[SCCC] Asking ChatGPT for CQWW advice

[Dick Stepanian]

Hi Clayton and all,
   This is Dick - W6TK.  I just saw your email.  I will try to respond more
later. I have a pretty full few days.  I often jumped in and did the best I
could with the time and station I had.  I didn't do a lot of pre-planning -
so if you can keep your BIC and have a working station with reasonable
antennas - you will probably beat the score.  At any rate go for it and
have fun....records are there to be beaten and they will be.

  I finally am getting my station on line up in OR.  I played about 10
hours in this weekend's SS-SSB - got all but SB for the Clean-sweep!  Still
fun and always a challenge1
.

73 and good luck,
Dick - W6TK
PS let me know if you have specific questions and I'll try my best to
answer.

On Mon, Nov 17, 2025 at 9:36 AM Paul Gacek via SCCC <sccc at contesting.com>
wrote:

> Hi Clayton
>
> ChatGPT is great.
>
> However, you might want to consider two other approaches to interfacing to
> a frontier LLM.
>
> Both are predicated on you mustering relevant information such as PDFs,
> web links etc. Once you’ve found a corpus of data then either;
>
> A) Use LangChain to create a RAG ( Retrieval Augmented Gen) and feed it
> your collection of data
> B) use NoteBookLM which is essentially a front end to a RAG based on
> Googles Gemini.
>
> Once either of the above is done then ask questions about the contest and
> see what you get.
>
> …..or as Bruce suggested seek advice of a real human!
>
> Paul
>
> > On Nov 17, 2025, at 8:53 AM, Bruce Horn <bhorn at hornucopia.com> wrote:
> >
> > Hi Clayton,
> >
> > Without spending a lot of time on the lengthy "analysis" ChatGPT
> produced, my observations are:
> >
> > - the "strategy" it produced seems to be a regurgitation of published
> information about contesting, rather than specific to LP or to operating
> from S.Calif, but that's what I would expect since there wouldn't be much
> data it trained on that would be specific to your needs. LLMs never say "I
> don't know."
> >
> > - the log analysis doesn't seem to reflect W6TK's log. Looks like Dick
> spent most of his time S&Ping, but ran for a couple of periods when his run
> rate exceeded his S&P rate. Overall, he averaged 33 Qs/hour based on his
> 3830scores posting, with an 82Q hour.
> >
> > - you'd learn more about his log by using the K5KA/N6TV CBS log analysis
> tool than using ChatGPT.
> >
> > - I suggest you ask Dick about his strategy, since he contested for many
> years from S.Calif before recently moving to Oregon.
> >
> > 73 de Bruce, WA7BNM   (bhorn at hornucopia.com)
> >
> >
> > ----- Original Message -----
> > From: "Clayton Nall" <clayton.nall at gmail.com>
> > To: "SCCC Reflector" <sccc at contesting.com>, "k5zd" <k5zd at cqww.com>
> > Sent: Sunday, November 16, 2025 8:06:56 PM
> > Subject: [SCCC] Asking ChatGPT for CQWW advice
> >
> > I’m planning a serious entry in CQWW CW in the LP classic overlay
> category
> > at the UCSB club station. I asked ChatGPT to give me strategy advice.  I
> > also fed it W6TK’a top-scoring log in the category and asked it to
> > summarize.
> >
> > Is this advice on point or yet another ChatGPT hallucination?
> >
> > https://chatgpt.com/share/691a9ab6-6f24-8003-9ab1-048fe22577b2
> >
> >
> > Clayton Nall
> > http://www.nallresearch.com
> > Cell: (617) 850-2062
> > _______________________________________________
> > SCCC mailing list
> > SCCC at contesting.com
> > http://lists.contesting.com/mailman/listinfo/sccc
> >
> > _______________________________________________
> > SCCC mailing list
> > SCCC at contesting.com
> > http://lists.contesting.com/mailman/listinfo/sccc
>
>
> _______________________________________________
> SCCC mailing list
> SCCC at contesting.com
> http://lists.contesting.com/mailman/listinfo/sccc
>