• 2024-01-19

    I can’t remember any time when more spaces for innovation and entrepreneurship were wide open than now

    Lenin supposedly said:

    There are decades where nothing happens, and there are weeks where decades happen.

    It’s the same in technology.

    I came to Silicon Valley in the mid-90’s, just in time to see the dot-com boom unfold. Lots happened very quickly in that time. There were a few more such periods of rapid change since, like when centralized social media got going, and when phones turned into real computers. But for many years now, not much has happened: we got used to the idea that there’s a very small number of ever-larger tech giants, which largely release incremental products and that’s just that. Nothing much happens.

    But over the last year or so, suddenly things are happening again. I think not only are the spaces for innovation and entrepreneurship now more open than they have been for at least a decade or more; it’s possible they have never been as open as they are now.


    • Everybody’s favorite subject: machine learning and AI. I don’t believe in much of what most people seem to believe about AI these days. I’m not part of the hype train. However, I do believe that machine learning is a fundamental innovation that allows us to program computers in a radically different way than we have in the past 50 and more years: instead of telling the computer what to do, we let it observe how it’s done and have it copy what it saw. Most of what today’s AI companies use machine learning for, in my view, is likely not going to stand the test of time. However, I do believe that this fundamentally different way of programming a computer is going to find absolutely astounding and beneficial applications at some point. It could be today: the space for invention, innovation and entrepreneurship is wide open.

    • The end of ever-larger economies of scale and network effects in tech. The dominant tech companies are very close to having pretty much all humans on the planet as customers. The number of their users is not going to double again. So the cost structure of their businesses is not going to get reduced any more simply by selling the same product to more customers, nor is the benefit of their product going to grow through growing network effects as much as in the past. It’s like they are running into a physical limit to the size of many things they can do. This opens space for innovation and successful competition.

      Most interesting, it allows the creation of bespoke products again; products that are optimized for particular markets, customer groups and use cases. Ever noticed that Facebook is the same product for everybody, whether you are rich or poor, whether you have lots of time, or none, whether you are CEO or a kid, whether you like in one place or another, whether you are interested in sports or not and so forth? It’s the same for products of the other big platform vendors. That is a side effect of the focus on economies of scale. All of a sudden, increased utility for the user will need to come from serving their specific needs, not insisting that all cars need to be black. For targeted products, large platforms have no competitive advantages over small organizations; in fact, they may be at a real disadvantage. Entrepreneurs, what are you waiting for?

    • The regulators suddenly have found their spine and aren’t kidding around, starting with the EU.

      • The Apple App Store got in the way of your business? They are about to force the App Store open and allow side loading and alternate app stores (although Apple is trying hard to impede this as much as possible; a fight is brewing; my money is on the regulators).

      • The big platforms hold all your data hostage? Well, in many jurisdictions around the world you now have the right to get all copy of all your data. Even better, the “continuous, real-time access” provision of the EU’s Digital Markets Act is about to come into force.

      • The platforms don’t let you interoperate or connect? Well, in the EU, a legal requirement for interoperability of messaging apps is already on the books, and more are probably coming. Meta’s embrace of ActivityPub as part of Threads is a sign of it.

      Imagine what you can do, as an entrepreneur, if you can distribute outside of app stores, use the same data on the customer that the platforms have, and you can interoperate with them? The mind boggles … many product categories that previous were impossible to compete with suddenly are in play again.

    • Social networking is becoming an open network through the embrace of ActivityPub by Meta’s Threads. While nobody outside of Meta completely understands why they are doing this, they undoubtedly are progressing towards interoperability with the Fediverse. Whatever the reasons, chances are that they also apply to other social media products, by Meta and others. All of a sudden competing with compelling social media application is possible again because you have a fully built-out network with its network effects from day one.

    • Consumers know tech has a problem. They are more willing to listen to alternatives to what they know than they have in a long time.

    • And finally, 3D / Spatial Computing a la Apple. (I’m not listing Meta here because clearly, they don’t have a compelling vision for it. Tens of billions spent and I still don’t know what they are trying to do.)

      Apple is creating an an entirely new interaction model for how humans can interact with technology. It used to be punch cards and line printers. Then we got interactive green-screen terminals. And then graphics displays, and mice. That was in the 1980’s. Over the next 40 years, basically nothing happened (except adding voice for some very narrow applications). By using the space around us as a canvas, Apple is making it possible to interact with computing in a radically different way. Admittedly, nobody knows so far how to really take advantage of the new medium, but once somebody does, I am certain amazing things will happen.

      Again, an opportunity ripe for the taking. If it works, it will have the same effects on established vendors as the arrival of the web had on established vendors: some managed to migrate, or the arrival graphical user interfaces on the vendors of software for character terminals; most failed to make the switch. So this is another ideal entrepreneurial territory.

