To get more productive reading time out of X/Twitter, rather than rely on the algorithm, I'm going to use Hermes Agent to give me a daily morning report. ``` Every morning at 6 a.m. 1. Use `xurl --app hermes-agent-xurl search "is:verified has:links agent" -n 12` to fetch a list of articles as JSON 2. Produce a report with a 1 or 2 line summary per article Deliver the report on Slack without preview. ``` Coming from a coding background, one needs to become more comfortable with prompting Hermes in this way. Examining `~/.hermes/cron/jobs.json`, one will be a little taken aback at first, as Hermes rewrites the prompt. It removes certain details and it adds others. What the human believed to be important may not be what the agent believes to be important, so one develop a degree of confidence that the agent gets this correct. I certainly cannot argue against the result, which turned out to be correct. That's where the confidence ultimately comes from. I expect this technique to be fruitful. I will use jobs throughout the day to deliver exactly what I want to read, and this will replace the algorithm-manipulated doom scrolling.

Twitter's search does not have query filters for articles. Here is a rough approximation that works. Add the following filters to your query. filter:verified filter:links For example, agent memory filter:verified filter:links Posting articles is a premium feature offered to verified users. Usually, an article contains links to a web site or git project. That's why this query pattern works for articles.

Apparently the version 2 api doesn't support this same query syntax as the user interface. This works: xurl search "is:verified has:links agent" -n 20

Attention is all you need, they said. Until ghetto ontology leads to such ambiguity and imprecision that no one knows what their peers are talking about in different contexts.

To be clear, we are not talking about simple loops in a procedural language. while : ; do echo 'hello world's; done The topic is about orchestrating large units of work (business outcomes). x.com/i/status/2063995010643…

I can empathize with non-native English speakers struggling to speak English early in their journey to become proficient. I cannot understand how someone who works professionally in English to not improve their speech skills. Perhaps when they are immersed in a community of fellow non-native speakers, and they all struggle with heavy accents and similar patterns of poor speech, they don't learn. That's the problem with offshore teams. I know this is not a problem on an individual level. I have seen native Chinese speakers immigrate from Hong Kong, and they end up speaking English perfectly without any accent. Females tend to be better at this than males. Maybe this is more of a male problem overall.

While heavy (unintelligible) accents are a problem for some, my biggest beef is a person's inability to express themselves concisely in well-formed sentences. I have the same criticism for native speakers, but the problem is especially prevalent with non-native speakers. When almost EVERY sentence ends with "right?" When almost EVERY sentence includes filler adverbs like "basically", "actually", and "probably". My chief complaint with native speakers, aside from filler words, is imprecision. I never want to hear "stuff like that" used to gloss over the fact that the rest of the sentence was too imprecise or off the mark. Hand-waving phrases are indicative that the person doesn't know what they are talking about.

This generalized problem exists almost everywhere that service providers engage with customers. When communication is initiated by the provider, they are diligent about authenticating the customer, but there is no reliable mechanism for the customer to authenticate the provider. This is especially true over voice calls. Email is easier because of the header metadata, which should have been a red flag in this specific case.