DLC Guitar Pedals, AI, and Tube Amps: Where Does Technology Leave the Modern Guitarist?

The electric guitar and communities around it have always had a peculiar relationship with technology. The instrument is, of course, defined by its dependence on electrical amplification technology for being played and heard properly. And yet, the culture around the instrument has also has a traditionalist faction that is afraid of doing anything differently from the 1950s-1970s era. Heck, as I write this, the "latest and greatest" line of instruments Fender is promoting heavily is Vintera® II Road Worn® with its "era accurate" pickups or whatever.

I played guitar (poorly) growing up, but took around a decade away from playing. This year, I've fallen in love all over again with playing, but I found a drastically changed landscape compared to that of my childhood.

I hope companies in this industry don't confuse enthusiasm for digital gear with enthusiasm for business models like subscription SaaS or low-effort corporate video game releases.

They say "Leo got it right the first time," but a lot of the people who say that are talking about very specific Strat and Tele design decisions. I suggest that this is basically an example of what overfitting is. Leo did get it right the first time, but the innovation wasn't the specific designs (as good as they were/are) but the fact that the electric guitar has been basically a totally stable modular interface this whole time. There's no dongle hell. There's no "Gibson-diameter cables" and "Fender". Indeed, Leo's designs have powered an ecosystem with remarkable backwards compatibility. My current laptop is basically a super computer for a Stratocaster manufactured in 2008, and yet I can plug my first Squire Strat into my interface like a guitar made in 2025. That is pretty unheard of in the consumer world. But when I took that Strat out to take some pictures for this article, it really struck me how antiquated of a design Fender has stuck to. The neck feels like a baseball bat, and it really looks and feels like something out of the 1950s (and not in the most flattering way) compared to a newer superstrat (even one from within the FMIC stable).

PX-1 Plugout

Boss recently announced and released a pedal called the PX-1 Plugout (I guess this is a wordplay on the concept of a DAW 'plugin') that includes digital models of various Boss pedals with plans to add additional later (that you will need to buy licenses for).

I don't think it's necessary to spend too much time rehashing things or beating a dead horse here, but I do have some thoughts on the matter based on my experience in software development that I want to share.

What I don't see people talk about with with this pedal is how absurdly user-hostile Boss is about firmware updates for it. There is a section of the firmware update page for the PX-1 called "CAUTION WHEN PERFORMING THE UPDATE" that concludes with this gem of a line:

We regret that we are unable to answer questions regarding the update procedure using this system program. Please perform the update responsibly, following the directions given in this document.

And not only can they not answer questions about the update process, the document they are talking about mentions at various points the possibility of destroying the PX-1 or putting it in a state where it needs to be serviced by Roland.

I don't think it would be fair to criticize Boss/Roland for wanting to be more of a software company. But when you want to be a software company, massive amounts of user-facing tech debt like this is not healthy or acceptable. It suggests inadequate resources for engineering and support for the scope of the ecosystem they are trying to build. When you are in the business of selling software, it is outright unacceptable to tell customers that you can't answer questions about updating the software. I thought that's what we were paying you for, Boss!

And this thing is, this isn't new for Boss. It's how they've operated with the Katana too, and it is even worse there. At some point, I will write an anti-katana post here because there is a lot about that amp that deserves to get skewered that just goes unnoticed apparently.

If you are using your pedals to help you make a living as a musician, why the heck would you ever trust a product like that?

Boss likes to emphasize that these "Model Pass" licenses are not subscriptions, but I think it's clear that they are tempted by the GTA 5 dream where they get us to buy their effects multiple times in different platforms/formats/generations.

Subscriptions for guitar amp/fx modeling are definitely starting to pop up, even if Boss isn't leading the way. I think we need to push back against it as much as we can. For me, subscriptions entirely contradict my personal philosophy towards the economics of buying gear. I have spent a fair chunk of change on guitars, pedals, and recording gear, but it's almost like I'm buying time to be creative over years to decades.