    But here’s the kicker: what if you combined all of the above? What can you build if your primary interaction model is 3D overlayed over the real world, with bespoke experiences for your specific needs, assisted by (some) intelligence that goes beyond what computers typically do today, accomplished by some form of machine learning, all fed by personal data collected by the platforms, and distributed outside of the straightjacket and business strategies of app stores?

    We have not seen as much opportunity as this in a long time; maybe ever.

  • 2023-12-08

    Meta/Threads Interoperating in the Fediverse Data Dialogue Meeting yesterday

    I participated in a meeting titled “Meta’s Threads Interoperating in the Fediverse Data Dialogue” at Meta in San Francisco yesterday. It brought together a good number of Meta/Threads people (across engineering, product, policy), some Fediverse entrepreneurs like myself, some people who have been involved in ActivityPub standardization, a good number of industry observers / commentators, at least one journalist, and people from independent organizations whose objective is to improve the state of the net. Altogether about 30 people.

    It was conducted under the Chatham House rule, so I am only posting my impressions, and I don’t identify people and what specific people said. (Although most attendees volunteered for a public group photo at the end; I will post a link when I get one. Photo added at the bottom of this post.)

    For the purposes of this post, I’m not going to comment about larger questions such as whether Meta is good or bad, should be FediBlock’ed immediately or not; I’m simply writing down some notes about this meeting.

    In no particular order:

    • The Threads team has been doing a number of meetings like this, in various geographies (London was mentioned), and with various stakeholders including the types of people that came to this meeting, as well as Fediverse instance operators, regulators and civil society.

    • Apparently many (most?) invitees to these meetings were invited because other invitees had been recommending them. I don’t know whether or what kind of future meetings like this they are planning, but I’d be happy to pass along names if we know each other and you get in touch. Thanks to – you know who you are – who passed along my name.

    • The Threads team comes across as quite competent and thoughtful at what they do.

    • On some subjects that are “obvious” to those of use who have hung around open federated systems long enough like myself, many attendees seemed strangely underinformed. I didn’t get the impression that they don’t want to know, but simply that integrating with the “there-is-nobody-in-charge” Fediverse is so different from other types of projects they have done in the past, they are still finding their bearings. I heard several: “A year ago, I did not know what the Fediverse was.”

    • Rolling out a large node – like Threads will be – in a complex, distributed system that’s as decentralized and heterogeneous as the Fediverse is not something anybody really has done before. It’s unclear what can go wrong, so the right approach appears to be to go step-by-step, feature by feature: try it, see how it works in practice, fix what needs fixing, and only then move on to the next feature.

    • That gradual approach opens them up to suspicions their implementation is one-sided and entirely self-serving. I guess that can’t be avoided until everything is deployed they publicly said they will deploy.

    • While there are many challenges, I did get the impression the project is proceeding more or less as planned, and there are no major obstacles.

    • Everybody knows and is aware Meta brings a “trust deficit” to the Fediverse. The best mitigation mentioned was to be as transparent as possible about all aspects of what they plan and do.

      I think that’s a good approach, but also that they can do far more on transparency than they have so far. For example, they could publicly share a roadmap and the engineering rationale for why the steps they identified need to be in this sequence.

    • There are many, many questions on many aspects of the Fediverse, from technical details, to operational best practices, to regulatory constraints and how they apply to a federated system. The group generally did not know, by and large, how to get them answered, but agreed that meetings like this serve as a means to connect with people who might know.

      I think this is a problem all across the Fediverse, not specific to Meta. We – the Fediverse – need to figure out a way to make that easier for new developers; certainly my own learning curve to catch up was steeper than I would have liked, too.

    • Many people did not know about FediForum, our Fediverse unconference, and I suspect a bunch of the meeting attendees will come to the next one (likely in March; we are currently working on picking a date). Many of the discussions at this meeting would have been right at home as FediForum sessions, and while I am clearly biased as FediForum organizer, I would argue that doing meetings like this in an open forum like FediForum could help substantially with the trust deficit mentioned above.

    • There’s significant interest in the Fediverse Test Suite we just got funding approval for from the EU’s NGI Zero program. There’s general agreement that the Fediverse could work “better”, be more reliable, and more be comprehensible to mainstream users, if we had better the test coverage than the Fediverse has today. This is of interest to all new (and existing) developers.

    • There was a very interesting side discussion on whether it would be helpful for Fediverse instances (including Threads) to share reputation information with other instances that each instance might maintain on individual ActivityPub actors for its own purposes already. Block lists as they are shared today are a (very primitive) version of this; a more multi-faceted version might be quite helpful across the Fediverse. This came up in a breakout group discussion, and was part of brainstorming; I didn’t hear that anybody actually worked on this.

