get to listen to and offer an opinion on Mexican crust-punk or forgotten psy-trance remixes or unsuccessful contestants from the country heats of 60s Eurovisions? Or whatever else our nominators rustled up.

In their heyday that number hit 200 – I think there were only one or two polls where we managed 200 votes on every single match, but many of them almost got there. Which underlined the main reason I loved doing the polls. The results were only half the point – less even; the main thing the pop polls were good for was as an engine of discovery, and discovery you were doing alongside other people.

More on that in the next section. Because the results were still important – it’s why we finished the polls off with a giant tournament of 256 of the previous winners. And it’s worth talking about those results and about how we kept the polls entertaining – like discovery, it really was a collective effort.

Twitter is not inherently a great place for music chat. Of course now it’s full of trolls, far-right accounts and pornbots it’s even worse, but the existing Twitter music ecosystem had two big problems for what I wanted to do. One was stans – the way Twitter gave fan armies the opportunity to organise themselves into communities. Nothing wrong with that – or rather the merits and problems of it have nothing to do with the polls – but any close encounter with a large enough stan contingent made a poll pointless. Still, there were things you could do about that – blocks, and rulings on distorted results and ultimately disqualification of some tracks – and those things mostly worked.

The second problem was trickier. The shorthand for it was “Music Twitter”, a catch-all for a free-floating conservatism which surfaces every time you get people to make lists or rank things collectively on social media. Everyone likes different music, but not entirely different music, and lists and prompts would get repetitive as the artists participants liked in common dominated. The loci of attention tended to be smart, bohemian rock music – the Beatles, Kinks, Velvet Underground at al in the 60s; US college rock, indie rock and UK Britpop in the 90s. Between them, the apex of Music Twitter appreciation, the post-punk and new-wave era and the clever, innovative pop and alternative music which came out of it.

Those are all good moments in pop history. In fact very few of the records Music Twitter liked are bad! But it’s the old problem of canon-building – it leaves a lot out, and it gets repetitive. It’s not particularly entertaining to see the same names and faces – pretty similar looking faces at that – surface again and again in music challenges and tournaments. You start to suspect that the hidden point of ‘challenges’ like that is not to challenge yourselves or anyone else, but just to remind yourself, again and again, of the things you already know you love. It’s a way of inspecting your collection in public. And while that’s lovely and comforting and I enjoy it myself, at scale it leads to extremely conservative outcomes. Probably the ultimate example of this was a “greatest albums of the 21st century” Twitter tournament, which concluded that, overwhelmingly, said greatest albums were made by white men in bands. This is not a 21st century I recognise, but it is a reflection of the kind of thing Music Twitter liked.

Still, the polls were about Music, and we were on Twitter. They were a collective endeavour just like any other music challenge. Which meant working out strategies to avoid the dullest outcomes and encourage diversity and lack of repetition. In terms of diversity we did pretty well. Our favourites were still hugely skewed to the 1977-1999 era, with some 00s and 60s/early 70s strongly in the mix, but when you look at the 256 tracks in the final poll, drawn from all the previous winners and high performers, only 24% are by white male artists or bands. The figure for the 2021 Rolling Stone Top 256 songs, by contrast is 43%. Representation is a blunt and problematic instrument, but our list still feels fresher, more exciting, downright better than theirs. (And let’s not forget even the Rolling Stone list was grumbled about by reactionaries as woke nonsense!)

How did we manage this? Three main ways.

The first was careful limits on nominations. If you offer people a wide variety of things to vote on, they will often actually take that chance! It was obvious after the first couple of open nomination polls – the Cover Versions and 1990 ones – that a thumb mildly placed on the scale was needed to make sure we heard as diverse a set of songs as possible. The result was the “Main” and “Bonus” system, where everyone got one free shot at a nomination, and then three other suggestions, and I’d fill out the poll from those suggestions. This produced a wide-ranging set of songs for the qualifying round (it didn’t always mean throttling the indie rock noms, but it did generally mean promoting genres other than rock and pop). It also meant that Nominations Day was a riotous, noisy highlight every month, when people would rush to get their picks in for a new poll and see what everyone else had chosen.

