Tech news has an unusual property: the people building the thing post about it directly, often before any publication covers it. A model release, an outage, a licence change, a shutdown. The primary source is frequently a person rather than a company, and following the right people gets you the news and the reaction to it in the same place.
Practitioners
@AndrejKarpathy posts infrequently and it is almost always worth reading. Explanations of how systems actually work, written by someone who built them, without the layer of translation a journalist has to add.
@simonw covers what is genuinely new in tooling and what merely sounds new, and tests things rather than describing them. Useful precisely because he publishes the failures alongside the successes.
@levie writes about enterprise software from the inside, with more humour than the category usually allows.
Company accounts
@OpenAI and @AnthropicAI announce releases first on their own accounts, so following them removes the lag between an announcement and the coverage of it. Both are marketing channels, which is fine as long as you file them that way rather than as reporting.
Put company accounts in a different lane from journalists. Mixing them makes it easy to read a launch post as though it were an assessment.
Journalism and analysis
@TechCrunch and @verge for volume and coverage breadth. Neither is fast in the way a practitioner is fast, but they catch the stories nobody posted about themselves, which is most of the funding and regulatory news.
@karaswisher for industry politics and the relationships behind the announcements. Opinionated, which is the point.
@profgalloway for the business argument, usually framed more sharply than the underlying numbers strictly support. Worth reading as a provocation rather than as analysis.
Hardware
@MKBHD remains the most reliable early read on consumer hardware, particularly on whether a device is pleasant to use rather than impressive on a spec sheet. Low posting volume, high signal.
The problem with a tech timeline
Tech is the category where the algorithm hurts most, because engagement and importance are almost inversely related. A confident thread about what a release means will outperform the release announcement itself, and a contrarian take will outperform both. Sort by engagement and you systematically see commentary before news.
The other issue is speed of decay. A model release matters for about a day. A timeline that resurfaces popular posts from yesterday is showing you a stale version of a field that moves weekly.
A layout
| Lane | Accounts | Why separate |
|---|---|---|
| Source | @OpenAI, @AnthropicAI, company accounts | Announcements, unfiltered |
| Practitioners | @AndrejKarpathy, @simonw, @levie | Low volume, easily buried |
| Press | @TechCrunch, @verge, @karaswisher | High volume, would swamp the rest |
The separation matters more here than in other categories because the posting rates differ so widely. A publication posts thirty times a day. Karpathy might post twice a week. In one merged feed the second is invisible, and the second is the one you actually wanted.
Running it as a passive ticker on a second screen suits the way tech news arrives, which is in bursts around releases and conferences rather than evenly.
Common questions
Has tech discussion moved off X?
Partly. A visible share of the research community now posts to Bluesky or Mastodon as well, though usually in addition rather than instead. Company announcements and the wider industry conversation still happen on X.
Should I follow individual researchers?
If you work in the field, yes, and pick them from papers you have actually read rather than from follower counts. For general awareness, a few well chosen practitioners cover more ground than a long list.
How do I keep up without reading all day?
Stop trying to read it and let it pass you instead. A ticker running in peripheral vision means the significant items register and the rest does not, which is the correct amount of attention for most of it.
A short list
@AndrejKarpathy, @simonw, @OpenAI, @AnthropicAI, and @verge covers releases, informed assessment, and the news nobody announced about themselves. Add a hardware account if that is your interest, and leave the commentary accounts for when you have time to argue with them.
TT Staff
The TweetTicker Editorial Team provides real-time insights for high-velocity data environments.