I'm not even a Boss hater, either. I think my eq-200 is my favorite pedal.

Can surveillance capitalism lead to good outcomes for buyers of music gear?

Nobody (this is probably an overstatement, but oh well) likes how every app and service is seemingly used to spy on us to target ads and tailor the content it shows us to maximize our consumption of ads.

As I mentioned before, there was a fairly long period of my life where I had fallen out of the habit of playing guitar. I didn't bring my guitar to university, so my roommate from college, didn't know I played. After a brief exchange of chats about how I love guitars, I backed out to the main menu of snapchat and saw a Fender ad sitting in my mailbox! Well, I'm no dummy, I saw what Fender was up to, and they weren't gonna get me! I'm not even in the market for a Fender. So I put my phone away amusedly and moved on. Or so I thought.

Before this, I had already made up my mind that I wanted to buy a Charvel. It seemed like the next step. I grew up with a Stratocaster, but it was never anywhere near optimal for the mix of Metallica and Pantera I grew up playing.

Later that week, I ended up ordering a hardtail Charvel Pro-mod So-Cal from Sweetwater after spending a few hours looking at the specs of various Charvel hardtails. I was super excited.

But then I was like, "Oh shit, Fender got me. Well played, gents."

To reiterate, I had already decided I was going to buy a hardtail Charvel "when I can afford it." But I couldn't could help but feel like the buy was algorithmically prompted/nudged. And of course, I was right! Fender's Snapchat ad in my inbox was indicative of a bigger problem, the AdTech industry had marked me as prey to sell guitars to!

In general, I try to use browsers that let you block ads. Unfortunately, Chrome on Android doesn't let you do this (desktop Chrome keeps getting enshittified too, BTW), and the OS seems to force you into Chrome-based viewing of web content in certain situations. But when I'm not in Firefox, this is what things can look on my phone:

How the fuck is a guy supposed to stay up to date on Wyoming's senate seats when his screen is covered in all this bullshit? Sheesh.

These recommendations are good. Almost too good (I already own 1 of those guitars and the algo hasn't figured out yet that an SSS or 2pt Charvel is what I am missing now). Dare I say, another example of overfitting. Obviously, the algorithm is pretty finely tuned to show me Superstrats in a particular price range, though.

I have encountered other examples like this that deserve their own articles, too. After being blasted with a gazillion Reverb ads in youtube shorts, I finally let one play long enough to hear Juan Alderete (of The Mars Volta, Racer X, etc.) get to the part where he talks about why he needs to sell his gear: he is cognitively disabled due to a bike accident where he went over the handlebars (with a helmet, he added). This struck me to the core because I've survived several very nasty ski crashes and this year started to become involved in disability advocacy.

Once I saw that Reverb ad that morning, it felt inevitable that I was going to buy something from him. I ended up buying an "Aluminum Green Knob Pedal" that was described as "An odd pedal." And I am glad I did! It has a second life on my pedalboard, and I would suggest that I literally needed to get bombarded with that many Reverb ads (or at least the starts of them) over the past 6 months for them to have a chance at me hearing one I "needed" to hear in a receptive way. At this point, I was otherwise against using Reverb since what you don't know can hurt you a lot when you are buying used gear.

I am sure this pedal will get its own article here at some point!

I detest surveillance capitalism, but it would seem that the AdTech industry had actually helped me buy musical gear that I am very happy I own. What a contraction!

Is it ethically, culturally/artistically 'bad' to train models on gear?

Although there are some talented artists out there parroting "AI is just a tool" lines (I don't hate that position as much as some people... but it is too reductive), I think largely the art community (including music but much more broadly too) has formed a consensus that generative AI trained on art (of others) is inherently problematic.

But I think "AI has no place in art" is an equally un-nuanced take.

There have been times where I've entertained the idea of having a generative AI drummer that I could give natural language instructions to. Wouldn't it be great, I've wondered, to be able to type something like "black metal blast beats evoking the ferocity of a winter storm" and have a drum performance generated that I could start tremolo picking along to? How is it any different from drum machines?