The second way I helped encourage diversity was by making sure like fought like. A big inspiration for the pop polls – as a proof point that you could build a community around twitter polling, for starters – were Kent M Beeson’s “Albums Of The 90s” tournaments, which went about polling very differently. Kent would run an enjoyable seeding ballot first, which meant the weaker LPs would fight the stronger ones in the first round, and the tournament would build towards bone-crunching matches between the absolute best.

Running a song-based poll meant more tracks on a shorter timescale, and I wanted the first-round matches to be as exciting as possible. The conclusion I came to is that if a song couldn’t come 1st or 2nd in a 4-track qualifying group, it had no chance of winning overall – so why not give big songs strong opponents early? And – to minimise the chances of people picking purely on which genre they liked – strong opponents of roughly the same type. The wails of commenter despair when a 1990 qualifier threw EMF, Primal Scream, and the Happy Mondays together confirmed I was onto something. But as an unexpected, but very welcome bonus, it turned out this cruel gladiatorial strategy also improved the range of successful tracks.

And the final way we ended up with more interesting winners was by, frankly, encouraging people who liked them and putting off people who didn’t. What did well in the pop polls set expectations for what might do well in future. If you turned away in disgust because “Love Me Do” got hammered by Dusty Springfield in the qualifiers of the Debut Singles poll, or never got what made “Get Ur Freak On” or “Buffalo Stance” a masterpiece, or simply couldn’t understand why Althea And Donna were so far ahead of the Sex Pistols in a 1977 group, then fine, other challenges were available. (Though if you felt all those things and stuck it out anyway, a sincere thank you, for making the polls less of an echo chamber) And the “like meets like” rule helped too, by creating lots of groups entirely made up of rap, or dance, or songs in other languages.

So voting the way we did helped us self-select for players who wanted to hear lots of different tracks in lots of different styles and weren’t just waiting around for the obvious songs to win (though sometimes they did, and that was fine too). It meant we had a community of voters who liked surprises and liked to talk about music and advocate for it. In the later parts of any poll, the chat around the songs was usually the draw, not just the results. Twitter was not a great format for music criticism – longform arguments or considered statements on records. But it was a very strong format for music chat – pithy descriptions, quick insights, trolling and squabbling, gags and memes. I loved the polls for how much and how well they let that happen.

III. POP

And yet I do think of the polls as an act of music criticism, even though Twitter sucked for it. Some of this is based on what I was just talking about at the most obvious level – the way our take on ‘the canon’, and the favourites we determined via play, holds up very well against other lists or definitions. Beyond the winners, the playlists for the polls were often something special. There were areas we were always weak on – metal and country being the two biggest ones – but I think the playlists for the four end of year polls (2020, 2021, 2022, 2023) were remarkable pictures of those years, capturing their vibrancy and breadth but also the sonic habits and fallback strategies that make a year what it is. And the run of 11 64-track playlists of Black music from 1968 to 1981 are fantastic. Those are a small material legacy for the polls.

They also point at what I think the bigger achievement was. Here I’m going right back to the beginning, where I said the polls solved an immediate problem and turned out to solve a much broader one too. The immediate problem was keeping me occupied in lockdown. The broader problem was music discovery.

Discovering new music is hard. It shouldn’t be – if you use any streaming service you have access to inconceivably huge amounts of new music. Even if you don’t, and have stuck primarily to buying music, bandcamp makes a deluge of digital new music available to you.

But still, discovery is really hard. Streaming services know this too – their algorithms work to find a balance between comfort and discovery, but even the discovery parts tend not to work well for people like the ones I just described, who like to hear a bunch of different things in different styles. Across the internet, the default recommendation mode tends to be “here’s something very like a thing you already like, but not quite as good”.

So what you’re getting recommended is sometimes great, but often kind of mid. That’s the first discovery problem. The second is that there’s so much of it, it’s easy to become overwhelmed, and also to not give new things the attention they might deserve: even a second listen can feel luxurious in these glutted times.

But the third problem is the most insidious. Algorithms isolate you through personalisation. You can negotiate with them to solve the first problem and get better results, learn how to pick trails through the data-maps they create to deal with the second problem, but ultimately it’s just you and an app and some headphones. And that means even when you do discover something remarkable, there’s a bit of a so what factor: the app outsources all social engagement with music, which is a lot of the joy of it.