    • When we think of privacy risks when Meta connects to the Fediverse, we usually think of what happens to data that moves from today’s Fediverse into Meta. I didn’t realize the opposite is also quite a challenge (personal data posted to Threads, making its way into the Fediverse) for an organization as heavily monitored by regulators around the world as is Meta.

    • There was very little talk (not none, but little) about the impact on regulation, such as the “continuous and real-time access” provision in the EU’s Digital Markets Act and whether that was a / the driver for Fediverse integration.

    • There was very little discussion on business models for Threads, and where exactly ads would go. For example, would ads just stay within the Threads app, or would they inject them into ActivityPub feeds, like some companies did with RSS back in the days? Of course, should that happen, as a non-Threads Fediverse user, one can always unfollow; there is no way for them to plaster non-Threads users with ads if they don’t interact with Threads accounts.

    • I came away convinced that the team working on Threads indeed genuinely wants to make federation happen, and have it happen in a “good” way. I did not get any sense whatsoever that any of the people I interacted were executing any secret agenda, whether embrace-and-extend, favoring Threads in some fashion or anything like that. (Of course, that is a limited data point, but I thought I convey it anyway.)

    • However, the meeting did not produce a clear answer to the elephant-in-the-room question that was raised repeatedly by several attendees in several versions, which is some version of: “All the openness with Threads, namely integration with the Fediverse, supporting account migration out from Threads etc, is the opposite of what Facebook/Meta has done over its history. What has fundamentally changed so that you now believe openness is the better strategy?” And: “In the past Facebook was a far more open system than it is today, you gradually locked it down. What guarantee is there that your bosses won’t follow the same playbook this time, even if you think they won’t?”

      Personally I believe this question needs a better answer than has been given publicly so far, and the answer needs to come from the very top of Meta. The statement must have staying power beyond what any one executive can deliver.

    I left the meeting with far more questions than I could get answered; but nobody wanted to stay all night :-)

    My gut feel is that it is safe to assume they will do a reasonably fair, responsible job with a gradual rollout of federation for at least the next year or two or such. Beyond that, and in particular if it turns out creators with large follower groups indeed move off Threads at a significant rate (one of the stated reasons why they are implementing Fediverse support as creators have asked for this), I don’t think we know at all what will happen. (I’m not sure that anybody knows at this point.) And of course, none of this addresses the larger issues that Meta has as an organization.

    In the hope this summary was useful …

    Group photo (opt-in)

    (Source. Thanks tantek.com for initiating this.)

  • 2023-08-16

    The Peer Computing Architecture

    I used that term frequently a long time ago, before the centralizing platforms took over all of technology. I believe it’s time to dig it back out, because for the first time in over a decade, perhaps even two, there’s real uptake in this computing architecture with protocols such as ActivityPub and Matrix.

    My definition is at peercomputing.org.

    Feedback appreciated.

    (Historical note: there was a time I called it the “four-point architecture”. My old blog tells me I wrote about it in, gasp, 2005. That means 18 years of walking in the desert, and only now do the doors open up again.)

  • 2023-07-21

    Meta’s plummeting Threads engagement numbers are good news for the Fediverse

    The Wall Street Journal reports that:

    User engagement on Threads has continued to fall after an initial surge in sign-ups … the number of daily active users [has fallen] to 13 million, down about 70% from a July 7 peak, according to estimates from market intelligence firm Sensor Tower. The average time users spend on the iOS and Android apps has also decreased to four minutes from 19 minutes.

    That’s quite a bit different than the headlines of “fastest app growth in history” and “more than 100 million accounts in a week”. It was always clear that only a fraction of those initial users were going to stick around, but more concerning – from Meta’s point of view – is that decrease in time spent with the app by the users that do continue to use the app: from 19 minutes to four minutes is a very significant decline.

    From the perspective of the (open) Fediverse, however, I think this is very good news:

    • At these numbers, Threads is still significantly larger than the rest of the Fediverse combined, but far less so than it looked like a week ago. There was a possibility that Meta would simply drop their plans to implement ActivityPub, or just do a very rudimentary job, because why would you spend much attention on this if all you can connect to is a network 1% the size of your own. At 10-20%, as it looks like right now, that argument is very different.
    • Rather than leaning back, and celebrating its success, Meta now needs to work hard to compete with Twitter. (And Twitter will have to compete back.) That is good for the world in general, and indirectly good for the Fediverse as the “market” for text-based microblogs is suddenly competitive again. People realize they have at least two (well-marketed) choices; it’s not a foregone conclusion that “Twitter owns this”. If there are two choices, maybe there are three of four, including the non-Meta Fediverse.
    • And perhaps, as a side effect of all of this, some of the dollars being spent may find their way into the Fediverse. Certainly Meta is now even more incentivized to highlight Threads’ differences with Twitter, and key to that is ActivityPub. I would think that should make them more willing, rather than less, to spend a bit of money on Fediverse infrastructure, like a test suite, that benefits the entire network.
    • Of course, it’s also good news for everybody who had been concerned that Threads would overwhelm the culture of the Fediverse the same way AOL overwhelmed in the culture of Usenet in the nineties.

