lukev

16 ms·

301.
by lukev 2y ago
Yeah, reviews are inherently social. I’m waiting for someone to build a review platform on top of ATProto (bluesky).
302.
by lukev 2y ago
What does this actually mean? Most MMOs don't let you convert your in-game currency back to real-world money. For those that do, is there actually a regulatory framework around that?
303.
by lukev 2y ago
Just curious, what kind of Elon scandal do you think could move the needle at this point? Anyone capable and willing to observe his behavior has probably already acted on it. What lines are left to cross?
304.
by lukev 2y ago
love this... would love to discuss more (but not quite sure who you are)... hit me up via other channels?
305.
by lukev 2y ago
They are not. It’s quite difficult to prompt a generative model to get details correct, and it takes a ton of tries if you can make it work at all. Practically impossible if the model wasn’t trained with data that makes a distinction in wha
306.
by lukev 2y ago
A) That article is from was 3 months ago, the network has grown almost 50% since then. B) You can set up your own PDS, right now. If you're comfortable with Docker it should take you under an hour: https://github.com/b
307.
by lukev 2y ago
The protocol is fully open and adoption is picking up rapidly. You can easily host your own data server in your preferred jurisdiction if that is important to you.
308.
by lukev 2y ago
I think there's something here. I'm not sure it's a bad thing. A negative framing might be that there's a sort of elitism. A more positive framing might be that it simply reflects a tension between different, valid views
309.
by lukev 2y ago
Nice! I'm waiting for (/ thinking of building) something similar for #atproto/Bluesky. You could even use OAuth to allow users to add responses inline, which would then also show up in the Bluesky feed (if you wanted), with a
310.
by lukev 2y ago
I don't know what you're talking about, 2 new versions of Clojure have come out since then, both of which have more substantive changes than the previous couple versions.
311.
by lukev 2y ago
[flagged]
312.
by lukev 2y ago
If I block you, then my client will prevent me from seeing your records. There's nothing at the protocol level that prevents you from seeing my records, even when blocked, because all records are public. Anyone (even without an account
313.
by lukev 2y ago
I haven't looked at how Threadsky does it, and I don't want to weigh in on the merits of different approaches. But blocking in ATProto happens at the application (AppView) layer. A different application can absolutely make differe
314.
by lukev 2y ago
XKCD #538 strikes again, and definitely extends to forcing people to lie about algorithms and possible backdoors. I don't think, however, that this means we need to give up on crypto entirely. Just... be aware of the threat model for w
315.
by lukev 2y ago
Even elected officials are not allowed to do stupid or insecure things! I seem to remember something about a quite famous incident involving a private email server. Which boggles the mind when you compare it to what's going on right no
316.
by lukev 2y ago
They do not. The majority of the country either did not vote, or voted against Trump. Most of those who did vote for Trump did so on the basis of immigration, economic or the culture war -- not so he could dismantle the federal government a
317.
by lukev 2y ago
This is absolutely the problem. But there is a line of sight; namely, combining LLMs with existing semantic data technologies (e.g, RDF.) This is why I'm building a federated query optimizer: we want to let the LLM reason and formula
318.
by lukev 2y ago
This is a key point. Given current overall levels of economic productivity, we should all be working 3-4 day weeks, at most. To whom do the benefits of all this newfound efficiency accrue?
319.
by lukev 2y ago
Well, I find myself forced to agree with you there. I think these models are tremendously useful for data processing, and that chasing "reasoning" and the production of artifacts for human consumption are entirely red herrings.
320.
by lukev 2y ago
I don't disagree that "is it useful" is the important question. The amount of money being invested is very clearly disproportionate to the current utility, and much of it is obviously based on the premise that LLMs can (or wi
321.
by lukev 2y ago
If you have evidence of how the secure enclaves on mobile devices are compromised, you should share those details. You can also believe that there is an industry-wide conspiracy in which everything is backdoored. But that's a philosoph
322.
by lukev 2y ago
And (as a complete side note) it has all worked out very poorly for Twitter, as a revenue-generating platform. No reason to think it will be better when applied to the federal government.
323.
by lukev 2y ago
The courts are just now beginning to order injuctions and restraining orders, for the stuff that happened last week. The process seems to lag by 2-3 business days. So hopefully we'll be seeing a lot more this week. How the administrati
324.
by lukev 2y ago
But that has to be an act of congress. Congress creates and funds government departments. The president can't just dismantle them by fiat. That he's (apparently) tasking people to do so who don't even have the requisite exper
325.
by lukev 2y ago
Network effects, lack of alternatives, and the difficulty of collective action are fully sufficient to explain this.
326.
by lukev 2y ago
Kind of, yes. It is a completely coherent position to dislike and distrust a platform, and still use it daily because that's where your friends are. Network effects are powerful. It's entirely possible for a platform to be popular
327.
by lukev 2y ago
SEO, marketing, and content generation are fundamentally parasitic bullshit. Unlike the initial .com wave, there is value to some companies, but basically no value to end consumers or (ultimately) humanity.
328.
by lukev 2y ago
All well and good, but how well will it work if the pattern continues that the best open models are less than a year behind what OpenAI is doing? How long can they maintain their position at the top without the insane cashflow?
329.
by lukev 2y ago
The inaccuracy is a feature, not a bug. Police have always wanted the power to arrest who they want, when they want. AI, like drug-sniffing dogs before it provides them the plausible deniability to do so.
330.
by lukev 2y ago
One thing I'll note is that I'm inclined to trust the poll of engineers who have actually been using this stuff for real work, as opposed to a N=1 example of it performing well (outside the context of a real software project) and

More ›