These three problems are what makes the current music landscape feel so bleak to, let’s just say it, a lot of people my age. Discovery of new music is unreliable (difficult to find good stuff), overwhelming (there’s too much of it) and lonely (there’s no social context for it). Lots of people are finding lots of different and good solutions to these, but they are still problems.

At the risk of sounding like a proper binmen guy, it wasn’t always like this. If you take something like music magazines or Top Of The Pops, or just a good radio station, they offered imperfect but real solutions to these problems – which we didn’t realise they were solving, at the time. Each one offered a reliable source of novelty, and there were enough different sources that you could find ones you were copacetic with fairly easily. They exposed you to a manageable amount of new music, and if you wanted more you could find different sources (something like TOTP or Radio 1 was a source in itself but also a possible gateway to other ones, when a song that spoke to you broke into the chart or playlist). And finally they were social nexus points – a common reference point you had with friends, something to talk about with new ones.

It’s easy to get dewy-eyed over those things but it isn’t useful. The media environment that allowed them to exist has been razed and isn’t coming back. If we think discovery of new music in a social context is a good thing we have to build new ways to make it happen, using the platforms and systems we’ve got.

The pop polls were one such way. Not intentionally: a happy accident! But they did solve the problems. The player base of nominators all trying to surprise, delight, and promote their own favourites made each poll into a pool of shared knowledge and enthusiasm – new music and new discoveries coming together every month, alongside old and rediscovered friends. That solved the reliability problem. A playlist of 250-320 tracks each month solved the glut problem – chunky, for sure, but not overwhelming, and organised so you could pick and choose each day which parts interested you, or tune out for a round and let your peers do the filtering for you. And all this happened in public, exploiting elements of social and streaming platforms (Twitter, Spotify, YouTube) to create a context in which you were discovering and arguing about things alongside friends.

As the polls matured I hit on different ways to emphasise the discovery angle. Early on I started the “Golden Beat” vote, where people could nominate the new-to-them track they most loved. (An “Ultimate Golden Beat” playlist collects the winners and a bunch of individual nominations for all time best discovery). Later tournaments involved a “B-Sides” competition, which siphoned off lesser-known tracks likely to be wiped out early on and gave them more chance to shine. Both these tweaks helped give exposure to ‘smaller’ tracks, with the occasional dramatic fuck-up where I had no idea how famous a song was.

All of this made the polls the most effective way I had of finding new music since my messageboard days of the early 00s. And, to my immense joy, a lot of other people felt the same. The polls were time consuming but they were rarely less than an absolute pleasure to run, with a friendly and easy-going vibe, certainly by Twitter standards. I blocked a bunch of accounts, but almost all for stan activity, and remarkably few for being an arsehole.

The polls solved the three discovery problems I’ve talked about. But there was something else about them too – something in the format itself – that made them work for me. The four song group format turned out to be excellent at rubbing similar and dissimilar songs up against one another, forcing them into conversation by the act of competition.

This is a kind of pop criticism too, or at least it encourages one. I’ve written before about how a lot of the media I loved growing up, and more of the media I just took for granted, came as a bundle. A copy of the NME, a NOW compilation, a 2000AD prog, a radio show, a newspaper, an edition of Top Of The Pops: all of them package deals, in which individual components might gain meaning from their proximity to each other. Bundles solved the problems of discovery I just outlined, balancing the public and the personal – the contents were something everyone had in common; the hierarchy and its meaning were up to you. And also balancing familiarity and surprise, delight and disgust – you might see your favourite artist on Top Of The Pops and be thrown into contempt or confusion by the next one along, but this was a space in which these two were part of a whole and you had to make your own sense of that. One of the things that I keep coming back to with criticism is that those spaces are valuable, however arbitrary they are.

The pop polls were a succession of tiny spaces like that, and they were very arbitrary indeed: to carve any space out of the information maelstrom you have to be arbitrary and artificial. But in forcing you to say “who should win here?” and encouraging you to come up with reasons why, they created something precious. And they made discovery a joy again.