    (Note: I am in favor of Meta implementing ActivityPub in Threads; in fact I would like it to implement it in all of its social media properties, and messenger services like WhatsApp – or perhaps a similar protocol like Matrix. I believe the same for non-Meta apps like LinkedIn, YouTube etc.

    IMHO, the ability to interoperate across platforms is a major disinfectant, and can be a major factor in reducing product lock-in, which can lead to better products, more choices, and fewer “forced” policies like privacy invasions, because users have the ability to go elsewhere. I understand those Fediverse users who do not want to touch anything that Meta does and sympathize; the good news is that you, as a user, have unprecedented control in the Fediverse over who you do and do not want to interact with, and more power to you if you exercise that control.)

  • 2023-06-28

    ActivityPub everywhere?

    In previous posts, I argued the EU’s Digital Services and Digital Markets Acts are the most likely reason why Meta is implementing ActivityPub in its upcoming Twitter competitor “P92”, and that ActivityPub is really the only game in town that meets Meta’s requirements for interop for “P92”, as they require protocols that are proven to work and governed in a lawyer-vetted process by a legitimate standards organization.

    But if that line of reasoning is correct for for why Meta would pick ActivityPub to meet its interoperability obligations under EU’s new rules, it is likely also correct for all other social networking products, by Meta and others, that are required to interop by the same new rules, notably:

    • Facebook (also Meta)
    • Instagram (also Meta)
    • Snapchat (Snap)
    • LinkedIn (Microsoft)
    • Twitter
    • and perhaps even YouTube (Google).

    (The full list is here although I think the EU puts them in different markets. Update 2023-06-30: Actually, this is the wrong list; the list of “gatekeepers” will only be established in a couple of months under the Digital Markets Act. Chances are it’s very similar to this list of “Very large on-line platforms”, however.)

    Let me say this again. The product teams of all of those products are very likely asking themselves the very same question right now that Meta asked itself: what are we going to implement to meet our interop obligations, and when are we going to do it?

    I believe most will arrive at the same conclusion: they will look for protocols that can demonstrably work for these kinds of interop requirements, and that are maintained by a well-understood standards organization. And will find that ActivityPub is the only game in town that meets those requirements.

    (It is possible that one or two of the organizations above come up with a “spoiler” strategy developing or picking an alternate protocol. To make up a random example, say, Microsoft might buy Bluesky and declare it to be its interop architecture. They would have to turn that protocol over to a standards org, however, before its competitors would even consider it. And that might not gain them much; sticking with ActivityPub is simpler.)

    (It is also possible that some of the above organization will choose to fight the EU in court; Twitter comes to mind. I’m not a lawyer, I don’t know how likely this might succeed; however, the EU has shown to be tenacious and unstoppably moving forward in this direction, so I wouldn’t bet on that strategy if I were them.)

    Which means a number of things – starting with the ActivityPub standardization process in the W3C:

    • Meta, Microsoft and others will suddenly show up in the W3C ActivityPub community. It’s easy for them to do, they already have been spending significant amounts of time in the W3C.

    • They will show up with requirements currently considered out of scope by the W3C SWICG, but driven by the need that the entire stack is standardized that is necessary to enable interop between, say LinkedIn and Facebook. If they decide that real interop is actually necessary – my premise – then an “assembly required” standard (like ActivityPub is today) is not sufficient for them.

    • Therefore the official Social Web Working Group will need to be re-constituted, and its charter will need to be broader than it was before.

    • There will be loud screaming in the existing W3C social community. (Because many of the members of the current SWIGC are extremely strongly in opposed to surveillance capitalism – the prevailing business model of most of the companies above – or, in some cases, that any kind of business has any role in desirable forms of social media at all.) An accommodation will have to be found, as the W3C as an organization wants to serve all willing participants in a standards process, not just those with a certain value system.

    But there more intriguing, and exciting possibility is this:

    • When everything is said and done and my line of reasoning is reasonably correct, interoperability via ActivityPub indeed might be “everywhere”: Facebook, Instagram, P92, LinkedIn, Snapchat, even Twitter.

    Social media would end up being a radically different beast. A potentially much better, beast as I believe. Ponder this.

    Updated 2023-06-30:

    • Added Digital Markets Act in addition to the Digital Services Act; it’s really both and the intent behind them that drive this.