Well, I think my hypothetical AI drummer differs from a drum machine in terms of user intention and performance. Art is simultaneously about pushing the limits of what's possible while accepting the limits and messy imperfection of human artists. A drum machine is something that an artist/producer programs within his skills as a part of making music. With my AI drummer hypothetical, we are no longer trying to actually control the drum performance. It is chasing the vibe of having a drummer, but the generated "performance" can't have any possible emotion or experience behind it. You are purely using it to replace other musicians for the stuff you don't know how to do, which is kind of weaksauce (yes a drum machine also does this, but it has the expression aspect I mentioned earlier). So, I think think using AI to replace human musical composition and performance is disgusting. Suno explicitly markets their crappy vibe plagiarism service with "give me instant results and replace musicians who know things I don't" messaging. Yuck. Creeps.

But what about training models on gear? This is not some science fiction question. The open-source Neural Amp Modeler (NAM) project "uses deep learning to create models of guitar amplifiers and pedals with state-of-the-art accuracy." It is free to download the plugin for your DAW if you are a guitarist, and it's free for developers to integrate the code from various parts of the project (there is also the model-training side too).

If you are in the "AI has absolutely no place in art" camp, it seems pretty obvious that you should detest NAM. But that position doesn't really stand on solid ground.

In contrast with my earlier "AI drummer," I think NAM usage is more akin to developing sample packs with your recordings of your physical/acoustic instruments. It is pretty uncontroversial that you should be allowed to record your own instrument and then use the samples in chopped-up ways in software like a DAW or an audio programing language (such as TidalCycles). Using our own instruments to generate data for digital workflows and software is an established tradition, so we're clearly not talking about a "stolen data" situation . And this raises an important question for me: if you are are anti-AI in general, is there any point at which "AI" becomes ethical if the data is sourced ethically, or are you arbitrarily against all forms of neural networks and machine learning? It's rather luddite to be against forms of math, no?

Theoretically, you could argue that modeling/profiling tech (which can include more deterministic applications too) cannibalizes the amp market, but I'm nowhere near convinced that's the real story.

First, as far as I'm aware, there has never been any court to ever have a problem with an amp or pedal sounding like another. The guitar world has a long history of products copying each other's circuits and sound and iterating on them. If you want to invent some contrived IP rights for amp manufacturers, go ahead and do it.

Tube amps are not terribly practical. If you want to have 20 different tube amps and constantly be trying different tubes on all of them and actively replacing suspected duds, you will quickly run out of money. I'm not going to pretend to be an expert in owning tube amps (I do have a Marshall combo that I love), but it seems like they might be best enjoyed by taking the time to get really good at dialing in one main amp (or two) and understanding which tubes, microphones, and eq settings you might want to try next in your tone-chasing quest (I think adding the Boss EQ-200 to my FX loop opens up my Marshall DSL5CR so much).

Maybe, instead of killing tube amps, modeling tech will make guitarists really well-researched buyers of tube amps. After all, if you can try simulated versions of a thousand different amps all at home, you will probably end up with a decent idea of what you are looking for when you finally shop for your 1 main tube amp that you are going to own for a while. Maybe, in the long run, more people will get deep enough into guitar to realize they want a tube amp. In my guitar childhood, I was totally ignorant to the entire world of tone outside my Fender Frontman.

I think there is also a sort of "right to repair" aspect to this too where I feel it is reasonable to digitally clone my physical setup for practicing and recording. What happens if my tubes start to go microphonic or noise but I can't immediately replace them? What happens if tube supply chains break down altogether? Of course, Boss doesn't want us thinking like this... they want to sell us a Roland cloud subscription for the right to use their pedals in a DAW. $200 annually at the time of writing. Sheesh. And it would seem that there's historical preservation value in modeling different rare amps as well.