IV. PEOPLE

When the pop polls had got themselves established, I thought they probably had five years in them before we exhausted the format. As it was, they made it to four, and that might have been a year too many. I don’t want to go over the downfall of Twitter, and the abortive move away: I’ll just say that yes, the platform is vastly worse now, just on the basic level of things not quite working smoothly, even before you get to Musk’s desire to use it to tilt the axis of politics harshly rightward. I will miss the polls an awful lot, but I won’t miss X at all.

The last two full, open-nomination polls we ran showed the format at its best – the end of year polls and the Black Pop History tournaments always brought out the best in nominators and were fantastic engines of discovery. But even there we were fighting a losing battle – the number of voters went into sharp decline, partly because of the loss of momentum in the move but partly because non-premium accounts just couldn’t get the reach they once did.

But this gets at one last point about the People’s Pop Polls – despite the name, they could never really scale. They were never meant to. Bundling is forbidding as well as enticing: the fact that a user-friendly group with lots of big hits would immediately be followed by the most obscure private-pressing underground sounds was designed to give exposure to the weird stuff, but it also put people off. And it’s fine to put people off – part of the job of anyone designing worthwhile online experiences now is using friction to make sure things reach the audience who’ll get most out of them. Yes, the polls were a great discovery engine, but they were a great discovery engine for a small group of weirdos who embraced the format.

(Could they have scaled? I don’t know. If someone wants to pay me money to do a corporate version of the Peoples Pop Polls I absolutely would sell out but I don’t think they’d get much joy from it.)

Anyway, this is the part of this long goodbye where I say a heartfelt thankyou to that small group of weirdos, the People of the Pop Polls. I say small, but there are too many of you to name you all, and I would be terrified of missing some of you out. There were dozens upon dozens of people whose regular comments and nominations I looked forward to seeing, and missed if they dropped away for a bit. There were people who I absolutely relied on to cover particular bases, and they always came through – one day, Jel, power metal will get its democratic due. A lot of the time putting together a group I would cackle at the anticipated reaction of some particular voter. (Any time you thought “this group is attacking me personally”, well… sometimes it was. Though attacking isn’t quite the word.) I have, accidentally or by design, built a few small music communities online. You lot were the friendliest and the best.

I do need to mention a few people who went above and beyond, though. Dan Perry, for creating and maintaining the PollHalla playlist for so long, and for helping make the first Black Pop poll a success along with Donna Brown. Graham Meikle, for completing the PollHalla playlist. Maura Johnston, for coming through behind the scenes so often with stuff we’d missed. All the suggestion box stuffers. Dave Cooper Moore, for the ABBAcus and other playlist help. Anyone who uploaded nominations when YouTube was being arsey. Arron Wright, for his own challenges and keeping track of the meta. Everyone who suggested a poll topic but especially Dan Pop-O-Matic for suggesting the Death poll, which could have been a disaster and turned into a triumph. Bartlebooth45 for the placenames map. Idca (and many others) for the memes. All the elves on the 2023 poll. The people who donated to our charity fundraisers. And looping right back to the genesis of the polls, my wife Isabel for listening, nominating swoonsome K-Pop ballads, and for sacrificing many Saturdays to Noms Day.

What next? The People’s Pop Polls on Twitter are over. There will be a pinned post with this piece and a few playlists, and that’s all. I’m going to get back to writing more than polling again. My main social presence now is on Bluesky, where I am easily findable. It is not remotely a perfect social network, but I’m enjoying myself there in a way I wasn’t on Twitter – polls aside – for a long time. If you were part of the Pop Polls, and you’re on there, I’d love to stay in touch.

Bluesky doesn’t have polls – yet. But there is a Peoples Pop account – peoples-pop-polls.bsky.social – in case one day it does. And I have promised to do something on there at the end of this year, some poll-like artefact to cover the music of 2024. It won’t be the same. I don’t exactly know what it will be. But the moment, as they say, has been prepared for.

And on here, over the rest of the summer, I’ll be running a bunch of posts called something like the Pollrunner’s Picks, highlighting discoveries I personally made via the pop polls, the clearest legacy of the honestly lovely times I’ve had doing them.

Thanks.