Anyway, I think NAM is great, and I intend to build products with it. That is the beauty of the open-source world.

There are definitely ways we can go too far in using ML in our signal chains. Imagine pickups that use AI to predict what you meant to play and adjust your signal to be perfectly in tune and without any mistakes. This would be a bridge too far! Yet, I think something like this will exist eventually. Probably sooner than we think. Hopefully us guitarists, in the aggregate, can reject products like this that totally destroy the authenticity of the experience of playing guitar.

I don't think we want tube amps to die out either. l don't even know if it would make a difference audible in terms of the recordings we would hear in our world, but practicing or playing on stage with a tube amp roaring near you is an experience that I don't think we want to lose. The amp modding hobby might go away with time, but the modding culture is shifting more towards making your own IRs and plugins now that we have modular digital

Can musicians "bond" with software like with guitars and amps?

I think the whole point of playing guitar is to try to play music that resonates emotionally with yourself and others. When guitarists buy a guitar or amp, we are seeking an emotional connection with it. We want the gear to help us express our emotions and even amplify them and cause feedback. If you've looked at enough reviews in online guitar stores, you've seen something along the lines of, "This is a beautiful guitar but I just never could form any sort of real connection with it. 3 stars." Musical gear is inherently emotional, so we naturally have to wonder if we can bond with software for music in the same way. Of course amp sims have portability and repeatability/consistency benefits. Of course you can get recordings that sound great on them (and I don't think I can tell any sort of a difference in my real mixes, but I am still pretty new to DAWs and mixing). But can we bond with these sims like we can with a great tube amp?

I have no idea honestly... I wouldn't say I've been playing on them long enough to say. But I have bonded emotionally with software as a musical instrument. 100%.

Tidalcycles (which I mentioned earlier) is a musical programming language that I definitely developed a connection with as I used it to release an EDM EP a few years back. It made me really happy to be able to explore music ideas like that and keep chaining functions together to experiment with the sound. I have vaguely the same kind of emotional connection to that programming as I do to my guitars, so I absolutely believe that the same thing happens with amp sims.

Closing thoughts

Like it or not, software is part of the modern guitarist's signal chain. That means that sometimes we have to apply thinking from software engineering and software procurement to our guitar tone.

There will always be some musical gear that is inherently consumable due to physical constraints. But poorly developed and licensed software could make all your gear inherently consumable. Can you ever really own your sound if support is going to be phased out for your software in the medium term? You definitely can't own your sound if your sound depends on subscriptions because that is the deal with subscriptions.

The guitar community has benefitted enormously from social media and the internet. Compared to the world in which I started to learn guitar (mid 2000s), access to quality information is seemingly infinite, but the flip side is that attention can be infinitely fleeting and diffuse.

Home recording has boomed over the past decade. While I'm not going to sit here and pretend that it is bad for music to become a more affordable part of people's lives, there is definitely a balance that needs to be found. Otherwise, we all just record music at home by ourselves. "Services" like Suno could accelerate that trend. We shouldn't want to optimize out the friction of making music with others.

I am not pro-modeling or anti-modeling, pro-software or anti-software (in musical performance and production), pro-AI or anti-AI (in the most general sense of the word). What I value is music that has human authorship, emotion, intent, mastery, and even imperfection ("perfect" djent is not interesting and never was, IMO). I suspect that many guitarists would agree even if social media algorithms don't properly incentives those traits in the "content" that we end up consuming. There are ways for software, even AI, to be beneficial or antagonistic to those values. The guitar community has always had debates about tradition and technology. I think this is healthy and can help us survive and thrive amidst what is happening (technology that threatens the nature of art itself).

As I was writing this article (originally for my personal website), I kept coming up with more and more angles that related to these ideas. And many of them were worthy of their own articles. So I chose to start this site so I can keep sharing my thoughts and various music-related products that I am cooking up. I am excited about the future, but I am also praying true art (created by emotionally-invested humans) can survive headwinds it faces.